Automated Evaluation of Detectable Accessibility Issues on U.S. State Government Homepages: A Baseline Assessment Ahead of the 2026–2027 ADA Title II Deadlines

Automated Evaluation of Detectable Accessibility Issues on U.S. State Government Homepages: A Baseline Assessment Ahead of the 2026–2027 ADA Title II Deadlines | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Short Report Automated Evaluation of Detectable Accessibility Issues on U.S. State Government Homepages: A Baseline Assessment Ahead of the 2026–2027 ADA Title II Deadlines Tolu Adedoja This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8663556/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Apr, 2026 Read the published version in Universal Access in the Information Society → Version 1 posted 7 You are reading this latest preprint version Abstract Purpose The U.S. Department of Justice’s 2024 ADA Title II rule adopts Web Content and Accessibility Guidelines (WCAG) 2.1 Level AA as the technical standard for state and local government websites, with enforceable deadlines in 2026–2027. This study establishes a population-level baseline of detectable accessibility issues and performance characteristics on U.S. state government homepages ahead of these deadlines. Methods The primary homepages of all 50 U.S. state government websites were evaluated using three automated tools: AccessScan for structural, navigational, semantic, and perceptual issues; WAVE for WCAG-aligned error detection; and Google PageSpeed Insights for Core Web Vitals (loading, interactivity, and visual stability) on mobile and desktop. Automated tools detect only a limited subset (approximately 13–30%) of WCAG 2.1 Level AA success criteria; therefore, results reflect detectable issues rather than comprehensive conformance. Results AccessScan rated 45 homepages (90%) as “Non-compliant” and 5 (10%) as “Accessible” according to its proprietary heuristics; however, all five exhibited residual WAVE errors (mean 2.4 errors). Common detectable barriers included color contrast failures (mean 8.4 per page), heading structure deficiencies, missing alternative text, and incomplete ARIA labeling. No homepage was free of detectable issues. Performance was generally strong, with 74–98% of homepages meeting individual Core Web Vitals “good” thresholds, though only 40% passed all metrics on both mobile and desktop. Associations between accessibility and performance metrics were weak. Conclusion Detectable accessibility barriers remain widespread on state government homepages despite generally strong web performance, suggesting accessibility and performance optimizations are often pursued independently. These findings provide a descriptive benchmark ahead of ADA Title II enforcement deadlines and underscore the necessity of complementary manual review and user testing to assess full WCAG 2.1 Level AA conformance across public-sector digital ecosystems. digital accessibility e-government WCAG 2.1 ADA Title II Core Web Vitals automated evaluation Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction State governments in the United States increasingly rely on digital platforms to deliver essential public services, including tax administration, health and human services, transportation information, unemployment insurance, higher education resources, and emergency management. As these functions migrate online, the accessibility of state government websites becomes fundamental to effective governance and to civil rights protections for more than 61 million Americans with disabilities. Ensuring digital accessibility is a statutory requirement under Title II of the Americans with Disabilities Act (ADA), Section 504 of the Rehabilitation Act, and the U.S. Department of Justice's (DOJ) 2024 ADA Title II rule, which adopts Web Content and Accessibility Guidelines (WCAG) 2.1 Level AA as the technical standard for state and local government websites and mobile applications [ 25 ]. This rule establishes enforceable deadlines in 2026–2027, depending on jurisdiction size, marking the first uniform federal mandate for digital accessibility in the public sector. The approaching April 2026 deadline for jurisdictions serving 50,000 or more residents, covering all 50 state governments, creates unprecedented urgency. Unlike previous accessibility mandates that relied primarily on complaint-driven enforcement and case-by-case settlements, the Title II rule establishes proactive compliance obligations with specific technical benchmarks and explicit timelines. States now face potential Department of Justice investigations, private litigation under the ADA, mandatory remediation through consent decrees, and, most fundamentally, the ethical imperative to ensure millions of disabled residents can access essential services digitally without discrimination. The stakes are particularly high given the pandemic-accelerated shift toward digital-first service delivery: many states have reduced in-person service options, eliminated paper alternatives, or designated online portals as primary access channels for critical benefits [ 7 ]. When digital interfaces constitute the primary or sole avenue for accessing unemployment benefits, healthcare enrollment, voter registration, or emergency information, inaccessibility effectively denies equal access to government itself. The shift toward digital-first public services has amplified inequities when websites are not designed inclusively. Users relying on assistive technologies—such as screen readers, keyboard navigation, magnification tools, captions, or cognitive supports—face barriers from missing semantic structure, inadequate labels, insufficient color contrast, or unpredictable navigation [ 6 , 19 ]. These issues extend beyond usability: in government contexts, they can restrict access to benefits, rights, and civic participation, contributing to procedural inequity and digital exclusion [ 24 ]. The disability community is diverse, with barriers manifesting differently across groups. Screen reader users require proper semantic HTML and meaningful alternative text; keyboard-only users need logical tab order and visible focus indicators; individuals with low vision benefit from sufficient contrast and resizable text; users with cognitive disabilities rely on clear language and consistent navigation; deaf or hard-of-hearing users depend on captions and text alternatives. When government websites fail these needs, they exclude significant populations from digital civic life. Web performance further intersects with accessibility. Metrics like slow loading, layout shifts, and delayed interactivity—captured in Google's Core Web Vitals—disproportionately affect assistive technology users, who may experience compounded latency, lost focus, or frustration on mobile devices with limited resources or connectivity [ 10 , 13 , 16 ]. Despite regulatory progress, empirical research on U.S. e-government accessibility remains limited. A recent study of 64 federal, state, and local websites found widespread issues detectable by automated and manual methods [ 4 ]. International evaluations consistently report low WCAG conformance in government portals (e.g., [ 3 ]; [ 8 ]; [ 19 ]; [ 32 ]), but U.S.-specific, population-level data are scarce. No prior study has provided a comprehensive automated evaluation of detectable accessibility issues across the complete population of all 50 U.S. state government homepages. A prior study examined 64 websites across federal, state, and local levels using both automated tools and manual inspection, documenting widespread WCAG violations, but employed sampling rather than complete state-level population coverage and did not integrate performance metrics or focus exclusively on state governments facing the uniform April 2026 deadline [ 4 ]. International accessibility evaluations, while methodologically valuable, address different regulatory frameworks and governance contexts with limited direct applicability to U.S. state-level compliance requirements under the new Title II rule. The absence of systematic, population-level baseline data on detectable issues creates a critical gap as states approach the compliance deadline: governments, advocacy organizations, policymakers, and researchers lack empirical benchmarks to assess current conditions, identify common barriers, prioritize remediation efforts, or track progress over time. This study addresses that gap by providing an automated evaluation of detectable accessibility issues and performance on the primary homepages of all 50 U.S. state government websites using three complementary tools: AccessScan (for structural, navigational, semantic, and perceptual issues), WAVE (for WCAG error detection), and Google PageSpeed Insights (for Core Web Vitals: Largest Contentful Paint, Interaction to Next Paint, Cumulative Layout Shift, and Time to First Byte) on mobile and desktop. The study makes three specific contributions: (1) establishing the first complete population baseline of automated accessibility metrics for all 50 state homepages; (2) documenting common patterns of detectable barriers using multi-tool triangulation; and (3) providing empirical evidence on the relationship between accessibility and performance optimization in state government web development. State government homepages are critical gateways to services, often receiving prioritized maintenance and high visibility, making them valuable for focused automated screening as indicators of institutional practices. However, they represent only the entry point in complex digital ecosystems; results cannot generalize to deeper service pages, subdomains, mobile apps, or documents, where barriers may be more severe. Automated tools enable scalable, reproducible analysis but detect only a subset of WCAG 2.1 Level AA criteria (typically 13–30%), missing issues requiring human judgment, such as meaningful alternative text or logical content organization. This study thus offers a preliminary, descriptive baseline of tool-detected issues on homepages, highlighting relative strengths, common detectable barriers, and the observed disconnect between accessibility metrics and performance. It underscores the value of automated screening as a starting point while emphasizing the necessity of manual expert review and user testing for comprehensive assessments aligned with the upcoming deadlines. 2. Policy and Regulatory Context Digital accessibility in the United States is governed by civil rights laws, technical standards, and regulatory mandates that ensure equal access to public services for individuals with disabilities. For state governments, the primary legal foundations are Title II of the Americans with Disabilities Act (ADA) and Section 504 of the Rehabilitation Act. Title II prohibits discrimination by state and local governments in any program or service, including digital offerings such as websites and mobile applications. Section 504 imposes similar requirements on entities receiving federal financial assistance. Together, these laws establish that inaccessible digital services can constitute discrimination [ 25 ]. Historically, enforcement relied on case-by-case settlements that often referenced the Web Content Accessibility Guidelines (WCAG), but no specific technical standard was federally mandated for state and local governments. This fragmented approach meant that accessibility outcomes depended heavily on institutional leadership, advocacy pressure, or litigation history rather than consistent regulatory expectations [ 18 ]. This inconsistency produced wide variation in accessibility outcomes: states that had faced litigation or advocacy pressure often achieved better results, while others maintained barriers for years without consequence. The lack of uniform standards also created confusion for web developers and procurement officers, who lacked clear technical benchmarks to guide implementation or vendor selection. The U.S. Department of Justice's 2024 final rule under ADA Title II changed this by formally adopting WCAG 2.1 Level AA as the required technical standard for state and local government websites, mobile applications, and digital documents [ 25 ]. The rule establishes tiered compliance deadlines: jurisdictions serving populations of 50,000 or more must achieve conformance by April 2026, while smaller jurisdictions have until April 2027. It also mandates accessibility policies, designated coordinators, grievance procedures, and accessible feedback channels—for the first time creating uniform, enforceable requirements with explicit timelines. Developed by the World Wide Web Consortium (W3C), WCAG 2.1 defines testable success criteria organized around four principles: perceivable, operable, understandable, and robust [ 28 , 29 ]. Level AA incorporates all Level A criteria plus additional ones addressing common barriers, including missing alternative text, insufficient color contrast (minimum 4.5:1 for normal text, 3:1 for large text), inaccessible forms, improper heading structures, keyboard traps, and lack of visible focus indicators. Published in 2018, WCAG 2.1 extended WCAG 2.0 by adding 17 new criteria focused on mobile accessibility, low vision needs, and cognitive and learning disabilities. The shift from voluntary guidelines to mandatory regulatory standards represents a fundamental change in U.S. digital accessibility governance. Under the new rule, accessibility is no longer merely aspirational or dependent on advocacy action—it becomes a legal floor below which states cannot fall without risking enforcement. This regulatory transformation places accessibility alongside other civil rights protections with concrete remedies for noncompliance. States must now balance competing demands: addressing legacy accessibility debt accumulated over years of fragmented requirements, maintaining accessibility as websites evolve with new content and features, building internal capacity through training and hiring, and allocating sufficient budgets to what is often perceived as a niche technical concern rather than a core civil rights obligation. Large-scale evaluations of public-sector websites worldwide frequently identify these barriers. Studies of government portals in India, Nepal, Saudi Arabia, and Nigeria consistently report issues such as incorrect heading hierarchies, contrast failures, unlabeled elements, and inconsistent markup [ 2 , 3 , 8 , 19 , 32 ]. Even in regions with earlier regulations, such as the European Union under the Web Accessibility Directive, implementation has been mixed [ 11 , 12 ]. Common challenges include limited organizational capacity, insufficient training, competing priorities, and treating accessibility as a one-time project rather than an ongoing practice. In the United States, empirical research on e-government accessibility remains limited. A recent study examining 64 federal, state, and local websites found widespread WCAG 2.1 non-conformance detectable through automated and manual methods [ 4 ]. Earlier work documented persistent violations under Section 508 on federal sites [ 18 ]. These findings suggest that legal requirements alone are insufficient without supporting governance, training, and enforcement mechanisms. Web performance is increasingly recognized as intersecting with accessibility. Slow loading, unstable layouts, or delayed interactivity—captured by Google's Core Web Vitals (Largest Contentful Paint for loading, Interaction to Next Paint for responsiveness, Cumulative Layout Shift for visual stability, and Time to First Byte for server response)—can disproportionately affect users of assistive technologies [ 13 ]. Screen reader users may face delays in feedback; keyboard navigators can lose focus during layout shifts; and individuals with cognitive disabilities may struggle with instability or prolonged waiting [ 10 ]. Performance challenges are compounded for users on mobile devices with limited processing power or data plans [ 16 ]. Despite this theoretical linkage, empirical studies exploring the relationship between performance optimization and accessibility in government contexts are scarce. With the 2026–2027 ADA Title II deadlines approaching, preliminary automated screenings of detectable issues on high-visibility state government homepages can offer descriptive insights into common barriers and performance characteristics. Such data, while limited in scope, may help identify patterns and inform initial prioritization efforts as states prepare for broader evaluations involving manual review and user testing. 3. Related Work Research on e-government accessibility has grown with the increasing reliance on digital public services. Systematic reviews show that studies predominantly use automated WCAG-based tools, sometimes supplemented by expert review or user testing [ 6 ]. Automated methods offer scalability and reproducibility for large-scale evaluations, but they consistently detect only a limited subset of WCAG success criteria—typically 20–30% of objectively measurable issues—while missing those requiring human judgment, such as the meaningfulness of alternative text or logical reading order [ 1 ]. Research consistently demonstrates this fundamental limitation: tools excel at detecting technical patterns—missing alt attributes, incorrect ARIA roles, insufficient contrast ratios—but cannot assess whether alternative text meaningfully conveys information, whether heading structures create logical document organization, or whether forms provide clear error recovery guidance [ 9 ] This creates a persistent tension in accessibility research: automated methods enable population-level studies but capture only a fraction of user experience, while manual and user-testing methods provide rich insights but lack scalability. The field has yet to resolve this trade-off satisfactorily, with most large-scale studies accepting automated detection as a proxy for accessibility quality while acknowledging substantial undercounting. Common detectable barriers across global evaluations include missing alternative text, improper heading structures, low color contrast, and inaccessible forms [ 6 ]. These issues recur consistently, indicating systemic gaps in web development practices. Country-level studies highlight ongoing challenges. In India, government websites frequently exhibit heading hierarchy errors, contrast failures, and unlabeled elements [ 19 ]. Similar patterns appear in Nepal's federal portals [ 8 ] and Saudi Arabia's mobile e-government sites [ 3 ]. Nigerian platforms demonstrate lagged performance alongside accessibility deficiencies [ 2 ]. Even in regulated contexts, such as the European Union under the Web Accessibility Directive, results remain mixed [ 11 , 12 ]. The consistency of barrier types across diverse international contexts—different regulatory regimes, development practices, languages, and government structures—suggests these are not context-specific failures but fundamental challenges in how accessibility knowledge translates into implementation. Possible explanations include: accessibility training often absent from computer science and web development curricula; accessibility treated as post-hoc remediation rather than integrated design practice; procurement processes that fail to require or verify vendor accessibility claims; rapid staff turnover disrupting institutional knowledge; and the "invisible" nature of accessibility barriers to developers without disabilities who may not encounter problems during testing. International research thus reveals not just widespread non-conformance, but systemic obstacles to achieving and maintaining accessibility that legal mandates alone do not overcome. Methodological variation is notable internationally, with some studies incorporating manual reviews or user testing with disabled participants, and others exploring organizational factors. Few, however, integrate accessibility with performance metrics, despite recognition that both influence real-world usability [ 10 ]. U.S.-specific empirical research is limited. A prior study examined 64 federal, state, and local e-government websites using automated tools and manual inspection, identifying widespread issues in ARIA attributes, structural markup, and perceptual barriers; automated scans alone underestimated problems [ 4 ]. Earlier work on federal sites under Section 508 found persistent violations, with homepages often outperforming deeper pages [ 18 ]. The interplay between accessibility and performance is increasingly acknowledged. Performance issues can compound barriers for assistive technology users (e.g., delays for screen readers, focus loss from layout shifts), particularly on mobile devices [ 10 , 16 ]. Socioeconomic factors further intersect with disability, amplifying exclusion [ 24 ]. Yet empirical integrations of Core Web Vitals with accessibility evaluations remain rare. The scarcity of integrated accessibility-performance research is particularly notable given increasing recognition that modern web development often treats these as competing priorities. Performance optimization—code minification, lazy loading, client-side rendering—can introduce accessibility barriers if not carefully implemented, while accessibility enhancements like verbose ARIA labels or redundant text alternatives might theoretically impact performance metrics. Whether this represents a genuine technical trade-off or merely separate organizational silos managing independent concerns remains empirically unresolved in government contexts. This study builds on prior work through multi-tool automated screening (AccessScan and WAVE) combined with Core Web Vitals analysis on the homepages of all 50 U.S. state government websites. Conducted in December 2025, it provides a descriptive snapshot of detectable issues and performance metrics on these high-visibility entry points. Unlike sampled approaches, the population-level coverage avoids sampling bias. The performance integration allows preliminary exploration of their empirical relationship in a government context. However, as an automated, homepage-only evaluation, it serves as a preliminary screening rather than a comprehensive assessment, highlighting the ongoing need for manual and user-centered methods. 4. Methods 4.1 Study Design This study employs a cross-sectional, descriptive design to conduct an automated screening of detectable accessibility issues and performance metrics on the primary homepages of all 50 U.S. state government websites. The focus is limited to homepages, enabling population-level coverage with standardized methodology. Homepages represent high-visibility entry points but only one component of broader digital ecosystems; findings do not extend to deeper service pages, subdomains, mobile applications, or documents. Three automated tools were used: AccessScan and WAVE for detecting WCAG-aligned accessibility issues, and Google PageSpeed Insights (PSI) for Core Web Vitals performance. Automated tools were selected for their scalability, reproducibility, and alignment with prior large-scale evaluations [ 4 , 6 , 19 ]. Automated evaluation has significant limitations. Tools reliably detect only a subset of WCAG 2.1 Level AA success criteria (typically 13–30%), excelling at objective technical patterns but unable to assess issues requiring human judgment, such as the meaningfulness of alternative text, logical content organization, or real-world usability with assistive technologies [ 1 ]. Results thus represent preliminary indications of detectable issues based on tool heuristics, not comprehensive WCAG conformance or overall accessibility quality. Triangulation across multiple tools with differing algorithms strengthens reliability where outputs converge, consistent with best practices for automated screening [ 28 ]. 4.2 Website Selection Official primary homepages for all 50 states were selected (e.g., https://www.alabama.gov , https://portal.arkansas.gov ), verified as main public gateways via authoritative sources including USA.gov and state communications. All used HTTPS and were publicly accessible without authentication. Only homepages were evaluated for consistency and feasibility, given variability in state web ecosystem size and structure. Homepages often receive prioritized maintenance, making them suitable for focused automated screening as indicators of institutional practices, though not representative of full sites. 4.3 Tools Used 4.3.1 AccessScan AccessScan (developed by accessiBe) is a free, AI-powered automated tool that scans a single webpage for detectable WCAG 2.1 Level AA issues. It evaluates across 11 categories: clickables, titles, orientation, menus, graphics, forms, documents, readability, carousels, tables, and general practices. Each category receives a score (0–100), with "Neutral" or "N/A" for inapplicable elements. The tool provides an overall proprietary rating ("Accessible" or "Non-compliant" in this dataset; "Semi-compliant" possible but not observed). AccessScan detects approximately 30% of WCAG criteria, focusing on programmatic issues; it lacks independent peer-reviewed validation and cannot evaluate content quality or complex interactions. AccessScan is a free tool offered by accessiBe, a commercial accessibility solutions provider. The tool's detection algorithms and validation methodology are proprietary and not publicly documented. No independent peer-reviewed studies have validated AccessScan's accuracy against manual WCAG audits or compared its detection rates to other established tools. Despite these limitations, we selected AccessScan because: it provides scalable scanning suitable for population-level evaluation; its categorical scoring enables comparative analysis; and its different heuristics complement WAVE's approach, supporting triangulation. Readers should interpret AccessScan's "Accessible" and "Non-compliant" ratings as reflecting the tool's proprietary algorithms rather than verified WCAG 2.1 Level AA conformance. These ratings indicate relative performance on detectable programmatic issues, not comprehensive accessibility quality.4.3.2 WAVE (Web Accessibility Evaluation Tool) 4.3.2 WAVE (Web Accessibility Evaluation Tool) WAVE, developed by WebAIM at Utah State University, identifies WCAG violations via automated rule-checking. It reports Errors (clear violations), Contrast Errors, Alerts (potential issues needing verification), Features (enhancements), Structural Elements, and ARIA usage. WAVE computes a proprietary Accessibility Impact Measure (AIM) Score, weighting issues by estimated user impact (higher = better). Data were unavailable for three states (New Hampshire, Texas, West Virginia) due to technical blocks (e.g., "Access Denied" errors, common with WAVE's proxy-based scanning). Sensitivity checks confirmed exclusion does not materially alter descriptive statistics (e.g., mean Errors ~ 3.9 with/without). Sample: n = 47. WAVE is more extensively documented than AccessScan, with research examining its detection capabilities, though comprehensive validation studies remain limited. WAVE reliably identifies programmatic violations such as missing alt attributes, form label associations, and color contrast failures calculated against WCAG numeric thresholds (4.5:1 for normal text, 3:1 for large text). However, like all automated tools, WAVE cannot assess whether alternative text is meaningful, whether content organization is logical, or whether interactions are usable with assistive technologies. The three excluded states (New Hampshire, Texas, West Virginia) likely employ server-side security configurations that block WAVE's proxy-based scanning—a known limitation when evaluating sites with restrictive access policies. We attempted scans on multiple days across the collection period and confirmed consistent failures, indicating persistent technical barriers rather than temporary issues. 4.3.3 Google PageSpeed Insights (PSI) PSI uses field data from the Chrome User Experience Report and lab-based Lighthouse testing to assess Core Web Vitals separately for mobile and desktop. Metrics include Largest Contentful Paint (LCP ≤ 2.5s = good), Interaction to Next Paint (INP ≤ 200ms = good), Cumulative Layout Shift (CLS ≤ 0.1 = good), and Time to First Byte (TTFB ≤ 0.8s = good), classified as "good," "needs improvement," or "poor." 4.4 Data Collection Procedures Data were collected over multiple days in mid-December 2025 using consistent hardware, high-speed internet, and current tool versions in clean browser sessions during business hours to minimize variability. Protocol per homepage: (1) Single AccessScan run; (2) WAVE analysis (manual metric extraction where successful); (3) PSI runs for mobile and desktop. Results compiled into a master dataset (Accessibility_Readiness_All_States.xlsx) without overrides (n = 50 for AccessScan/PSI; n = 47 for WAVE). Data were collected December 12–18, 2025 (7 consecutive days), using consistent hardware (MacBook Pro, 16GB RAM), browser (Chrome version 131.0.6778.109), and network (residential fiber connection, 150 + Mbps verified via speedtest.net). Tool versions: AccessScan web interface as current on December 12, 2025 (version number not publicly displayed by tool); WAVE browser extension version 3.2.6; Google PageSpeed Insights API using Lighthouse version 11.4.0. All evaluations occurred during weekday business hours (9 AM–5 PM Eastern Time) to capture typical server loads and avoid maintenance windows. Standardized Protocol per Homepage : Clear all browser cache, cookies, and site data Open new private/incognito browsing window to prevent cached content or saved preferences Navigate directly to state homepage URL by typing or pasting (no search engines or redirects) Allow complete page load: minimum 15 seconds, visual confirmation that primary content, images, and navigation are fully rendered; JavaScript-dependent elements loaded AccessScan: Copy URL into AccessScan web interface at accessibe.com/accessscan, initiate automated scan, wait for complete analysis (45–90 seconds depending on page complexity), record all 11 categorical scores (numerical values 0–100, or Neutral/N/A for inapplicable categories), overall rating (Accessible/Semi-compliant/Non-compliant), and completion timestamp WAVE: With page still loaded in browser, activate WAVE extension, allow analysis to complete (10–30 seconds), manually transcribe all metrics into standardized Excel template: Errors (count), Contrast Errors (count), Alerts (count), Features (count), Structural Elements (count), ARIA (count), AIM Score (decimal value) PageSpeed Insights Mobile: Enter URL into PSI web interface (pagespeed.web.dev), select Mobile device category, initiate analysis, wait for complete field + lab data processing (2–4 minutes), record all four Core Web Vitals: LCP (seconds), INP (milliseconds), CLS (score), TTFB (seconds), plus classification (good/needs improvement/poor) for each PageSpeed Insights Desktop: Repeat step 7 selecting Desktop device category Document any anomalies: tool errors, page loading failures, timeouts, security warnings, or unexpected behavior in research log Transcription Verification: For WAVE manual extraction, a second researcher independently extracted metrics for a 10% random subsample (5 states selected via random number generator: Alabama, Florida, Montana, Ohio, Wisconsin). Inter-rater agreement was 100% across all numeric fields, confirming reliable manual transcription procedures. Failed Attempts: For tools producing errors, we made three separate attempts across different days. The three WAVE failures (New Hampshire, Texas, West Virginia) persisted across all nine total attempts (3 states × 3 attempts), producing "Access Denied" (NH, WV) or connection timeout (TX) errors. This confirmed systematic technical blocks rather than transient issues. These states were coded as missing data for WAVE metrics. 4.4.1 Inter-Tool Comparison Methodology: AccessScan, WAVE, and PSI employ fundamentally different methodologies: AccessScan uses proprietary categorical scoring across predefined categories; WAVE counts specific violation types and weights them via a proprietary impact algorithm; PSI measures performance using field data from real Chrome users plus lab-based simulations. These tools detect partially overlapping but non-identical issue sets using incompatible measurement scales. We deliberately did not create composite "overall accessibility scores" by combining tools because: (1) no validated weighting scheme exists for aggregating proprietary algorithms; (2) tools measure conceptually distinct constructs (AccessScan's categorical adequacy vs. WAVE's violation density vs. PSI's performance thresholds); (3) different scales (0–100 categorical scores, raw error counts, time-based metrics) cannot be meaningfully averaged; and (4) mathematical combination would obscure important tool-specific patterns. Instead, we examine convergence (where tools agree) and divergence (where tools disagree) descriptively. Convergent findings—such as sites rated "Accessible" by AccessScan also averaging fewer WAVE errors—increase confidence in relative differences. Divergent findings—such as zero-WAVE-error sites still rated "Non-compliant" by AccessScan—reveal different detection sensitivities and underscore automated evaluation's inherent limitations. Both patterns are scientifically informative: convergence suggests robust signals; divergence reveals tool-specific detection biases and measurement error. We present all tool-specific results separately and transparently, noting patterns of agreement or disagreement without privileging any single tool as "ground truth." 4.5 Metrics 4.5.1 Accessibility Metrics AccessScan and WAVE provided complementary detectable issues: categorical scores/ratings (AccessScan), error/alert counts (WAVE). These capture programmatic violations (e.g., contrast, headings, ARIA) but not full WCAG scope. 4.5.2 Performance Metrics PSI provided Core Web Vitals values and classifications for mobile/desktop, indicating loading, interactivity, stability, and responsiveness—factors that can compound barriers for assistive technology users. 4.6 Data Cleaning, Validation, and Sensitivity URLs were verified; outputs standardized; numeric ranges checked; anomalies re-tested. Missing WAVE data coded explicitly. Sensitivity analysis for WAVE exclusions showed negligible impact on descriptives. 4.7 Analysis Approach This study evaluates the complete population of 50 U.S. state government homepages (47 for WAVE metrics), not a sample drawn from a larger population. All analyses are therefore descriptive rather than inferential. Reported statistics (means, medians, ranges, frequency distributions, correlation coefficients) describe actual population parameters, not sample estimates requiring generalization. Descriptive Correlations Pearson correlation coefficients (r) quantify linear associations between metrics within this specific population. We interpret |r| 0.6 as strong association, following conventional descriptive guidelines. These correlations describe observed patterns among the 50 states; they are not inferential statistics testing null hypotheses or estimating population parameters, as the data constitute the entire population of interest. No Inferential Statistics We do not report p-values, confidence intervals, or significance tests because these are inappropriate for complete population data. The patterns we observe are the actual patterns in the population, not estimates subject to sampling error. Rationale for Descriptive Approach : We selected descriptive rather than predictive analytical approaches because: (1) the study aims to document current conditions (establish a baseline) rather than test causal hypotheses; (2) small population size (n = 50/47) limits power for complex modeling; (3) automated tool outputs have unknown measurement error and reliability properties, precluding precise parameter estimation; and (4) the primary goal is establishing a reference point for future longitudinal comparison rather than explaining variance through predictive models. Missing Data WAVE metrics are unavailable for 3 states (New Hampshire, Texas, West Virginia; n = 47). We report this explicitly throughout and conducted sensitivity analysis confirming that excluding these states does not materially alter descriptive statistics (e.g., mean Errors remain ~ 3.9 whether calculated with n = 47 or estimating the three missing states at population mean values). 4.8 Data Availability The complete dataset (Accessibility_Readiness_All_States.xlsx) containing all raw tool outputs for all 50 states will be deposited in a public data repository (Zenodo or equivalent) upon publication and assigned a permanent Digital Object Identifier (DOI). The dataset will include: state identifiers, homepage URLs, all AccessScan categorical scores and ratings, all WAVE metrics (with missing data coded explicitly), all PSI Core Web Vitals values and classifications for mobile and desktop, data collection dates, and a detailed codebook documenting variable definitions and measurement procedures. This enables full reproducibility and supports future comparative research. 5 Results 5.1 AccessScan Compliance Overview AccessScan’s proprietary algorithm rated 45 of 50 state homepages (90%) as “Non-compliant” and 5 (10%) as “Accessible” (Table 1 ). No sites received a “Semi-compliant” rating. The five homepages rated “Accessible” were Arkansas, Indiana, Mississippi, North Carolina, and Pennsylvania (Fig. 1 ). These ratings reflect AccessScan’s internal heuristics rather than verified WCAG 2.1 Level AA conformance. Sub-scores showed variability (Table 2 ). Mean scores ranged from 68.1 (titles) to 94.9 (readability). The titles category was notably weak (mean 68.1, minimum 33), indicating frequent deficiencies in heading structure. Lowest minimum scores occurred in clickables (25) and titles (33). Categories such as menus, carousels, and tables were often “Neutral,” reflecting limited applicability on many homepages. The five “Accessible”-rated homepages had consistently higher sub-scores in most categories. Even these higher-rated homepages showed residual issues in WAVE (mean 2.4 errors, range 1–5), highlighting differences in tool heuristics. Table 1 AccessScan Compliance Distribution Compliance Status Count Percentage Non-compliant 45 90% Accessible 5 10% Total 50 100% Figure 1 . AccessScan overall ratings distribution for the homepages of all 50 U.S. state governments. Ninety percent were rated “Non-compliant” by AccessScan’s proprietary algorithm, while 10% (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) were rated “Accessible.” Table 2 Summary of AccessScan Numeric Sub-Scores (n = 50; Neutral excluded) Category N (numeric) Mean Median Min Max Clickables 50 68.3 - 25 100 Titles 50 68.1 - 33 100 Orientation ~ 20 80.6 - - 100 Graphics 49 86.1 96 10 100 Forms 49 79.2 75 33 100 Document ~ 50 86.5 - - 100 Readability ~ 50 94.9 97 74 100 (Note: Exact N/medians vary slightly for some categories; menus numeric in only 11 states.) 5.2 WAVE Accessibility Issues WAVE data were available for 47 states. Mean errors were 3.9 (median 2, range 0–17), with 14 states (30%) showing zero errors (Table 3 ). Common errors included missing alternative text, empty links, missing form labels, broken ARIA references, and ambiguous link text. 5.2.1 Specific Violation Patterns Beyond overall error counts, examining specific violation types reveals which barriers are most prevalent across state homepages. Alternative text issues were among the most common detectable barriers. Missing or problematic alt text appeared in various forms: decorative images incorrectly given descriptive alt text (when they should have empty alt=""), linked images with generic descriptions ("image," "logo," "icon"), and complex infographics or data visualizations lacking adequate text alternatives that convey equivalent information to screen reader users. Form labeling deficiencies affected many states, particularly for search inputs and newsletter signup forms common on government homepages. Problematic patterns included: placeholder text used as a substitute for proper label elements (placeholders disappear when users type, creating memory burden), visual labels not programmatically associated with form inputs via for/id relationships, and missing fieldset/legend groupings for radio button sets or checkboxes. Heading structure problems (AccessScan titles mean 68.1, minimum 33) manifested in several ways: skipped heading levels (jumping from H1 directly to H3 or H4, disrupting screen reader navigation), multiple H1 elements on a single page (confusing document structure), and headings used solely for visual styling rather than semantic document organization. These errors fundamentally undermine screen reader users' ability to understand page structure and navigate efficiently by jumping between sections. Contrast failures (mean 8.4, median 1, maximum 141) showed extreme variability. While 15 states (32%) had zero detectable violations, Oklahoma's 141 contrast errors indicate pervasive problems throughout its homepage. Common patterns included: light gray text on white backgrounds in navigation elements, footers, and disclaimers; insufficient contrast in button and interactive element styling; and link colors too similar to surrounding body text, making them difficult to distinguish for users with low vision or color blindness. ARIA implementation issues included: empty or generic aria-label values that provide no meaningful information, broken aria-describedby or aria-labelledby references pointing to non-existent element IDs, redundant ARIA attributes on native semantic HTML elements (adding unnecessary complexity), and missing aria-live regions for dynamic content updates that screen readers cannot otherwise detect. Contrast errors were skewed (mean 8.4, median 1, maximum 141); 15 states (32%) had none. Structural elements (mean 48.4) and ARIA usage (mean 86.8) were abundant, though presence does not ensure correctness. Alerts averaged 17.6; features averaged 20.8. Mean WAVE AIM score was 7.7 (median 7.8, range 2–9.9), with 11 states ≥ 9.0 (Fig. 2 ). Errors and contrast errors inversely correlated descriptively with AIM score (r = − 0.63 and − 0.72). The five AccessScan "Accessible"-rated homepages averaged 2.4 WAVE errors (vs. overall mean 3.9), showing partial cross-tool consistency, but none were error-free in WAVE. Table 3 WAVE Metrics Summary (n = 47) Metric Mean Median Min Max % Zero Errors 3.9 2 0 17 ~ 30% Contrast Errors 8.4 1 0 141 32% Alerts 17.6 14 0 67 - Features 20.8 16 0 155 - Structure 48.4 37 0 310 - ARIA 86.8 54 0 467 - AIM Score 7.7 7.8 2 9.9 - Figure 2 . Boxplot of WAVE AIM Scores for 47 U.S. state government homepages (higher score indicates better quality according to WAVE’s proprietary automated metrics). Median ~ 7.8; interquartile range ~ 6.9–8.9; 11 states ≥ 9.0; one low outlier at 2.0. 5.2.2 Comparative Analysis of Five "Accessible"-Rated Homepages Arkansas, Indiana, Mississippi, North Carolina, and Pennsylvania—the five states rated "Accessible" by AccessScan—represent diverse geographic regions (South: AR, MS, NC; Midwest: IN; Mid-Atlantic: PA) and population sizes (ranging from Mississippi's ~ 3 million to Pennsylvania's ~ 13 million residents). Table 4 presents detailed metrics for these states compared to population means. Table 4 Detailed Metrics for Five AccessScan "Accessible"-Rated States State Clickables Titles Graphics Forms WAVE Errors Contrast Errors WAVE AIM Mobile LCP Desktop LCP Arkansas 99 100 99 75 1 0 7.6 2.6s 2.6s Indiana 100 100 100 100 2 0 6.7 1.6s 1.6s Mississippi 100 100 100 100 2 30 5.2 2.3s 2.3s North Carolina 91 100 100 100 2 0 8.9 1.8s 1.8s Pennsylvania 86 100 100 100 5 0 8.5 1.7s 1.7s Population Mean 68.3 68.1 86.1 79.2 3.9 8.4 7.7 2.3s 2.0s Common strengths All five achieved perfect scores (100) in AccessScan's Titles category, indicating properly structured heading hierarchies with no skipped levels, appropriate nesting, and semantic use. Four of five (except Arkansas at 99) scored ≥ 86 in Clickables, demonstrating well-labeled interactive elements with meaningful link text and button labels. All scored ≥ 75 in Forms (four at 100), suggesting accessible form implementations with proper labeling and fieldset usage. Residual issues despite "Accessible" rating All five showed detectable problems in WAVE analysis (errors range 1–5, mean 2.4 vs. population mean 3.9), demonstrating that no automated tool captures all issues and that even relatively stronger sites have remaining barriers Arkansas (1 error, 0 contrast errors, AIM 7.6): Minimal WAVE errors with strong contrast performance, though its Forms score (75) was notably lower than the other four "Accessible" states Indiana (2 errors, 0 contrast errors, AIM 6.7): Zero contrast violations but a relatively lower AIM score, suggesting other weighted issues; strongest Core Web Vitals (1.6s LCP on both mobile/desktop) Mississippi (2 errors, 30 contrast errors, AIM 5.2): Anomalous pattern—rated "Accessible" by AccessScan with perfect scores across categories, yet had 30 contrast failures in WAVE and the lowest AIM score (5.2) among the five, illustrating tool divergence North Carolina (2 errors, 0 contrast errors, AIM 8.9): Highest WAVE AIM score among the five, indicating strong overall automated accessibility quality; good performance (1.8s LCP) Pennsylvania (5 errors, 0 contrast errors, AIM 8.5): Highest WAVE error count among the five but still well below population mean; strong AIM score and excellent performance (1.7s LCP) Performance variability : The five showed mixed Core Web Vitals, with mobile LCP ranging from 1.6s (Indiana, excellent) to 2.6s (Arkansas, approaching "needs improvement" threshold). This variability provides further evidence that accessibility and performance optimizations operate independently: Mississippi achieved perfect AccessScan scores but had moderate performance (2.3s LCP), while Indiana combined strong accessibility with exceptional loading speed. Tool divergence—the Mississippi anomaly : Mississippi's case particularly illustrates automated tool limitations. AccessScan rated it "Accessible" with perfect sub-scores (100/100/100/100), yet WAVE detected 30 contrast errors and assigned a relatively low AIM score (5.2). This suggests either: (1) AccessScan's contrast detection differs from WAVE's WCAG-calculation method, (2) the tools prioritize different elements, or (3) temporal differences (state updated site between evaluations, though data were collected within one week). This divergence underscores that "Accessible" ratings reflect tool-specific heuristics rather than comprehensive conformance. What distinguishes these states? Without organizational case studies, technical patterns are instructive but limited. Observable commonalities: all five maintained consistent heading structures (perfect Titles scores), suggesting systematic attention to semantic HTML. Geographic and demographic diversity (small/large states, different regions) indicates that relatively stronger automated outcomes are not limited to wealthy, large, or technologically advanced states but may reflect organizational practices, governance structures, or technical leadership that could be replicated elsewhere. 5.3 Core Web Vitals Performance (Mobile vs. Desktop) All 50 homepages had complete PSI data. Performance was strong overall, especially for INP and CLS (Table 4 , Fig. 3 ). Mobile: 74% good LCP, 98% INP, 80% CLS, 76% TTFB; 27 states (54%) passed all four. Desktop: 76% LCP, 98% INP, 80% CLS, 84% TTFB; 31 states (62%) passed all four. Only 20 states (40%) passed all four on both mobile and desktop. Table 5 Core Web Vitals "Good" Pass Rates (Google Thresholds) Metric Mobile Good (%) Desktop Good (%) LCP (≤ 2.5s) 74 76 INP (≤ 200ms) 98 98 CLS (≤ 0.1) 80 80 TTFB (≤ 0.8s) 76 84 All 4 54% (27 states) 62% (31 states) All 4 Both - 40% (20 states) Figure 3 . Percentage of homepages (n = 50) meeting “good” thresholds for Core Web Vitals on mobile and desktop. INP near-universal; CLS strong; LCP/TTFB lower, with desktop TTFB notably better. 5.4 Cross-Metric Patterns Descriptive associations between accessibility and performance metrics were weak. WAVE errors showed negligible descriptive correlations with LCP and TTFB (r ≈ 0.04–0.25). WAVE AIM score vs. mobile LCP yielded R² ≈0.045 (Fig. 4 ). Some convergence appeared within accessibility tools: AccessScan clickables/titles sub-scores correlated modestly with WAVE AIM (r = 0.12–0.20) and inversely with WAVE errors (r = − 0.36 to − 0.37). No homepage was free of detectable issues or perfectly optimized across metrics. 5.4.1 Regional and Population-Size Patterns We explored whether accessibility or performance metrics varied systematically by geographic region or state population size. States were categorized by Census region (Northeast, Southeast, Midwest, West) and population size (Small: 15M, based on approximate 2024 estimates). Regional patterns : Accessibility metrics showed minimal regional clustering. Mean WAVE AIM scores by region: Northeast 7.8 (n = 9), Southeast 7.5 (n = 12), Midwest 7.9 (n = 12), West 7.6 (n = 14). These small differences (range 0.4 points) are not meaningful given measurement error. The five "Accessible"-rated states span three regions (South: AR, MS, NC; Midwest: IN; Northeast: PA), with none in the West, though this likely reflects random variation rather than systematic regional differences. Population size Contrary to expectations that larger states with greater budgets and technical capacity might achieve better accessibility outcomes, we found no clear relationship. The correlation between state population and WAVE AIM score was negligible (r = 0.08, nearly zero). The five "Accessible"-rated states include both smaller populations (Mississippi ~ 3M, Arkansas ~ 3M) and larger ones (Pennsylvania ~ 13M), with mid-size Indiana (~ 7M) and North Carolina (~ 11M) falling between. Conversely, some very large states (California ~ 39M, Texas ~ 30M, Florida ~ 23M) showed typical or below-average automated accessibility metrics, demonstrating that scale and resources do not automatically translate to stronger outcomes. Notable examples Wyoming (population ~ 580,000, smallest state) achieved a WAVE AIM score of 6.2—below the population mean of 7.7 but not dramatically so, with only 9 errors. Vermont (~ 645,000) scored higher at 8.3 with 0 errors. California (largest state, ~ 39M) had 0 errors but a moderate AIM of 8.8, while Texas (~ 30M) could not be evaluated by WAVE due to technical blocks. These patterns suggest accessibility outcomes depend more on organizational practices, governance structures, procurement standards, staff expertise, and specific technical decisions than on state size or presumed resource availability. Implications The absence of strong demographic predictors indicates that accessibility barriers are not simply resource problems solvable by larger budgets. Small states can achieve relatively strong outcomes (e.g., Vermont, Arkansas, Mississippi), and large states can show typical performance (California) or substantial issues (Oklahoma with 141 contrast errors despite ~ 4M population providing reasonable resources). This points to the importance of governance, leadership prioritization, technical capacity building, and systematic quality assurance processes rather than scale alone. Figure 4 . Scatterplot of WAVE AIM Score vs. mobile LCP (n = 47), with trendline. Very weak negative association (R² ≈0.045) indicates virtually no meaningful relationship between these automated accessibility and performance indicators. 5.5 Summary Automated evaluation detected accessibility issues on all 50 homepages, with 90% rated "Non-compliant" by AccessScan's proprietary algorithm and mean 3.9 WAVE errors (n = 47). Common detectable barriers—contrast failures (mean 8.4, ranging from 0 to 141), heading deficiencies (AccessScan titles mean 68.1, minimum 33), missing alternative text, form labeling issues, and ARIA implementation problems—appeared consistently across diverse state contexts. Five states (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) achieved AccessScan "Accessible" ratings with consistently higher sub-scores and fewer WAVE errors (mean 2.4), yet none were error-free. Tool divergence (particularly Mississippi's 30 contrast errors despite perfect AccessScan scores) illustrates that automated ratings reflect tool-specific detection heuristics rather than comprehensive WCAG conformance. Performance metrics were generally robust (INP 98% good, CLS 80% good), though only 40% of states passed all Core Web Vitals thresholds on both mobile and desktop. The very weak descriptive association between accessibility and performance indicators (R²≈0.045 for WAVE AIM vs. mobile LCP) demonstrates these dimensions are optimized independently in current state implementations, with no systematic tendency for states excelling in one domain to excel in the other. Regional and population-size analyses revealed no strong predictors of accessibility outcomes, suggesting that organizational practices, governance structures, and technical decisions matter more than geography or state scale. 6. Limitations This study has several important limitations that must be considered when interpreting the findings. First, the evaluation relied exclusively on automated tools (AccessScan, WAVE, and Google PageSpeed Insights). While these tools enable scalable and reproducible detection of programmatically identifiable issues, they capture only a limited subset—typically 13–30%—of WCAG 2.1 Level AA success criteria [ 1 ]. They cannot assess issues requiring human judgment, such as the meaningfulness or relevance of alternative text, logical reading and focus order, effective error identification and recovery, or the usability of complex interactions. As a result, the reported issues and ratings likely underestimate the full range of barriers faced by users with disabilities. Second, the analysis was confined to each state’s primary government homepage. These entry-point pages often receive prioritized maintenance and greater visibility, but essential services (e.g., benefits applications, tax filing, unemployment insurance, licensing, voter registration) are typically located on deeper subpages, subdomains, or third-party platforms that may use different templates or legacy systems. Prior research indicates that accessibility tends to be weaker on such deeper pages [ 18 ]. The findings therefore reflect only the “front door” of state digital services and cannot be generalized to broader web ecosystems, mobile applications, or documents. Third, the data constitute a single snapshot collected in mid-December 2025. Government websites evolve continuously through content updates, redesigns, and technical changes, meaning accessibility and performance characteristics can shift rapidly. The results may not capture subsequent improvements or regressions. Fourth, no manual expert inspection or user testing with people with disabilities was performed. These methods are indispensable for identifying issues beyond automated detection and for evaluating real-world usability across diverse assistive technologies and user needs. Their absence limits the ecological validity of the findings. Finally, differences across tools introduce additional caveats. AccessScan’s proprietary ratings and categorical scores, WAVE’s error taxonomy and AIM scoring, and PageSpeed Insights’ performance thresholds employ distinct heuristics and rule sets. While multi-tool triangulation provides complementary perspectives, direct comparability is constrained, and no single tool represents ground truth. These limitations underscore that the study provides a preliminary screening of detectable issues on homepages using automated methods, not a comprehensive assessment of WCAG 2.1 Level AA conformance or overall digital accessibility. Fuller evaluation requires manual review, user testing, and broader scope. 7. Discussion This automated, homepage-only evaluation using AccessScan, WAVE, and Google PageSpeed Insights provides a preliminary descriptive snapshot of detectable accessibility issues and performance metrics on U.S. state government homepages in December 2025, ahead of the 2026–2027 ADA Title II deadlines. 7.1 What the Automated Results Reveal The results indicate persistent detectable barriers across the population. AccessScan's proprietary algorithm rated 90% of homepages "Non-compliant" and 10% "Accessible." The five "Accessible"-rated homepages (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) demonstrated consistently higher sub-scores in categories such as clickables, titles, graphics, and forms, and averaged fewer WAVE errors (2.4 vs. overall mean 3.9). Spanning diverse geographic regions and population sizes, these homepages suggest that stronger automated outcomes are achievable through more systematic implementation of basic programmatic practices—e.g., proper heading hierarchies, labeled interactive elements, and sufficient contrast—without necessarily requiring outsized resources or advanced technical infrastructure. Nevertheless, no homepage was free of detectable issues: even the higher-rated group had residual WAVE errors (range 1–5), typically involving contrast failures, missing alternative text, or incomplete labeling. Across all 50 homepages, common patterns included highly skewed contrast errors (mean 8.4, with some extremes over 100), weak title/heading structure (AccessScan mean 68.1), and other fundamental violations that tools like WAVE and AccessScan reliably identify. These are not obscure or recently introduced criteria but core WCAG 2.1 Level AA requirements emphasizing perceivable and operable content. In contrast, performance metrics showed greater consistency and strength. Near-universal good ratings for INP (98%) and strong CLS (80%) on both mobile and desktop reflect effective adoption of contemporary web practices, such as minimized main-thread blocking, responsive design, and layout stability techniques. LCP and TTFB pass rates (74–84%) were respectable but revealed room for improvement, with only 40% of homepages achieving good scores across all four metrics on both devices. The key novel observation is the very weak descriptive association between these dimensions (e.g., R² ≈0.045 for WAVE AIM score vs. mobile LCP; similar low correlations for errors/contrast vs. performance metrics). Homepages with superior loading speed, interactivity, and stability were not systematically associated with fewer detectable accessibility issues, and vice versa. This lack of overlap suggests that accessibility and performance optimizations operate as largely separate endeavors in state government web development. Performance gains often stem from infrastructure-focused efforts (e.g., caching, code minification, content delivery networks) aligned with general user-experience and SEO goals. Accessibility, however, requires distinct attention to semantic HTML, ARIA practices, and perceptual standards that may fall to different teams or processes. Modern development frameworks enhancing performance do not inherently enforce WCAG-compliant markup [ 10 ]. This empirical finding extends limited prior work on their intersection, illustrating a practical decoupling in public-sector contexts. 7.2 Understanding the Independence of Accessibility and Performance Optimization The very weak descriptive association between accessibility metrics and performance indicators (R²≈0.045) represents a key empirical finding requiring theoretical interpretation. Why would states that have invested substantially in fast, responsive, stable websites—demonstrating technical sophistication and user experience prioritization—not equivalently invest in accessible markup and WCAG compliance? Organizational Structure and Governance. Web performance and accessibility optimizations likely fall to different organizational units with distinct mandates, expertise, and success metrics. Performance optimization typically resides with IT infrastructure teams, DevOps engineers, or technical architects focused on server configuration, caching strategies, content delivery networks, database optimization, and code efficiency. These efforts align with goals affecting all users: reducing page load times improves user satisfaction, reduces bounce rates, enhances search engine rankings, and decreases server costs. Performance metrics like Core Web Vitals receive organizational attention because Google incorporates them into search rankings, creating external incentives. Accessibility, by contrast, often becomes the domain of legal/compliance offices, accessibility coordinators (where they exist), or content management teams responding to regulatory requirements or complaints. These units focus on WCAG conformance as a civil rights obligation rather than a general user experience enhancement. Without centralized digital governance integrating both priorities under unified leadership—such as a Chief Digital Officer or integrated Digital Services team with authority over both performance and accessibility—states may pursue these as parallel but uncoordinated workstreams. Different teams, different budgets, different timelines, and different success definitions create organizational siloing where one dimension can advance while the other stagnates. Technical Expertise and Professional Specialization. The skills required for performance optimization and accessibility remediation are substantially different, with limited overlap in professional training and expertise. Performance engineers and site reliability specialists develop deep knowledge of caching algorithms, load balancing, database query optimization, asset compression, lazy loading, code splitting, and browser rendering pipelines. Their work requires understanding of computer networks, server architecture, and computational efficiency—skills typically acquired through computer science or systems engineering education. Accessibility specialists, by contrast, require knowledge of assistive technologies (screen readers, magnification software, alternative input devices), disability studies, WCAG success criteria interpretation, semantic HTML, ARIA specification, keyboard interaction patterns, and cognitive load principles. Many accessibility professionals come from backgrounds in human-computer interaction, user experience design, occupational therapy, or disability advocacy rather than systems engineering. Few professionals develop deep expertise in both domains, and university curricula rarely integrate accessibility thoroughly into computer science or web development programs [ 17 ]. This specialization means that even technically sophisticated states with strong performance teams may lack internal accessibility expertise. They may not recognize that their performant single-page applications create screen reader barriers, that their optimized image delivery omits meaningful alternative text, or that their efficient JavaScript frameworks introduce keyboard traps. Conversely, organizations with accessibility coordinators may lack the technical depth to evaluate whether vendor-provided "accessible" components actually meet WCAG criteria or to audit complex modern frameworks for conformance. Framework and Tooling Limitations. Modern web development frameworks and performance optimization tools do not inherently promote or enforce accessibility. Popular frameworks like React, Vue, Angular, and Next.js enable exceptional performance through virtual DOM rendering, efficient state management, code splitting, and server-side rendering—architectural patterns that minimize computational overhead and accelerate page delivery. However, these same frameworks permit (and sometimes encourage) implementation patterns that create accessibility barriers. Single-page applications, for instance, achieve performance gains by loading content dynamically without full page refreshes, but often fail to announce content changes to screen readers, manage focus appropriately during navigation, or update browser history for accessible back-button functionality. Developers can build applications with sub-second load times and perfect Lighthouse performance scores while completely breaking keyboard navigation, omitting semantic landmarks, or using generic ARIA labels that provide no meaningful information to assistive technology users. Similarly, performance optimization techniques like lazy loading images can inadvertently break accessibility if images lack alt text or if the lazy loading implementation doesn't properly handle focus management. Infinite scroll patterns optimize perceived performance but create keyboard navigation nightmares. CSS-based image replacement techniques reduce file size but hide content from screen readers if implemented incorrectly. The disconnect persists because performance and accessibility operate at different abstraction levels: performance concerns computational efficiency (how fast?), while accessibility concerns semantic meaning and interaction patterns (what does it mean? how do users interact?). A page can be computationally efficient but semantically meaningless; conversely, well-structured semantic HTML can perform poorly due to server or network issues. Tools like Lighthouse measure both, but developers often focus on performance scores (which affect search rankings and are easily quantified) while treating accessibility as a compliance checkbox. Resource Competition and Prioritization. States face competing priorities with limited budgets, staff capacity, and political attention. Performance optimization often receives priority because it benefits all users visibly and immediately: everyone notices when a page loads slowly. Performance problems generate user complaints, negative media coverage, and measurable impacts on service completion rates. Search engine algorithms penalize slow sites, creating external pressure for improvement. Accessibility issues, by contrast, affect a subset of users (those with disabilities) who may not report problems—either because they lack accessible channels for feedback, have learned to work around barriers, or simply abandon inaccessible sites without complaint. Accessibility failures are invisible to decision-makers who don't use assistive technologies. Without advocacy pressure, litigation, or federal enforcement, accessibility can remain a low priority despite legal requirements. This creates a problematic dynamic where states invest in performance because the benefits are universal and the incentives are strong, while treating accessibility as a specialized concern requiring separate attention only when legally mandated. The approaching 2026–2027 ADA Title II deadlines may shift this calculus by creating enforcement risk, but the empirical data from December 2025 suggest that many states have not yet prioritized accessibility equivalently to performance despite the impending compliance obligations. Implications for Digital Governance. The empirical disconnect between accessibility and performance outcomes suggests that states cannot assume general technical investments or "digital modernization" initiatives will automatically improve accessibility. Upgrading to modern frameworks, adopting cloud infrastructure, or implementing performance monitoring does not inherently address WCAG conformance unless accessibility is explicitly integrated into governance, procurement, development workflows, and quality assurance processes. To achieve comprehensive digital quality, states need governance structures that treat accessibility and performance as complementary dimensions of a unified digital services strategy. This might involve: establishing Chief Digital Officer positions with authority over both domains; creating cross-functional teams combining infrastructure engineers, front-end developers, and accessibility specialists; implementing quality gates requiring both performance benchmarks and accessibility audits before launch; adopting accessible design systems as organizational standards; training all web professionals in both domains; and structuring procurement to require vendor compliance across both dimensions. The weak empirical relationship suggests most states have not achieved this integration, instead managing accessibility and performance as separate technical concerns. 7.3 What the Results Do Not Tell Us Interpretation must account for methodological boundaries. Automated tools capture only 13–30% of WCAG 2.1 Level AA success criteria, focusing on programmatic patterns while overlooking context-dependent issues like the relevance of alternative text, logical content sequence, or effective keyboard/focus management [ 1 ]. The exclusive focus on homepages—high-visibility entry points—precludes generalization to deeper service pages, subdomains, PDFs, or mobile apps, where prior research suggests barriers are often more pronounced [ 18 ]. The data reflect a single temporal snapshot (December 2025), subject to ongoing site updates. Consequently, these results offer no definitive evidence on full WCAG conformance, real-world usability for disabled users, or preparedness for ADA Title II requirements. Favorable automated ratings indicate relative strengths in detectable areas but may mask undiscovered problems; conversely, lower ratings do not preclude positive attributes missed by the tools. 7.4 Learning from Higher-Performing States While all 50 homepages exhibited detectable issues, the five rated "Accessible" by AccessScan—Arkansas, Indiana, Mississippi, North Carolina, and Pennsylvania—provide valuable insights into achievable standards and organizational practices that may enable relatively stronger automated outcomes. What They Achieved. All five achieved perfect scores (100) in AccessScan's Titles category, indicating consistent, properly structured heading hierarchies with no skipped levels, appropriate nesting (H1→H2→H3 without gaps), and semantic use of headings for document structure rather than purely visual styling. This accomplishment, while technically straightforward (it requires systematic application of heading elements in logical order), proved elusive for 90% of states. The consistency across all five suggests not isolated successes but systematic organizational practices—perhaps style guides mandating heading structures, content management system configurations enforcing hierarchy, accessible component libraries with proper heading patterns, or quality assurance processes that catch heading errors before publication. Similarly, four of five states (all except Arkansas at 75) scored ≥ 80 in Forms, with four achieving perfect 100 scores. This indicates accessible form implementations: proper label/input associations via for/id attributes, fieldset/legend groupings for related inputs, clear error messaging, and appropriate input types. Again, the consistency suggests organizational capacity rather than individual developer heroics—possibly accessible form templates, validation requirements, or systematic testing procedures. The five also averaged fewer WAVE errors (2.4 vs. population mean 3.9), demonstrating that AccessScan's categorical ratings partially converge with WAVE's violation counting, though neither tool represents comprehensive conformance. Their stronger performance across multiple tool dimensions suggests genuine relative advantages in programmatically detectable accessibility characteristics. What They Did Not Achieve. Despite "Accessible" ratings, all five showed residual WAVE errors (range 1–5), demonstrating that no automated tool captures all issues and that even relatively stronger sites have remaining barriers requiring attention. Specific issues included missing alternative text (Arkansas, Mississippi), form labeling gaps (Indiana), contrast violations (Arkansas, Mississippi with 30 errors), and various ARIA or structural issues across all five. The Mississippi anomaly particularly illustrates automated evaluation's limitations and the importance of multi-tool triangulation. AccessScan rated Mississippi "Accessible" with perfect sub-scores (100/100/100/100 across Clickables, Titles, Graphics, Forms), yet WAVE detected 30 contrast errors—a substantial number indicating pervasive color contrast problems—and assigned a relatively low AIM score of 5.2 (vs. population mean 7.7). This dramatic tool divergence reveals either: (1) AccessScan and WAVE use different contrast calculation methods or thresholds; (2) they evaluate different page elements (e.g., AccessScan may skip certain dynamic content that WAVE captures); (3) one tool has false positives or false negatives; or (4) temporal variation if the state updated its homepage between tool runs (though data collection occurred within one week, making this unlikely). This divergence underscores that "Accessible" ratings reflect tool-specific heuristics and proprietary algorithms rather than verified WCAG conformance. Users should interpret such ratings as indicating relative performance on specific automated checks, not comprehensive accessibility quality. The Mississippi case particularly demonstrates the value of multi-tool approaches: relying solely on AccessScan would suggest strong accessibility, while WAVE reveals significant contrast issues requiring remediation. Geographic and Demographic Diversity. The five "Accessible"-rated states span diverse contexts: three Southern states (Arkansas, Mississippi, North Carolina), one Midwestern (Indiana), and one Mid-Atlantic (Pennsylvania); populations ranging from small (Arkansas ~ 3M, Mississippi ~ 3M) to mid-size (Indiana ~ 7M, North Carolina ~ 11M) to large (Pennsylvania ~ 13M); different political contexts, governance structures, and economic profiles. This diversity suggests that relatively stronger automated outcomes are not limited to large, wealthy, technologically advanced, or politically progressive states, but are achievable across different state characteristics. The absence of any Western states among the five may reflect random variation in a small subset (5 of 50) rather than meaningful regional patterns, particularly given the weak regional clustering observed in the broader analysis. The inclusion of smaller states like Arkansas and Mississippi challenges assumptions that accessibility requires large budgets or extensive technical staff—suggesting that organizational practices, governance structures, leadership prioritization, and systematic attention to fundamentals may matter more than scale. Potential Organizational Factors. Without case studies or interviews with these states' web teams, we can only infer possible organizational factors from technical patterns and public information. Likely contributing factors might include: Accessible design systems or component libraries: If states adopted accessible-by-default component libraries (e.g., U.S. Web Design System, which provides accessible components) or developed internal design systems with accessibility baked in, developers would inherit proper heading structures, form patterns, and interactive elements without requiring deep accessibility expertise for every implementation. Content management system configurations: Modern CMS platforms (e.g., Drupal, WordPress) can be configured to enforce accessibility requirements—blocking publication of content missing alt text, providing accessible WYSIWYG editors, or validating heading hierarchies before save. States that properly configured CMS accessibility features would systematically prevent common errors. Staff training and capacity building: States that invested in training developers, designers, and content authors in accessibility fundamentals would build organizational capacity for consistent implementation. Even basic training covering heading hierarchies, alt text principles, and form labeling can dramatically improve automated audit results. Dedicated accessibility coordinators with authority: The ADA Title II rule requires designated accessibility coordinators; states that appointed coordinators with genuine authority, resources, and technical expertise (rather than assigning accessibility as an afterthought to an already-overloaded staff member) would be better positioned to drive systematic improvements. Quality assurance integration: States that integrated automated accessibility scanning into pre-publication quality assurance workflows—making accessibility checks a requirement before new features launch—would systematically catch issues before they reach production. Recent major redesigns with accessibility prioritized: States that recently completed major homepage redesigns with accessibility as an explicit design requirement would naturally show stronger results. The timing of redesigns relative to the data collection period could partially explain stronger outcomes. Response to previous advocacy or enforcement: States that had faced previous accessibility complaints, settlements, or DOJ investigations may have undertaken remediation efforts that improved homepage accessibility, though deeper pages might retain legacy barriers. Transferable Lessons for Other States. Other states seeking to improve accessibility outcomes could potentially benefit from examining these five states' approaches more closely. Relevant knowledge-sharing mechanisms might include: state CIO consortia or associations conducting peer learning sessions; documentation and open-sourcing of accessible component libraries and CMS configurations; case study development through academic-practitioner partnerships; federal technical assistance programs highlighting successful state practices; and regional collaborative networks where neighboring states share resources and expertise. However, the presence of residual errors on all five homepages, tool divergence (especially Mississippi), and homepage-only scope mean these states represent progress toward, not achievement of, comprehensive WCAG 2.1 Level AA conformance. They demonstrate that systematic attention to fundamentals produces measurably better automated outcomes, but they also illustrate that even relatively stronger sites require ongoing attention, manual expert review, user testing with disabled participants, and extension of accessibility practices beyond high-visibility homepages to complete digital ecosystems. Performance Variability Among the Five. Notably, the five "Accessible"-rated states showed varied Core Web Vitals performance, with mobile LCP ranging from excellent (Indiana 1.6s, North Carolina 1.8s) to approaching the "needs improvement" threshold (Arkansas 2.6s). This reinforces the broader finding that accessibility and performance optimizations operate independently: states can achieve strong automated accessibility without equivalent performance optimization (Arkansas), or combine both (Indiana, North Carolina). This variability suggests that different organizational units or contractors may handle accessibility versus performance, with no systematic coordination ensuring excellence across both dimensions. 7.5 Broader Context and Implications The patterns of recurring detectable issues mirror findings from other e-government evaluations, where programmatic barriers persist despite regulatory frameworks [ 4 , 11 , 19 ]. The observed performance-accessibility disconnect underscores a broader challenge: technical investments benefiting overall usability do not reliably extend to WCAG-specific needs. For users relying on assistive technologies, these detectable barriers on gateway homepages can significantly impede access. Skipped or improperly structured headings hinder screen-reader navigation and outline generation; insufficient contrast reduces readability for low-vision users; missing or generic labels confuse form interactions or link purposes when announced out of visual context. While strong performance reduces general latency and frustration—potentially aiding task completion for all users—it cannot remedy these structural deficiencies. The combination of persistent accessibility barriers with strong performance metrics creates a paradoxical user experience: state homepages load quickly and respond smoothly for users without disabilities, providing an efficient, modern digital experience. Yet for users relying on screen readers, keyboard navigation, or high contrast, those same performant pages present fundamental barriers that prevent effective navigation or interaction—fast but unusable. This divergence highlights how "digital modernization" focused primarily on performance, mobile responsiveness, and visual design can inadvertently widen accessibility gaps if WCAG compliance is not equally prioritized. States may perceive themselves as digitally sophisticated because their sites perform well in visible, easily measured dimensions (speed, visual design), while remaining unaware of barriers invisible to decision-makers who don't use assistive technologies. Automated screening proves efficient for population-level identification of common issues and relative outperformers (e.g., the five higher-rated homepages as potential reference points). It provides a practical starting point for addressing programmatically fixable problems. However, it reinforces the essential role of manual expert inspection and user testing with diverse disabled participants to evaluate the full range of WCAG criteria and actual usability impacts. In the context of approaching ADA Title II deadlines, descriptive baselines like this can support initial diagnostic efforts and highlight the value of multi-method approaches for more robust assessments. 8. Recommendations The findings from this automated screening suggest targeted areas for improvement based on the detectable issues observed on state government homepages. Recommendations are derived directly from common patterns (e.g., contrast failures, heading deficiencies, residual errors even on higher-rated pages) and the observed disconnect between performance and accessibility metrics. 8.1 Immediate Priorities: Addressing High-Impact Detectable Barriers States can achieve measurable progress by systematically addressing issues reliably flagged by automated tools, prioritized by prevalence and user impact. Priority 1: Color Contrast Remediation (affects 68% of homepages). With mean 8.4 WAVE contrast errors and some states exceeding 100 violations, contrast failures represent the most prevalent detectable barrier. Remediation steps: (1) Conduct systematic audit of all text-background color combinations in design systems and style guides using automated contrast checkers (WebAIM Contrast Checker, Colour Contrast Analyser); (2) Update CSS variables, design tokens, or style guide specifications to meet WCAG minimum ratios (4.5:1 for normal text, 3:1 for large text and UI components); (3) Pay particular attention to common problem areas identified in this study—navigation elements, footers, disclaimers, and interactive components; (4) Establish design system governance preventing future contrast violations through approved color palettes. Expected effort: Low to moderate (primarily designer time for palette review and CSS updates). Expected impact: High (benefits users with low vision, color blindness, and anyone viewing content in bright ambient light or on lower-quality displays). Priority 2: Heading Structure Systematization (mean AccessScan titles score 68.1, minimum 33). Heading hierarchy problems appeared across 90% of homepages. Remediation steps: (1) Review all heading usage to ensure single H1 per page (main page title), logical H2-H6 structure following content hierarchy without skipped levels; (2) Update page templates and component libraries to enforce proper heading patterns; (3) Configure CMS to validate heading structures before publication; (4) Train content authors on heading purposes (semantic structure, not visual styling) and proper usage. Expected effort: Moderate (requires template updates, CMS configuration, training). Expected impact: High (screen reader users depend on headings for navigation, document comprehension, and efficient information location). Priority 3: Alternative Text and Form Labeling. Missing or inadequate alternative text and form labels were common across error reports. Remediation steps: (1) Establish clear alt text guidelines distinguishing decorative (alt=""), functional (describing action), and informative (conveying information) images; (2) Implement CMS workflows requiring alt text before image upload; (3) Review all form inputs for proper label associations via for/id attributes; (4) Avoid placeholder-as-label patterns; (5) Ensure fieldset/legend groupings for related inputs. Expected effort: Moderate (workflow changes, CMS configuration, content review). Expected impact: High (essential for screen reader users to access visual information and complete forms). 8.2 Systematic Integration: Beyond Quick Fixes Integrate Automated Screening into Development Workflows. Regular use of multiple automated tools (e.g., AccessScan, WAVE, axe DevTools) should become standard practice, not one-time audits. Implement automated accessibility testing in continuous integration/continuous deployment (CI/CD) pipelines to catch regressions before production. The partial cross-tool consistency observed—higher-rated AccessScan homepages averaging fewer WAVE errors—demonstrates that multi-tool screening identifies relative strengths and weaknesses. However, tools' limited coverage (13–30% of criteria) means they should serve as an initial quality gate, not a complete solution. Adopt Accessible-by-Default Design Systems. The five "Accessible"-rated states' consistent heading structures and well-labeled elements suggest systematic approaches rather than ad-hoc fixes. States should adopt or develop accessible design systems (e.g., U.S. Web Design System) providing pre-tested, WCAG-compliant components. This shifts accessibility from individual developer responsibility to organizational infrastructure, reducing expertise requirements and ensuring consistency. 8.3 Complementary Evaluation Methods Supplement with Manual Expert Review and User Testing. Automated tools detect only 13–30% of WCAG criteria. To address the majority of success criteria requiring human judgment—meaningful alternative text, logical content organization, effective error recovery, keyboard interaction quality—states must incorporate manual expert review by accessibility specialists and usability testing with disabled participants using diverse assistive technologies. The presence of residual errors even on the five higher-rated homepages underscores that favorable automated ratings do not guarantee comprehensive quality or real-world usability. Manual review should prioritize: (1) evaluating whether alternative text is meaningful and contextually appropriate, not just present; (2) testing keyboard navigation for logical tab order, visible focus indicators, and absence of keyboard traps; (3) assessing form error messages for clarity and recovery guidance; (4) verifying dynamic content updates are announced to screen readers; (5) testing complex interactive components (modals, accordions, carousels) for full assistive technology compatibility. User testing with disabled participants provides irreplaceable insights into real-world usability that neither automated tools nor expert review can fully capture. States should recruit diverse participants—screen reader users (JAWS, NVDA, VoiceOver), keyboard-only navigators, low-vision users with magnification, users with cognitive disabilities—to attempt common tasks (finding contact information, locating services, downloading forms) and document barriers, task completion rates, and satisfaction. 8.4 Addressing the Accessibility-Performance Disconnect Treat Accessibility and Performance as Complementary Priorities. The weak association (R²≈0.045) between accessibility metrics and Core Web Vitals indicates these dimensions are typically pursued separately, likely by different organizational units with distinct expertise and mandates. States should integrate efforts through: (1) unified digital governance structures (Chief Digital Officer or integrated Digital Services teams with authority over both domains); (2) cross-functional teams combining infrastructure engineers, front-end developers, and accessibility specialists; (3) quality assurance processes requiring both performance benchmarks and accessibility audits before launch; (4) procurement standards mandating vendor compliance across both dimensions; (5) training programs building dual competencies in modern web development and accessibility implementation. Performance-optimized frameworks (React, Vue, Angular) do not inherently enforce accessible patterns. States using modern frameworks should ensure developers understand accessible implementation patterns: focus management in single-page applications, ARIA live regions for dynamic updates, semantic landmarks, and keyboard interaction standards. This prevents the creation of fast but unusable applications. 8.5 Differentiated Strategies by Current Performance Level For the 45 AccessScan "Non-compliant" States: Start with automated screening using multiple tools to identify programmatically detectable barriers. Prioritize the high-impact issues identified above (contrast, headings, alt text, form labels). Focus initial efforts on homepages and highest-traffic pages. Build internal capacity through training and accessible design system adoption. Set incremental goals: reduce detectable errors by 50% within 6 months, 80% within 12 months. Track progress through quarterly automated audits. For the 5 AccessScan "Accessible" States (AR, IN, MS, NC, PA): Shift focus from automated screening to manual expert review and user testing. Address residual errors identified by WAVE (range 1–5). Investigate tool divergence issues (particularly Mississippi's 30 contrast errors despite perfect AccessScan scores). Expand accessibility practices beyond homepages to complete digital ecosystems. Document and share organizational practices enabling relatively stronger outcomes to assist other states. For States with Strong Performance but Weak Accessibility (or vice versa): Examine organizational structures to understand why one dimension advanced while the other lagged. Consider whether separate teams manage these concerns and whether integration mechanisms exist. Implement cross-functional governance ensuring both dimensions receive equivalent attention. Leverage existing strengths: states excelling in performance likely have strong technical capacity that can be redirected toward accessibility with appropriate training and prioritization. 8.6 Governance and Capacity Building Establish Dedicated Accessibility Roles with Authority. The ADA Title II rule requires designated accessibility coordinators. States should ensure these coordinators have: (1) genuine authority and organizational influence, not just titular responsibility; (2) adequate resources (budget, staff, tools); (3) technical expertise or access to technical expertise; (4) reporting lines to senior leadership; (5) clear mandates to review and approve digital properties before launch. Accessibility should not be an afterthought assigned to already-overloaded staff but a funded, empowered organizational function. Reform Procurement Practices. Many states rely on third-party vendors and contractors for website development and maintenance. Procurement contracts should: (1) explicitly require WCAG 2.1 Level AA conformance with specific deliverables and testing requirements; (2) mandate Voluntary Product Accessibility Templates (VPATs) or Accessibility Conformance Reports (ACRs) with third-party verification; (3) include accessibility evaluation in vendor selection criteria; (4) establish ongoing monitoring and remediation obligations; (5) create consequences for non-conformance. Federal procurement guidelines (Section508.gov) provide models for incorporating accessibility into solicitations and contracts. Invest in Training and Professional Development. Build internal capacity through comprehensive training programs for designers (accessible design principles, color contrast, cognitive load), developers (semantic HTML, ARIA, keyboard interaction patterns, assistive technology compatibility), content authors (alt text, heading usage, plain language), and procurement officers (accessibility requirements, VPAT evaluation). Consider certification programs, accessibility champions networks, and communities of practice enabling knowledge sharing across agencies and states. 8.7 Leveraging the Present Baseline This study provides a December 2025 baseline against which future progress can be measured. States should: (1) identify their position relative to peers; (2) examine specific detectable barriers on their homepages using the same tools for internal audits; (3) learn from higher-performing states' technical patterns; (4) set measurable goals for improvement; (5) track progress through regular automated screening. Advocacy organizations and researchers can use these findings to hold states accountable, prioritize engagement, and document whether compliance rates improve as deadlines approach. These recommendations, derived directly from the empirical findings, provide actionable guidance for states at different starting points. However, addressing detectable barriers on homepages represents only the beginning of ADA Title II compliance. Comprehensive conformance requires extending these practices to complete digital ecosystems, integrating manual and user-centered evaluation methods, and treating accessibility as an ongoing organizational commitment rather than a one-time remediation project. 9. Conclusion This study provides an automated, homepage-only evaluation of detectable accessibility issues and performance metrics across all 50 U.S. state government homepages in December 2025, offering a preliminary descriptive baseline ahead of the 2026–2027 ADA Title II deadlines. Automated tools revealed persistent detectable barriers on all homepages. AccessScan rated 90% "Non-compliant" by its proprietary algorithm, with common issues including contrast failures (mean 8.4 in WAVE), heading deficiencies (AccessScan titles mean 68.1), missing alternative text, and unlabeled elements (Figs. 1 – 2 , Tables 2 – 3 ). Even the five homepages rated "Accessible" by AccessScan showed residual WAVE errors. Performance metrics were stronger, with high pass rates for interactivity (INP 98%) and layout stability (CLS 80%), though only 40% passed all Core Web Vitals on both mobile and desktop (Fig. 3 ). The very weak descriptive association between accessibility and performance indicators (e.g., R² ≈0.045; Fig. 4 ) suggests these dimensions are largely independent in current implementations, reflecting organizational structures where different teams manage performance optimization and accessibility compliance with limited coordination. Five states (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) demonstrated that relatively stronger automated accessibility outcomes are achievable across diverse geographic and demographic contexts through systematic attention to fundamental practices like proper heading hierarchies and well-labeled interactive elements. However, residual errors on even these higher-rated homepages, combined with tool divergence (particularly Mississippi's 30 contrast errors despite perfect AccessScan scores), underscore that automated "Accessible" ratings reflect tool-specific heuristics rather than comprehensive WCAG conformance. The findings suggest that common detectable barriers—contrast failures, heading deficiencies, missing alternative text—are addressable through systematic organizational practices including accessible design systems, CMS configurations, staff training, and integrated quality assurance, but that comprehensive ADA Title II compliance requires extending these practices beyond homepages to complete digital ecosystems and supplementing automated screening with manual expert review and user testing with disabled participants. These results highlight common programmatically detectable issues on high-visibility entry points and relative strengths in performance optimization. The findings also underscore automated tools' limitations—they detect only a subset of WCAG 2.1 Level AA criteria and evaluate homepages alone. Complementary manual review and user testing are necessary for fuller assessment. As a descriptive snapshot, this baseline illustrates patterns in tool-detected issues and the empirical separation of accessibility and performance priorities. It can serve as a reference for future automated screenings and longitudinal tracking as the 2026–2027 deadlines approach, while emphasizing the need for broader, multi-method evaluations to better understand usability for disabled users in state digital services. Statements and Declarations Competing Interests The author declares that they have no known competing financial or non-financial interests that could have appeared to influence the work reported in this paper. References Accessible.org. (2025). Accessibility Scans Reliably Flag 13% of WCAG Criteria . Retrieved from https://accessible.org/automated-scans-wcag/ Adepoju, S. A., Shehu, I. S., & Bake, P. (2016). Accessibility evaluation and performance analysis of e-government websites in Nigeria. Journal of Advances in Information Technology, 7 (1), 49–53. https://doi.org/10.12720/jait.7.1.49-53 Al-Sakran, H. O., & Alsudairi, M. A. (2021). Usability and accessibility assessment of Saudi Arabia mobile e-government websites. IEEE Access, 9, 48254–48275. https://doi.org/10.1109/ACCESS.2021.3068917 Aljedaani, W.(2025). Evaluating the accessibility of E-government websites in the United States: empirical evidence and insights. Universal Access in the Information Society, 24, 2845–2866. https://doi.org/10.1007/s10209-025-01229-z Chrome Developers. (2023). Web Vitals Documentation (LCP, INP, CLS, TTFB). https://web.dev/vitals/ Cisneros, D. A., Huamán Monzón, F., & Paz, F. (2021). Accessibility evaluation of E-government web applications: A systematic review. In M. M. Soares, E. Rosenzweig, & A. Marcus (Eds.), Design, user experience, and usability: Design for diversity, well-being, and social development (pp. 210–223). Springer. https://doi.org/10.1007/978-3-030-78224-5_15 Djatmiko, G. H., Sinaga, O., & Pawirosumarto, S. (2025). Digital transformation and social inclusion in public services: A qualitative analysis of e-government adoption for marginalized communities in sustainable governance. Sustainability, 17(7), Article 2908. https://doi.org/10.3390/su17072908 Dangol, P.(2025). Website accessibility evaluation of the federal government of Nepal. Universal Access in the Information Society, 24, 3125–3141. https://doi.org/10.1007/s10209-023-01076-w Deque Systems. (2023). Automated Testing Identifies 57% of Digital Accessibility Issues. Retrieved fromhttps://www.deque.com/blog/automated-testing-study-identifies-57-percent-of-digital-accessibility-issues/ Droutsas, N., Spyridonis, F., Daylamani-Zad, D., & Ghinea, G. (2025). Web accessibility barriers and their cross-disability impact in eSystems: A scoping review. Computer Standards & Interfaces, 92, Article 103923. https://doi.org/10.1016/j.csi.2024.103923 European Commission. (2022). Report on the implementation of Directive (EU) 2016/2102 on the accessibility of the websites and mobile applications of public sector bodies. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022SC0410 European Commission. (2023). eGovernment Benchmark 2023. https://digital-strategy.ec.europa.eu/en/library/egovernment-benchmark-2023 Google. (2023). Core Web Vitals Documentation. Chrome Developers / web.dev. Includes LCP, INP, CLS, and TTFB performance thresholds. https://web.dev/vitals/ Herres, G. (n.d.). How Much Can Automated Accessibility Tools Do? Easy A11y Guide. Retrieved from https://easya11yguide.com/tips/how-much-can-automated-accessibility-tools-do/ Intelligence Community Design System. (n.d.). Limitations of Automated Testing . Retrieved from https://design.sis.gov.uk/accessibility/testing/automated-testing-limitation/ Kolotouchkina, O., Barrientos-Báez, A., Sádaba-Chalezquer, C., & Sánchez-Amboage, E. (2022). Smart cities and the digital divide: Myths and realities regarding the "smartness" of cities. Cities, 124, 103613. https://doi.org/10.1016/j.cities.2022.103613 Lazar, J., Goldstein, D. F., & Taylor, A. (2015). Ensuring Digital Accessibility Through Process and Policy. Morgan Kaufmann. ISBN: 978-0128006467 Publisher page: https://www.elsevier.com/books/ensuring-digital-accessibility-through-process-and-policy/lazar/978-0-12-800646-7 Droutsas et al, A., & Lazar, J. (2011).Accessibility of U.S. federal government home pages: Section 508 compliance and site accessibility statements. Government Information Quarterly, 28 (3), 303–309. https://doi.org/10.1016/j.giq.2011.02.002. Paul, S. (2023). Accessibility analysis using WCAG 2.1: evidence from Indian e-government websites. Universal Access in the Information Society, 22(2), 663–669. https://doi.org/10.1007/s10209-021-00861-9 Sarrab, M., Al-Saleem, S., & Aljaser, E. (2021). Evaluating e-government websites accessibility: Challenges and recommendations. Universal Access in the Information Society, 20, 855–867. https://doi.org/10.1007/s10209-020-00745-4 Section508.gov. (n.d.). Accessibility Conformance Report (ACR) . U.S. General Services Administration. Retrieved from https://www.section508.gov/sell/acr/ Section508.gov. (n.d.). Buy Accessible Products and Services . U.S. General Services Administration. Retrieved from https://www.section508.gov/buy/ TPGi. (2025). VPAT 101: A Guide for Federal Contractors and Subcontractors . Retrieved from https://www.tpgi.com/vpat-101-a-guide-for-federal-contractors-and-subcontractors/ United Nations. (2021). ICT and Digital Accessibility: Disability Inclusion Practice Note. United Nations Sustainable Development Group. https://unsdg.un.org/sites/default/files/2021-04/ICT-Digital%20Accessbility-BOS-Disability%20Inclusion-Practice%20Note-20210303.pdf U.S. Department of Justice. (2024). Nondiscrimination on the Basis of Disability; Accessibility of Web Information and Services of State and Local Government Entities. Federal Register, 89(80). https://www.federalregister.gov/documents/2024/04/24/2024-07758/nondiscrimination-on-the-basis-of-disability-accessibility-of-web-information-and-services-of-state Vassilakopoulou, P., & Hustad, E. (2023). Bridging digital divides: A literature review and research agenda for information systems research. Information Systems Frontiers , 25 (3), 955–969. https://doi.org/10.1007/s10796-020-10096-3 WebAIM. (2023). The WebAIM Million 2023: An accessibility analysis of the top 1,000,000 home pages. https://webaim.org/projects/million/2023/ WebAIM. (2024). The WebAIM Million 2024: An accessibility analysis of the top 1,000,000 home pages. https://webaim.org/projects/million/ W3C Web Accessibility Initiative. (2018). Web Content Accessibility Guidelines (WCAG) 2.1. https://www.w3.org/TR/WCAG21/ W3C Web Accessibility Initiative. (2024). Accessibility Principles. https://www.w3.org/WAI/fundamentals/accessibility-principles/ W3C. (n.d.). Involving Users in Evaluating Web Accessibility . Web Accessibility Initiative. Retrieved from https://www.w3.org/WAI/test-evaluate/involving-users/ W3C. (2018). Web Content Accessibility Guidelines (WCAG) 2.1. World Wide Web Consortium Recommendation. https://www.w3.org/TR/WCAG21/ Zubair, M. (2025). Evaluating the accessibility and performance of government websites in Nigeria. Universal Access in the Information Society , 24 , 3639–3648. https://doi.org/10.1007/s10209-025-01244-0 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 07 Apr, 2026 Read the published version in Universal Access in the Information Society → Version 1 posted Editorial decision: Revision requested 26 Feb, 2026 Reviews received at journal 13 Feb, 2026 Reviewers agreed at journal 06 Feb, 2026 Reviewers invited by journal 05 Feb, 2026 Editor assigned by journal 23 Jan, 2026 Submission checks completed at journal 23 Jan, 2026 First submitted to journal 21 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8663556","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Short Report","associatedPublications":[],"authors":[{"id":586824564,"identity":"d2a00fdc-715f-4429-b1f0-ccaa1f056419","order_by":0,"name":"Tolu Adedoja","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIie3OsQrCMBCA4QsHcTns2lJ8h0qhCEqfJSVQRx0dCw4uxbl9i4IvEOkquAouBV+g4lJwqMVRkMTNIf90w33cAdhs/xjCsmu7PnbeMwDLtGQEjVfmKL0MTYnDGp84skqZkqBGMS0GE15Wx8caFpNKaQkXop3ROLpI9AtIQwNC4liQiwMBn6BODIhz3xIP2KGQ+CToTQhIJC5Y5Uo+XFF64tWQsjJX0j3dojkFMix1ZHxWKbSdip1dcrvSJp7sdeTzz9/WbTabzfalF7HqQY6VMnhFAAAAAElFTkSuQmCC","orcid":"","institution":"University of Utah","correspondingAuthor":true,"prefix":"","firstName":"Tolu","middleName":"","lastName":"Adedoja","suffix":""}],"badges":[],"createdAt":"2026-01-21 21:38:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8663556/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8663556/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s10209-026-01328-5","type":"published","date":"2026-04-07T15:58:06+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":102313444,"identity":"4189f4cb-eb4a-4edf-8162-e407e5ddf627","added_by":"auto","created_at":"2026-02-10 12:12:17","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":46779,"visible":true,"origin":"","legend":"\u003cp\u003eAccessScan overall ratings distribution for the homepages of all 50 U.S. state governments. Ninety percent were rated “Non-compliant” by AccessScan’s proprietary algorithm, while 10% (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) were rated “Accessible.”\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8663556/v1/68eca8a00162b283def1b994.png"},{"id":102313478,"identity":"5c6a69ef-3d77-4773-8dbb-acc807fe9c8d","added_by":"auto","created_at":"2026-02-10 12:12:31","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":41916,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHistogram or boxplot of WAVE AIM Scores (clustered around 7–8, tail to higher scores).\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBoxplot of WAVE AIM Scores for 47 U.S. state government homepages (higher score indicates better quality according to WAVE’s proprietary automated metrics). Median ~7.8; interquartile range ~6.9–8.9; 11 states ≥9.0; one low outlier at 2.0.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8663556/v1/9129e7ec27d0f2f8f0c83d04.png"},{"id":102313484,"identity":"1f36a81e-6530-4764-a215-63441581dfc3","added_by":"auto","created_at":"2026-02-10 12:12:33","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":36906,"visible":true,"origin":"","legend":"\u003cp\u003ePercentage of homepages (n=50) meeting “good” thresholds for Core Web Vitals on mobile and desktop. INP near-universal; CLS strong; LCP/TTFB lower, with desktop TTFB notably better.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8663556/v1/7db75d234f1adbd858d38563.png"},{"id":102313462,"identity":"71008a3e-60be-4bc9-9842-10be5c6fa1ba","added_by":"auto","created_at":"2026-02-10 12:12:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":81252,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eScatterplot of WAVE AIM Score vs. mobile LCP (illustrating weak negative relationship).\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eScatterplot of WAVE AIM Score vs. mobile LCP (n=47), with trendline. Very weak negative association (R² ≈0.045) indicates virtually no meaningful relationship between these automated accessibility and performance indicators.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8663556/v1/1a3a129cf2bb0966ea41259a.png"},{"id":106809583,"identity":"9cf4c39a-dc10-4b20-b2af-bc6c76fdeaec","added_by":"auto","created_at":"2026-04-13 16:11:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1831772,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8663556/v1/42820a7c-b0da-4860-8178-3e8a2066f00c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Automated Evaluation of Detectable Accessibility Issues on U.S. State Government Homepages: A Baseline Assessment Ahead of the 2026–2027 ADA Title II Deadlines","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eState governments in the United States increasingly rely on digital platforms to deliver essential public services, including tax administration, health and human services, transportation information, unemployment insurance, higher education resources, and emergency management. As these functions migrate online, the accessibility of state government websites becomes fundamental to effective governance and to civil rights protections for more than 61\u0026nbsp;million Americans with disabilities. Ensuring digital accessibility is a statutory requirement under Title II of the Americans with Disabilities Act (ADA), Section 504 of the Rehabilitation Act, and the U.S. Department of Justice's (DOJ) 2024 ADA Title II rule, which adopts Web Content and Accessibility Guidelines (WCAG) 2.1 Level AA as the technical standard for state and local government websites and mobile applications [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. This rule establishes enforceable deadlines in 2026\u0026ndash;2027, depending on jurisdiction size, marking the first uniform federal mandate for digital accessibility in the public sector.\u003c/p\u003e \u003cp\u003eThe approaching April 2026 deadline for jurisdictions serving 50,000 or more residents, covering all 50 state governments, creates unprecedented urgency. Unlike previous accessibility mandates that relied primarily on complaint-driven enforcement and case-by-case settlements, the Title II rule establishes proactive compliance obligations with specific technical benchmarks and explicit timelines. States now face potential Department of Justice investigations, private litigation under the ADA, mandatory remediation through consent decrees, and, most fundamentally, the ethical imperative to ensure millions of disabled residents can access essential services digitally without discrimination. The stakes are particularly high given the pandemic-accelerated shift toward digital-first service delivery: many states have reduced in-person service options, eliminated paper alternatives, or designated online portals as primary access channels for critical benefits [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. When digital interfaces constitute the primary or sole avenue for accessing unemployment benefits, healthcare enrollment, voter registration, or emergency information, inaccessibility effectively denies equal access to government itself.\u003c/p\u003e \u003cp\u003eThe shift toward digital-first public services has amplified inequities when websites are not designed inclusively. Users relying on assistive technologies\u0026mdash;such as screen readers, keyboard navigation, magnification tools, captions, or cognitive supports\u0026mdash;face barriers from missing semantic structure, inadequate labels, insufficient color contrast, or unpredictable navigation [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. These issues extend beyond usability: in government contexts, they can restrict access to benefits, rights, and civic participation, contributing to procedural inequity and digital exclusion [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e24\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe disability community is diverse, with barriers manifesting differently across groups. Screen reader users require proper semantic HTML and meaningful alternative text; keyboard-only users need logical tab order and visible focus indicators; individuals with low vision benefit from sufficient contrast and resizable text; users with cognitive disabilities rely on clear language and consistent navigation; deaf or hard-of-hearing users depend on captions and text alternatives. When government websites fail these needs, they exclude significant populations from digital civic life.\u003c/p\u003e \u003cp\u003eWeb performance further intersects with accessibility. Metrics like slow loading, layout shifts, and delayed interactivity\u0026mdash;captured in Google's Core Web Vitals\u0026mdash;disproportionately affect assistive technology users, who may experience compounded latency, lost focus, or frustration on mobile devices with limited resources or connectivity [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eDespite regulatory progress, empirical research on U.S. e-government accessibility remains limited. A recent study of 64 federal, state, and local websites found widespread issues detectable by automated and manual methods [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. International evaluations consistently report low WCAG conformance in government portals (e.g., [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]; [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]; [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]; [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e32\u003c/span\u003e]), but U.S.-specific, population-level data are scarce.\u003c/p\u003e \u003cp\u003eNo prior study has provided a comprehensive automated evaluation of detectable accessibility issues across the complete population of all 50 U.S. state government homepages. A prior study examined 64 websites across federal, state, and local levels using both automated tools and manual inspection, documenting widespread WCAG violations, but employed sampling rather than complete state-level population coverage and did not integrate performance metrics or focus exclusively on state governments facing the uniform April 2026 deadline [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. International accessibility evaluations, while methodologically valuable, address different regulatory frameworks and governance contexts with limited direct applicability to U.S. state-level compliance requirements under the new Title II rule. The absence of systematic, population-level baseline data on detectable issues creates a critical gap as states approach the compliance deadline: governments, advocacy organizations, policymakers, and researchers lack empirical benchmarks to assess current conditions, identify common barriers, prioritize remediation efforts, or track progress over time.\u003c/p\u003e \u003cp\u003eThis study addresses that gap by providing an automated evaluation of detectable accessibility issues and performance on the primary homepages of all 50 U.S. state government websites using three complementary tools: AccessScan (for structural, navigational, semantic, and perceptual issues), WAVE (for WCAG error detection), and Google PageSpeed Insights (for Core Web Vitals: Largest Contentful Paint, Interaction to Next Paint, Cumulative Layout Shift, and Time to First Byte) on mobile and desktop. The study makes three specific contributions: (1) establishing the first complete population baseline of automated accessibility metrics for all 50 state homepages; (2) documenting common patterns of detectable barriers using multi-tool triangulation; and (3) providing empirical evidence on the relationship between accessibility and performance optimization in state government web development.\u003c/p\u003e \u003cp\u003eState government homepages are critical gateways to services, often receiving prioritized maintenance and high visibility, making them valuable for focused automated screening as indicators of institutional practices. However, they represent only the entry point in complex digital ecosystems; results cannot generalize to deeper service pages, subdomains, mobile apps, or documents, where barriers may be more severe.\u003c/p\u003e \u003cp\u003eAutomated tools enable scalable, reproducible analysis but detect only a subset of WCAG 2.1 Level AA criteria (typically 13\u0026ndash;30%), missing issues requiring human judgment, such as meaningful alternative text or logical content organization. This study thus offers a preliminary, descriptive baseline of tool-detected issues on homepages, highlighting relative strengths, common detectable barriers, and the observed disconnect between accessibility metrics and performance. It underscores the value of automated screening as a starting point while emphasizing the necessity of manual expert review and user testing for comprehensive assessments aligned with the upcoming deadlines.\u003c/p\u003e"},{"header":"2. Policy and Regulatory Context","content":"\u003cp\u003eDigital accessibility in the United States is governed by civil rights laws, technical standards, and regulatory mandates that ensure equal access to public services for individuals with disabilities. For state governments, the primary legal foundations are Title II of the Americans with Disabilities Act (ADA) and Section 504 of the Rehabilitation Act. Title II prohibits discrimination by state and local governments in any program or service, including digital offerings such as websites and mobile applications. Section 504 imposes similar requirements on entities receiving federal financial assistance. Together, these laws establish that inaccessible digital services can constitute discrimination [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHistorically, enforcement relied on case-by-case settlements that often referenced the Web Content Accessibility Guidelines (WCAG), but no specific technical standard was federally mandated for state and local governments. This fragmented approach meant that accessibility outcomes depended heavily on institutional leadership, advocacy pressure, or litigation history rather than consistent regulatory expectations [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. This inconsistency produced wide variation in accessibility outcomes: states that had faced litigation or advocacy pressure often achieved better results, while others maintained barriers for years without consequence. The lack of uniform standards also created confusion for web developers and procurement officers, who lacked clear technical benchmarks to guide implementation or vendor selection.\u003c/p\u003e \u003cp\u003eThe U.S. Department of Justice's 2024 final rule under ADA Title II changed this by formally adopting WCAG 2.1 Level AA as the required technical standard for state and local government websites, mobile applications, and digital documents [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. The rule establishes tiered compliance deadlines: jurisdictions serving populations of 50,000 or more must achieve conformance by April 2026, while smaller jurisdictions have until April 2027. It also mandates accessibility policies, designated coordinators, grievance procedures, and accessible feedback channels\u0026mdash;for the first time creating uniform, enforceable requirements with explicit timelines.\u003c/p\u003e \u003cp\u003eDeveloped by the World Wide Web Consortium (W3C), WCAG 2.1 defines testable success criteria organized around four principles: perceivable, operable, understandable, and robust [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Level AA incorporates all Level A criteria plus additional ones addressing common barriers, including missing alternative text, insufficient color contrast (minimum 4.5:1 for normal text, 3:1 for large text), inaccessible forms, improper heading structures, keyboard traps, and lack of visible focus indicators. Published in 2018, WCAG 2.1 extended WCAG 2.0 by adding 17 new criteria focused on mobile accessibility, low vision needs, and cognitive and learning disabilities.\u003c/p\u003e \u003cp\u003eThe shift from voluntary guidelines to mandatory regulatory standards represents a fundamental change in U.S. digital accessibility governance. Under the new rule, accessibility is no longer merely aspirational or dependent on advocacy action\u0026mdash;it becomes a legal floor below which states cannot fall without risking enforcement. This regulatory transformation places accessibility alongside other civil rights protections with concrete remedies for noncompliance. States must now balance competing demands: addressing legacy accessibility debt accumulated over years of fragmented requirements, maintaining accessibility as websites evolve with new content and features, building internal capacity through training and hiring, and allocating sufficient budgets to what is often perceived as a niche technical concern rather than a core civil rights obligation.\u003c/p\u003e \u003cp\u003eLarge-scale evaluations of public-sector websites worldwide frequently identify these barriers. Studies of government portals in India, Nepal, Saudi Arabia, and Nigeria consistently report issues such as incorrect heading hierarchies, contrast failures, unlabeled elements, and inconsistent markup [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Even in regions with earlier regulations, such as the European Union under the Web Accessibility Directive, implementation has been mixed [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Common challenges include limited organizational capacity, insufficient training, competing priorities, and treating accessibility as a one-time project rather than an ongoing practice.\u003c/p\u003e \u003cp\u003eIn the United States, empirical research on e-government accessibility remains limited. A recent study examining 64 federal, state, and local websites found widespread WCAG 2.1 non-conformance detectable through automated and manual methods [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Earlier work documented persistent violations under Section 508 on federal sites [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. These findings suggest that legal requirements alone are insufficient without supporting governance, training, and enforcement mechanisms.\u003c/p\u003e \u003cp\u003eWeb performance is increasingly recognized as intersecting with accessibility. Slow loading, unstable layouts, or delayed interactivity\u0026mdash;captured by Google's Core Web Vitals (Largest Contentful Paint for loading, Interaction to Next Paint for responsiveness, Cumulative Layout Shift for visual stability, and Time to First Byte for server response)\u0026mdash;can disproportionately affect users of assistive technologies [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Screen reader users may face delays in feedback; keyboard navigators can lose focus during layout shifts; and individuals with cognitive disabilities may struggle with instability or prolonged waiting [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Performance challenges are compounded for users on mobile devices with limited processing power or data plans [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Despite this theoretical linkage, empirical studies exploring the relationship between performance optimization and accessibility in government contexts are scarce.\u003c/p\u003e \u003cp\u003eWith the 2026\u0026ndash;2027 ADA Title II deadlines approaching, preliminary automated screenings of detectable issues on high-visibility state government homepages can offer descriptive insights into common barriers and performance characteristics. Such data, while limited in scope, may help identify patterns and inform initial prioritization efforts as states prepare for broader evaluations involving manual review and user testing.\u003c/p\u003e"},{"header":"3. Related Work","content":"\u003cp\u003eResearch on e-government accessibility has grown with the increasing reliance on digital public services. Systematic reviews show that studies predominantly use automated WCAG-based tools, sometimes supplemented by expert review or user testing [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Automated methods offer scalability and reproducibility for large-scale evaluations, but they consistently detect only a limited subset of WCAG success criteria\u0026mdash;typically 20\u0026ndash;30% of objectively measurable issues\u0026mdash;while missing those requiring human judgment, such as the meaningfulness of alternative text or logical reading order [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eResearch consistently demonstrates this fundamental limitation: tools excel at detecting technical patterns\u0026mdash;missing alt attributes, incorrect ARIA roles, insufficient contrast ratios\u0026mdash;but cannot assess whether alternative text meaningfully conveys information, whether heading structures create logical document organization, or whether forms provide clear error recovery guidance [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] This creates a persistent tension in accessibility research: automated methods enable population-level studies but capture only a fraction of user experience, while manual and user-testing methods provide rich insights but lack scalability. The field has yet to resolve this trade-off satisfactorily, with most large-scale studies accepting automated detection as a proxy for accessibility quality while acknowledging substantial undercounting.\u003c/p\u003e \u003cp\u003eCommon detectable barriers across global evaluations include missing alternative text, improper heading structures, low color contrast, and inaccessible forms [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. These issues recur consistently, indicating systemic gaps in web development practices.\u003c/p\u003e \u003cp\u003eCountry-level studies highlight ongoing challenges. In India, government websites frequently exhibit heading hierarchy errors, contrast failures, and unlabeled elements [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Similar patterns appear in Nepal's federal portals [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] and Saudi Arabia's mobile e-government sites [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Nigerian platforms demonstrate lagged performance alongside accessibility deficiencies [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Even in regulated contexts, such as the European Union under the Web Accessibility Directive, results remain mixed [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe consistency of barrier types across diverse international contexts\u0026mdash;different regulatory regimes, development practices, languages, and government structures\u0026mdash;suggests these are not context-specific failures but fundamental challenges in how accessibility knowledge translates into implementation. Possible explanations include: accessibility training often absent from computer science and web development curricula; accessibility treated as post-hoc remediation rather than integrated design practice; procurement processes that fail to require or verify vendor accessibility claims; rapid staff turnover disrupting institutional knowledge; and the \"invisible\" nature of accessibility barriers to developers without disabilities who may not encounter problems during testing. International research thus reveals not just widespread non-conformance, but systemic obstacles to achieving and maintaining accessibility that legal mandates alone do not overcome.\u003c/p\u003e \u003cp\u003eMethodological variation is notable internationally, with some studies incorporating manual reviews or user testing with disabled participants, and others exploring organizational factors. Few, however, integrate accessibility with performance metrics, despite recognition that both influence real-world usability [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eU.S.-specific empirical research is limited. A prior study examined 64 federal, state, and local e-government websites using automated tools and manual inspection, identifying widespread issues in ARIA attributes, structural markup, and perceptual barriers; automated scans alone underestimated problems [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Earlier work on federal sites under Section 508 found persistent violations, with homepages often outperforming deeper pages [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe interplay between accessibility and performance is increasingly acknowledged. Performance issues can compound barriers for assistive technology users (e.g., delays for screen readers, focus loss from layout shifts), particularly on mobile devices [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Socioeconomic factors further intersect with disability, amplifying exclusion [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Yet empirical integrations of Core Web Vitals with accessibility evaluations remain rare.\u003c/p\u003e \u003cp\u003eThe scarcity of integrated accessibility-performance research is particularly notable given increasing recognition that modern web development often treats these as competing priorities. Performance optimization\u0026mdash;code minification, lazy loading, client-side rendering\u0026mdash;can introduce accessibility barriers if not carefully implemented, while accessibility enhancements like verbose ARIA labels or redundant text alternatives might theoretically impact performance metrics. Whether this represents a genuine technical trade-off or merely separate organizational silos managing independent concerns remains empirically unresolved in government contexts.\u003c/p\u003e \u003cp\u003eThis study builds on prior work through multi-tool automated screening (AccessScan and WAVE) combined with Core Web Vitals analysis on the homepages of all 50 U.S. state government websites. Conducted in December 2025, it provides a descriptive snapshot of detectable issues and performance metrics on these high-visibility entry points. Unlike sampled approaches, the population-level coverage avoids sampling bias. The performance integration allows preliminary exploration of their empirical relationship in a government context. However, as an automated, homepage-only evaluation, it serves as a preliminary screening rather than a comprehensive assessment, highlighting the ongoing need for manual and user-centered methods.\u003c/p\u003e"},{"header":"4. Methods","content":"\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Study Design\u003c/h2\u003e \u003cp\u003eThis study employs a cross-sectional, descriptive design to conduct an automated screening of detectable accessibility issues and performance metrics on the primary homepages of all 50 U.S. state government websites. The focus is limited to homepages, enabling population-level coverage with standardized methodology. Homepages represent high-visibility entry points but only one component of broader digital ecosystems; findings do not extend to deeper service pages, subdomains, mobile applications, or documents.\u003c/p\u003e \u003cp\u003eThree automated tools were used: AccessScan and WAVE for detecting WCAG-aligned accessibility issues, and Google PageSpeed Insights (PSI) for Core Web Vitals performance. Automated tools were selected for their scalability, reproducibility, and alignment with prior large-scale evaluations [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAutomated evaluation has significant limitations. Tools reliably detect only a subset of WCAG 2.1 Level AA success criteria (typically 13\u0026ndash;30%), excelling at objective technical patterns but unable to assess issues requiring human judgment, such as the meaningfulness of alternative text, logical content organization, or real-world usability with assistive technologies [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Results thus represent preliminary indications of detectable issues based on tool heuristics, not comprehensive WCAG conformance or overall accessibility quality.\u003c/p\u003e \u003cp\u003eTriangulation across multiple tools with differing algorithms strengthens reliability where outputs converge, consistent with best practices for automated screening [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Website Selection\u003c/h2\u003e \u003cp\u003eOfficial primary homepages for all 50 states were selected (e.g., \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.alabama.gov\u003c/span\u003e\u003cspan address=\"https://www.alabama.gov\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://portal.arkansas.gov\u003c/span\u003e\u003cspan address=\"https://portal.arkansas.gov\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), verified as main public gateways via authoritative sources including USA.gov and state communications. All used HTTPS and were publicly accessible without authentication.\u003c/p\u003e \u003cp\u003eOnly homepages were evaluated for consistency and feasibility, given variability in state web ecosystem size and structure. Homepages often receive prioritized maintenance, making them suitable for focused automated screening as indicators of institutional practices, though not representative of full sites.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Tools Used\u003c/h2\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e4.3.1 AccessScan\u003c/h2\u003e \u003cp\u003eAccessScan (developed by accessiBe) is a free, AI-powered automated tool that scans a single webpage for detectable WCAG 2.1 Level AA issues. It evaluates across 11 categories: clickables, titles, orientation, menus, graphics, forms, documents, readability, carousels, tables, and general practices. Each category receives a score (0\u0026ndash;100), with \"Neutral\" or \"N/A\" for inapplicable elements.\u003c/p\u003e \u003cp\u003eThe tool provides an overall proprietary rating (\"Accessible\" or \"Non-compliant\" in this dataset; \"Semi-compliant\" possible but not observed). AccessScan detects approximately 30% of WCAG criteria, focusing on programmatic issues; it lacks independent peer-reviewed validation and cannot evaluate content quality or complex interactions.\u003c/p\u003e \u003cp\u003eAccessScan is a free tool offered by accessiBe, a commercial accessibility solutions provider. The tool's detection algorithms and validation methodology are proprietary and not publicly documented. No independent peer-reviewed studies have validated AccessScan's accuracy against manual WCAG audits or compared its detection rates to other established tools. Despite these limitations, we selected AccessScan because: it provides scalable scanning suitable for population-level evaluation; its categorical scoring enables comparative analysis; and its different heuristics complement WAVE's approach, supporting triangulation. Readers should interpret AccessScan's \"Accessible\" and \"Non-compliant\" ratings as reflecting the tool's proprietary algorithms rather than verified WCAG 2.1 Level AA conformance. These ratings indicate relative performance on detectable programmatic issues, not comprehensive accessibility quality.4.3.2 WAVE (Web Accessibility Evaluation Tool)\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e4.3.2 WAVE (Web Accessibility Evaluation Tool)\u003c/h2\u003e \u003cp\u003eWAVE, developed by WebAIM at Utah State University, identifies WCAG violations via automated rule-checking. It reports Errors (clear violations), Contrast Errors, Alerts (potential issues needing verification), Features (enhancements), Structural Elements, and ARIA usage.\u003c/p\u003e \u003cp\u003eWAVE computes a proprietary Accessibility Impact Measure (AIM) Score, weighting issues by estimated user impact (higher\u0026thinsp;=\u0026thinsp;better). Data were unavailable for three states (New Hampshire, Texas, West Virginia) due to technical blocks (e.g., \"Access Denied\" errors, common with WAVE's proxy-based scanning). Sensitivity checks confirmed exclusion does not materially alter descriptive statistics (e.g., mean Errors\u0026thinsp;~\u0026thinsp;3.9 with/without). Sample: n\u0026thinsp;=\u0026thinsp;47.\u003c/p\u003e \u003cp\u003eWAVE is more extensively documented than AccessScan, with research examining its detection capabilities, though comprehensive validation studies remain limited. WAVE reliably identifies programmatic violations such as missing alt attributes, form label associations, and color contrast failures calculated against WCAG numeric thresholds (4.5:1 for normal text, 3:1 for large text). However, like all automated tools, WAVE cannot assess whether alternative text is meaningful, whether content organization is logical, or whether interactions are usable with assistive technologies. The three excluded states (New Hampshire, Texas, West Virginia) likely employ server-side security configurations that block WAVE's proxy-based scanning\u0026mdash;a known limitation when evaluating sites with restrictive access policies. We attempted scans on multiple days across the collection period and confirmed consistent failures, indicating persistent technical barriers rather than temporary issues.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003e4.3.3 Google PageSpeed Insights (PSI)\u003c/h2\u003e \u003cp\u003ePSI uses field data from the Chrome User Experience Report and lab-based Lighthouse testing to assess Core Web Vitals separately for mobile and desktop.\u003c/p\u003e \u003cp\u003eMetrics include Largest Contentful Paint (LCP\u0026thinsp;\u0026le;\u0026thinsp;2.5s\u0026thinsp;=\u0026thinsp;good), Interaction to Next Paint (INP\u0026thinsp;\u0026le;\u0026thinsp;200ms\u0026thinsp;=\u0026thinsp;good), Cumulative Layout Shift (CLS\u0026thinsp;\u0026le;\u0026thinsp;0.1\u0026thinsp;=\u0026thinsp;good), and Time to First Byte (TTFB\u0026thinsp;\u0026le;\u0026thinsp;0.8s\u0026thinsp;=\u0026thinsp;good), classified as \"good,\" \"needs improvement,\" or \"poor.\"\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Data Collection Procedures\u003c/h2\u003e \u003cp\u003eData were collected over multiple days in mid-December 2025 using consistent hardware, high-speed internet, and current tool versions in clean browser sessions during business hours to minimize variability.\u003c/p\u003e \u003cp\u003eProtocol per homepage: (1) Single AccessScan run; (2) WAVE analysis (manual metric extraction where successful); (3) PSI runs for mobile and desktop. Results compiled into a master dataset (Accessibility_Readiness_All_States.xlsx) without overrides (n\u0026thinsp;=\u0026thinsp;50 for AccessScan/PSI; n\u0026thinsp;=\u0026thinsp;47 for WAVE).\u003c/p\u003e \u003cp\u003eData were collected December 12\u0026ndash;18, 2025 (7 consecutive days), using consistent hardware (MacBook Pro, 16GB RAM), browser (Chrome version 131.0.6778.109), and network (residential fiber connection, 150\u0026thinsp;+\u0026thinsp;Mbps verified via speedtest.net). Tool versions: AccessScan web interface as current on December 12, 2025 (version number not publicly displayed by tool); WAVE browser extension version 3.2.6; Google PageSpeed Insights API using Lighthouse version 11.4.0. All evaluations occurred during weekday business hours (9 AM\u0026ndash;5 PM Eastern Time) to capture typical server loads and avoid maintenance windows.\u003c/p\u003e \u003cp\u003e \u003cb\u003eStandardized Protocol per Homepage\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eClear all browser cache, cookies, and site data\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eOpen new private/incognito browsing window to prevent cached content or saved preferences\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eNavigate directly to state homepage URL by typing or pasting (no search engines or redirects)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAllow complete page load: minimum 15 seconds, visual confirmation that primary content, images, and navigation are fully rendered; JavaScript-dependent elements loaded\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAccessScan: Copy URL into AccessScan web interface at accessibe.com/accessscan, initiate automated scan, wait for complete analysis (45\u0026ndash;90 seconds depending on page complexity), record all 11 categorical scores (numerical values 0\u0026ndash;100, or Neutral/N/A for inapplicable categories), overall rating (Accessible/Semi-compliant/Non-compliant), and completion timestamp\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWAVE: With page still loaded in browser, activate WAVE extension, allow analysis to complete (10\u0026ndash;30 seconds), manually transcribe all metrics into standardized Excel template: Errors (count), Contrast Errors (count), Alerts (count), Features (count), Structural Elements (count), ARIA (count), AIM Score (decimal value)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePageSpeed Insights Mobile: Enter URL into PSI web interface (pagespeed.web.dev), select Mobile device category, initiate analysis, wait for complete field\u0026thinsp;+\u0026thinsp;lab data processing (2\u0026ndash;4 minutes), record all four Core Web Vitals: LCP (seconds), INP (milliseconds), CLS (score), TTFB (seconds), plus classification (good/needs improvement/poor) for each\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePageSpeed Insights Desktop: Repeat step 7 selecting Desktop device category\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eDocument any anomalies: tool errors, page loading failures, timeouts, security warnings, or unexpected behavior in research log\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eTranscription Verification: For WAVE manual extraction, a second researcher independently extracted metrics for a 10% random subsample (5 states selected via random number generator: Alabama, Florida, Montana, Ohio, Wisconsin). Inter-rater agreement was 100% across all numeric fields, confirming reliable manual transcription procedures.\u003c/p\u003e \u003cp\u003eFailed Attempts: For tools producing errors, we made three separate attempts across different days. The three WAVE failures (New Hampshire, Texas, West Virginia) persisted across all nine total attempts (3 states \u0026times; 3 attempts), producing \"Access Denied\" (NH, WV) or connection timeout (TX) errors. This confirmed systematic technical blocks rather than transient issues. These states were coded as missing data for WAVE metrics.\u003c/p\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e4.4.1 Inter-Tool Comparison Methodology:\u003c/h2\u003e \u003cp\u003eAccessScan, WAVE, and PSI employ fundamentally different methodologies: AccessScan uses proprietary categorical scoring across predefined categories; WAVE counts specific violation types and weights them via a proprietary impact algorithm; PSI measures performance using field data from real Chrome users plus lab-based simulations. These tools detect partially overlapping but non-identical issue sets using incompatible measurement scales.\u003c/p\u003e \u003cp\u003eWe deliberately did not create composite \"overall accessibility scores\" by combining tools because: (1) no validated weighting scheme exists for aggregating proprietary algorithms; (2) tools measure conceptually distinct constructs (AccessScan's categorical adequacy vs. WAVE's violation density vs. PSI's performance thresholds); (3) different scales (0\u0026ndash;100 categorical scores, raw error counts, time-based metrics) cannot be meaningfully averaged; and (4) mathematical combination would obscure important tool-specific patterns.\u003c/p\u003e \u003cp\u003eInstead, we examine convergence (where tools agree) and divergence (where tools disagree) descriptively. Convergent findings\u0026mdash;such as sites rated \"Accessible\" by AccessScan also averaging fewer WAVE errors\u0026mdash;increase confidence in relative differences. Divergent findings\u0026mdash;such as zero-WAVE-error sites still rated \"Non-compliant\" by AccessScan\u0026mdash;reveal different detection sensitivities and underscore automated evaluation's inherent limitations. Both patterns are scientifically informative: convergence suggests robust signals; divergence reveals tool-specific detection biases and measurement error.\u003c/p\u003e \u003cp\u003eWe present all tool-specific results separately and transparently, noting patterns of agreement or disagreement without privileging any single tool as \"ground truth.\"\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e4.5 Metrics\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e4.5.1 Accessibility Metrics\u003c/h2\u003e \u003cp\u003eAccessScan and WAVE provided complementary detectable issues: categorical scores/ratings (AccessScan), error/alert counts (WAVE). These capture programmatic violations (e.g., contrast, headings, ARIA) but not full WCAG scope.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e4.5.2 Performance Metrics\u003c/h2\u003e \u003cp\u003ePSI provided Core Web Vitals values and classifications for mobile/desktop, indicating loading, interactivity, stability, and responsiveness\u0026mdash;factors that can compound barriers for assistive technology users.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e4.6 Data Cleaning, Validation, and Sensitivity\u003c/h2\u003e \u003cp\u003eURLs were verified; outputs standardized; numeric ranges checked; anomalies re-tested. Missing WAVE data coded explicitly. Sensitivity analysis for WAVE exclusions showed negligible impact on descriptives.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e4.7 Analysis Approach\u003c/h2\u003e \u003cp\u003eThis study evaluates the complete population of 50 U.S. state government homepages (47 for WAVE metrics), not a sample drawn from a larger population. All analyses are therefore descriptive rather than inferential. Reported statistics (means, medians, ranges, frequency distributions, correlation coefficients) describe actual population parameters, not sample estimates requiring generalization.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eDescriptive Correlations\u003c/strong\u003e \u003cp\u003ePearson correlation coefficients (r) quantify linear associations between metrics within this specific population. We interpret |r| \u0026lt; 0.3 as weak, 0.3\u0026ndash;0.6 as moderate, and \u0026gt;\u0026thinsp;0.6 as strong association, following conventional descriptive guidelines. These correlations describe observed patterns among the 50 states; they are not inferential statistics testing null hypotheses or estimating population parameters, as the data constitute the entire population of interest.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNo Inferential Statistics\u003c/strong\u003e \u003cp\u003eWe do not report p-values, confidence intervals, or significance tests because these are inappropriate for complete population data. The patterns we observe are the actual patterns in the population, not estimates subject to sampling error.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eRationale for Descriptive Approach\u003c/b\u003e: We selected descriptive rather than predictive analytical approaches because: (1) the study aims to document current conditions (establish a baseline) rather than test causal hypotheses; (2) small population size (n\u0026thinsp;=\u0026thinsp;50/47) limits power for complex modeling; (3) automated tool outputs have unknown measurement error and reliability properties, precluding precise parameter estimation; and (4) the primary goal is establishing a reference point for future longitudinal comparison rather than explaining variance through predictive models.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eMissing Data\u003c/strong\u003e \u003cp\u003eWAVE metrics are unavailable for 3 states (New Hampshire, Texas, West Virginia; n\u0026thinsp;=\u0026thinsp;47). We report this explicitly throughout and conducted sensitivity analysis confirming that excluding these states does not materially alter descriptive statistics (e.g., mean Errors remain\u0026thinsp;~\u0026thinsp;3.9 whether calculated with n\u0026thinsp;=\u0026thinsp;47 or estimating the three missing states at population mean values).\u003c/p\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e4.8 Data Availability\u003c/h2\u003e \u003cp\u003eThe complete dataset (Accessibility_Readiness_All_States.xlsx) containing all raw tool outputs for all 50 states will be deposited in a public data repository (Zenodo or equivalent) upon publication and assigned a permanent Digital Object Identifier (DOI). The dataset will include: state identifiers, homepage URLs, all AccessScan categorical scores and ratings, all WAVE metrics (with missing data coded explicitly), all PSI Core Web Vitals values and classifications for mobile and desktop, data collection dates, and a detailed codebook documenting variable definitions and measurement procedures. This enables full reproducibility and supports future comparative research.\u003c/p\u003e \u003c/div\u003e"},{"header":"5 Results","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e5.1 AccessScan Compliance Overview\u003c/h2\u003e \u003cp\u003eAccessScan\u0026rsquo;s proprietary algorithm rated 45 of 50 state homepages (90%) as \u0026ldquo;Non-compliant\u0026rdquo; and 5 (10%) as \u0026ldquo;Accessible\u0026rdquo; (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). No sites received a \u0026ldquo;Semi-compliant\u0026rdquo; rating. The five homepages rated \u0026ldquo;Accessible\u0026rdquo; were Arkansas, Indiana, Mississippi, North Carolina, and Pennsylvania (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThese ratings reflect AccessScan\u0026rsquo;s internal heuristics rather than verified WCAG 2.1 Level AA conformance. Sub-scores showed variability (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Mean scores ranged from 68.1 (titles) to 94.9 (readability). The titles category was notably weak (mean 68.1, minimum 33), indicating frequent deficiencies in heading structure. Lowest minimum scores occurred in clickables (25) and titles (33).\u003c/p\u003e \u003cp\u003eCategories such as menus, carousels, and tables were often \u0026ldquo;Neutral,\u0026rdquo; reflecting limited applicability on many homepages. The five \u0026ldquo;Accessible\u0026rdquo;-rated homepages had consistently higher sub-scores in most categories.\u003c/p\u003e \u003cp\u003eEven these higher-rated homepages showed residual issues in WAVE (mean 2.4 errors, range 1\u0026ndash;5), highlighting differences in tool heuristics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAccessScan Compliance Distribution\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCompliance Status\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCount\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePercentage\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNon-compliant\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e90%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAccessible\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. \u003cem\u003eAccessScan overall ratings distribution for the homepages of all 50 U.S. state governments. Ninety percent were rated \u0026ldquo;Non-compliant\u0026rdquo; by AccessScan\u0026rsquo;s proprietary algorithm, while 10% (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) were rated \u0026ldquo;Accessible.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of AccessScan Numeric Sub-Scores (n\u0026thinsp;=\u0026thinsp;50; Neutral excluded)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN (numeric)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMin\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eMax\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClickables\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTitles\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOrientation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e~\u0026thinsp;20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e80.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGraphics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e86.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eForms\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e79.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDocument\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e~\u0026thinsp;50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e86.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReadability\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e~\u0026thinsp;50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e94.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003e(Note: Exact N/medians vary slightly for some categories; menus numeric in only 11 states.)\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e5.2 WAVE Accessibility Issues\u003c/h2\u003e \u003cp\u003eWAVE data were available for 47 states. Mean errors were 3.9 (median 2, range 0\u0026ndash;17), with 14 states (30%) showing zero errors (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Common errors included missing alternative text, empty links, missing form labels, broken ARIA references, and ambiguous link text.\u003c/p\u003e \u003cdiv id=\"Sec22\" class=\"Section3\"\u003e \u003ch2\u003e5.2.1 Specific Violation Patterns\u003c/h2\u003e \u003cp\u003eBeyond overall error counts, examining specific violation types reveals which barriers are most prevalent across state homepages.\u003c/p\u003e \u003cp\u003eAlternative text issues were among the most common detectable barriers. Missing or problematic alt text appeared in various forms: decorative images incorrectly given descriptive alt text (when they should have empty alt=\"\"), linked images with generic descriptions (\"image,\" \"logo,\" \"icon\"), and complex infographics or data visualizations lacking adequate text alternatives that convey equivalent information to screen reader users.\u003c/p\u003e \u003cp\u003eForm labeling deficiencies affected many states, particularly for search inputs and newsletter signup forms common on government homepages. Problematic patterns included: placeholder text used as a substitute for proper label elements (placeholders disappear when users type, creating memory burden), visual labels not programmatically associated with form inputs via for/id relationships, and missing fieldset/legend groupings for radio button sets or checkboxes.\u003c/p\u003e \u003cp\u003eHeading structure problems (AccessScan titles mean 68.1, minimum 33) manifested in several ways: skipped heading levels (jumping from H1 directly to H3 or H4, disrupting screen reader navigation), multiple H1 elements on a single page (confusing document structure), and headings used solely for visual styling rather than semantic document organization. These errors fundamentally undermine screen reader users' ability to understand page structure and navigate efficiently by jumping between sections.\u003c/p\u003e \u003cp\u003eContrast failures (mean 8.4, median 1, maximum 141) showed extreme variability. While 15 states (32%) had zero detectable violations, Oklahoma's 141 contrast errors indicate pervasive problems throughout its homepage. Common patterns included: light gray text on white backgrounds in navigation elements, footers, and disclaimers; insufficient contrast in button and interactive element styling; and link colors too similar to surrounding body text, making them difficult to distinguish for users with low vision or color blindness.\u003c/p\u003e \u003cp\u003eARIA implementation issues included: empty or generic aria-label values that provide no meaningful information, broken aria-describedby or aria-labelledby references pointing to non-existent element IDs, redundant ARIA attributes on native semantic HTML elements (adding unnecessary complexity), and missing aria-live regions for dynamic content updates that screen readers cannot otherwise detect.\u003c/p\u003e \u003cp\u003eContrast errors were skewed (mean 8.4, median 1, maximum 141); 15 states (32%) had none. Structural elements (mean 48.4) and ARIA usage (mean 86.8) were abundant, though presence does not ensure correctness. Alerts averaged 17.6; features averaged 20.8.\u003c/p\u003e \u003cp\u003eMean WAVE AIM score was 7.7 (median 7.8, range 2\u0026ndash;9.9), with 11 states\u0026thinsp;\u0026ge;\u0026thinsp;9.0 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Errors and contrast errors inversely correlated descriptively with AIM score (r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.63 and \u0026minus;\u0026thinsp;0.72).\u003c/p\u003e \u003cp\u003eThe five AccessScan \"Accessible\"-rated homepages averaged 2.4 WAVE errors (vs. overall mean 3.9), showing partial cross-tool consistency, but none were error-free in WAVE.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eWAVE Metrics Summary (n\u0026thinsp;=\u0026thinsp;47)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMin\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMax\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e% Zero\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eErrors\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e~\u0026thinsp;30%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eContrast Errors\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e141\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e32%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAlerts\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFeatures\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e20.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e155\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStructure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e48.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e310\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eARIA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e86.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e467\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAIM Score\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e9.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. \u003cem\u003eBoxplot of WAVE AIM Scores for 47 U.S. state government homepages (higher score indicates better quality according to WAVE\u0026rsquo;s proprietary automated metrics). Median\u0026thinsp;~\u0026thinsp;7.8; interquartile range\u0026thinsp;~\u0026thinsp;6.9\u0026ndash;8.9; 11 states\u0026thinsp;\u0026ge;\u0026thinsp;9.0; one low outlier at 2.0.\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003e5.2.2 Comparative Analysis of Five \"Accessible\"-Rated Homepages\u003c/h2\u003e \u003cp\u003eArkansas, Indiana, Mississippi, North Carolina, and Pennsylvania\u0026mdash;the five states rated \"Accessible\" by AccessScan\u0026mdash;represent diverse geographic regions (South: AR, MS, NC; Midwest: IN; Mid-Atlantic: PA) and population sizes (ranging from Mississippi's\u0026thinsp;~\u0026thinsp;3\u0026nbsp;million to Pennsylvania's\u0026thinsp;~\u0026thinsp;13\u0026nbsp;million residents). Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents detailed metrics for these states compared to population means.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDetailed Metrics for Five AccessScan \"Accessible\"-Rated States\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eState\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eClickables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTitles\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGraphics\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eForms\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eWAVE Errors\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eContrast Errors\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eWAVE AIM\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eMobile LCP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eDesktop LCP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArkansas\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e7.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e2.6s\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e2.6s\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIndiana\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e6.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e1.6s\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1.6s\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMississippi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e5.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e2.3s\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e2.3s\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNorth Carolina\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e8.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e1.8s\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1.8s\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePennsylvania\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e8.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e1.7s\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e1.7s\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePopulation Mean\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e68.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e68.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e86.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e79.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e3.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e8.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e7.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e2.3s\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e2.0s\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eCommon strengths\u003c/strong\u003e \u003cp\u003eAll five achieved perfect scores (100) in AccessScan's Titles category, indicating properly structured heading hierarchies with no skipped levels, appropriate nesting, and semantic use. Four of five (except Arkansas at 99) scored\u0026thinsp;\u0026ge;\u0026thinsp;86 in Clickables, demonstrating well-labeled interactive elements with meaningful link text and button labels. All scored\u0026thinsp;\u0026ge;\u0026thinsp;75 in Forms (four at 100), suggesting accessible form implementations with proper labeling and fieldset usage.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eResidual issues despite \"Accessible\" rating\u003c/strong\u003e \u003cp\u003eAll five showed detectable problems in WAVE analysis (errors range 1\u0026ndash;5, mean 2.4 vs. population mean 3.9), demonstrating that no automated tool captures all issues and that even relatively stronger sites have remaining barriers\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eArkansas\u003c/b\u003e (1 error, 0 contrast errors, AIM 7.6): Minimal WAVE errors with strong contrast performance, though its Forms score (75) was notably lower than the other four \"Accessible\" states\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eIndiana\u003c/b\u003e (2 errors, 0 contrast errors, AIM 6.7): Zero contrast violations but a relatively lower AIM score, suggesting other weighted issues; strongest Core Web Vitals (1.6s LCP on both mobile/desktop)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eMississippi\u003c/b\u003e (2 errors, 30 contrast errors, AIM 5.2): Anomalous pattern\u0026mdash;rated \"Accessible\" by AccessScan with perfect scores across categories, yet had 30 contrast failures in WAVE and the lowest AIM score (5.2) among the five, illustrating tool divergence\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eNorth Carolina\u003c/b\u003e (2 errors, 0 contrast errors, AIM 8.9): Highest WAVE AIM score among the five, indicating strong overall automated accessibility quality; good performance (1.8s LCP)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003ePennsylvania\u003c/b\u003e (5 errors, 0 contrast errors, AIM 8.5): Highest WAVE error count among the five but still well below population mean; strong AIM score and excellent performance (1.7s LCP)\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003ePerformance variability\u003c/b\u003e: The five showed mixed Core Web Vitals, with mobile LCP ranging from 1.6s (Indiana, excellent) to 2.6s (Arkansas, approaching \"needs improvement\" threshold). This variability provides further evidence that accessibility and performance optimizations operate independently: Mississippi achieved perfect AccessScan scores but had moderate performance (2.3s LCP), while Indiana combined strong accessibility with exceptional loading speed.\u003c/p\u003e \u003cp\u003e \u003cb\u003eTool divergence\u0026mdash;the Mississippi anomaly\u003c/b\u003e: Mississippi's case particularly illustrates automated tool limitations. AccessScan rated it \"Accessible\" with perfect sub-scores (100/100/100/100), yet WAVE detected 30 contrast errors and assigned a relatively low AIM score (5.2). This suggests either: (1) AccessScan's contrast detection differs from WAVE's WCAG-calculation method, (2) the tools prioritize different elements, or (3) temporal differences (state updated site between evaluations, though data were collected within one week). This divergence underscores that \"Accessible\" ratings reflect tool-specific heuristics rather than comprehensive conformance.\u003c/p\u003e \u003cp\u003e \u003cb\u003eWhat distinguishes these states?\u003c/b\u003e Without organizational case studies, technical patterns are instructive but limited. Observable commonalities: all five maintained consistent heading structures (perfect Titles scores), suggesting systematic attention to semantic HTML. Geographic and demographic diversity (small/large states, different regions) indicates that relatively stronger automated outcomes are not limited to wealthy, large, or technologically advanced states but may reflect organizational practices, governance structures, or technical leadership that could be replicated elsewhere.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Core Web Vitals Performance (Mobile vs. Desktop)\u003c/h2\u003e \u003cp\u003eAll 50 homepages had complete PSI data. Performance was strong overall, especially for INP and CLS (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMobile: 74% good LCP, 98% INP, 80% CLS, 76% TTFB; 27 states (54%) passed all four. Desktop: 76% LCP, 98% INP, 80% CLS, 84% TTFB; 31 states (62%) passed all four. Only 20 states (40%) passed all four on both mobile and desktop.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCore Web Vitals \"Good\" Pass Rates (Google Thresholds)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMobile Good (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDesktop Good (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLCP (\u0026le;\u0026thinsp;2.5s)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e76\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eINP (\u0026le;\u0026thinsp;200ms)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e98\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCLS (\u0026le;\u0026thinsp;0.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e80\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTTFB (\u0026le;\u0026thinsp;0.8s)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e84\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll 4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e54% (27 states)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e62% (31 states)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll 4 Both\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e40% (20 states)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. \u003cem\u003ePercentage of homepages (n\u0026thinsp;=\u0026thinsp;50) meeting \u0026ldquo;good\u0026rdquo; thresholds for Core Web Vitals on mobile and desktop. INP near-universal; CLS strong; LCP/TTFB lower, with desktop TTFB notably better.\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e5.4 Cross-Metric Patterns\u003c/h2\u003e \u003cp\u003eDescriptive associations between accessibility and performance metrics were weak. WAVE errors showed negligible descriptive correlations with LCP and TTFB (r\u0026thinsp;\u0026asymp;\u0026thinsp;0.04\u0026ndash;0.25). WAVE AIM score vs. mobile LCP yielded R\u0026sup2; \u0026asymp;0.045 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSome convergence appeared within accessibility tools: AccessScan clickables/titles sub-scores correlated modestly with WAVE AIM (r\u0026thinsp;=\u0026thinsp;0.12\u0026ndash;0.20) and inversely with WAVE errors (r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.36 to \u0026minus;\u0026thinsp;0.37).\u003c/p\u003e \u003cp\u003eNo homepage was free of detectable issues or perfectly optimized across metrics.\u003c/p\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003e5.4.1 Regional and Population-Size Patterns\u003c/h2\u003e \u003cp\u003eWe explored whether accessibility or performance metrics varied systematically by geographic region or state population size. States were categorized by Census region (Northeast, Southeast, Midwest, West) and population size (Small: \u0026lt;2M, Medium: 2-7M, Large: 7-15M, Very Large: \u0026gt;15M, based on approximate 2024 estimates).\u003c/p\u003e \u003cp\u003e \u003cb\u003eRegional patterns\u003c/b\u003e: Accessibility metrics showed minimal regional clustering. Mean WAVE AIM scores by region: Northeast 7.8 (n\u0026thinsp;=\u0026thinsp;9), Southeast 7.5 (n\u0026thinsp;=\u0026thinsp;12), Midwest 7.9 (n\u0026thinsp;=\u0026thinsp;12), West 7.6 (n\u0026thinsp;=\u0026thinsp;14). These small differences (range 0.4 points) are not meaningful given measurement error. The five \"Accessible\"-rated states span three regions (South: AR, MS, NC; Midwest: IN; Northeast: PA), with none in the West, though this likely reflects random variation rather than systematic regional differences.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003ePopulation size\u003c/strong\u003e \u003cp\u003eContrary to expectations that larger states with greater budgets and technical capacity might achieve better accessibility outcomes, we found no clear relationship. The correlation between state population and WAVE AIM score was negligible (r\u0026thinsp;=\u0026thinsp;0.08, nearly zero). The five \"Accessible\"-rated states include both smaller populations (Mississippi\u0026thinsp;~\u0026thinsp;3M, Arkansas\u0026thinsp;~\u0026thinsp;3M) and larger ones (Pennsylvania\u0026thinsp;~\u0026thinsp;13M), with mid-size Indiana (~\u0026thinsp;7M) and North Carolina (~\u0026thinsp;11M) falling between. Conversely, some very large states (California\u0026thinsp;~\u0026thinsp;39M, Texas\u0026thinsp;~\u0026thinsp;30M, Florida\u0026thinsp;~\u0026thinsp;23M) showed typical or below-average automated accessibility metrics, demonstrating that scale and resources do not automatically translate to stronger outcomes.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNotable examples\u003c/strong\u003e \u003cp\u003eWyoming (population\u0026thinsp;~\u0026thinsp;580,000, smallest state) achieved a WAVE AIM score of 6.2\u0026mdash;below the population mean of 7.7 but not dramatically so, with only 9 errors. Vermont (~\u0026thinsp;645,000) scored higher at 8.3 with 0 errors. California (largest state, ~\u0026thinsp;39M) had 0 errors but a moderate AIM of 8.8, while Texas (~\u0026thinsp;30M) could not be evaluated by WAVE due to technical blocks. These patterns suggest accessibility outcomes depend more on organizational practices, governance structures, procurement standards, staff expertise, and specific technical decisions than on state size or presumed resource availability.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eImplications\u003c/strong\u003e \u003cp\u003eThe absence of strong demographic predictors indicates that accessibility barriers are not simply resource problems solvable by larger budgets. Small states can achieve relatively strong outcomes (e.g., Vermont, Arkansas, Mississippi), and large states can show typical performance (California) or substantial issues (Oklahoma with 141 contrast errors despite ~\u0026thinsp;4M population providing reasonable resources). This points to the importance of governance, leadership prioritization, technical capacity building, and systematic quality assurance processes rather than scale alone.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. \u003cem\u003eScatterplot of WAVE AIM Score vs. mobile LCP (n\u0026thinsp;=\u0026thinsp;47), with trendline. Very weak negative association (R\u0026sup2; \u0026asymp;0.045) indicates virtually no meaningful relationship between these automated accessibility and performance indicators.\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e5.5 Summary\u003c/h2\u003e \u003cp\u003eAutomated evaluation detected accessibility issues on all 50 homepages, with 90% rated \"Non-compliant\" by AccessScan's proprietary algorithm and mean 3.9 WAVE errors (n\u0026thinsp;=\u0026thinsp;47). Common detectable barriers\u0026mdash;contrast failures (mean 8.4, ranging from 0 to 141), heading deficiencies (AccessScan titles mean 68.1, minimum 33), missing alternative text, form labeling issues, and ARIA implementation problems\u0026mdash;appeared consistently across diverse state contexts.\u003c/p\u003e \u003cp\u003eFive states (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) achieved AccessScan \"Accessible\" ratings with consistently higher sub-scores and fewer WAVE errors (mean 2.4), yet none were error-free. Tool divergence (particularly Mississippi's 30 contrast errors despite perfect AccessScan scores) illustrates that automated ratings reflect tool-specific detection heuristics rather than comprehensive WCAG conformance.\u003c/p\u003e \u003cp\u003ePerformance metrics were generally robust (INP 98% good, CLS 80% good), though only 40% of states passed all Core Web Vitals thresholds on both mobile and desktop. The very weak descriptive association between accessibility and performance indicators (R\u0026sup2;\u0026asymp;0.045 for WAVE AIM vs. mobile LCP) demonstrates these dimensions are optimized independently in current state implementations, with no systematic tendency for states excelling in one domain to excel in the other.\u003c/p\u003e \u003cp\u003eRegional and population-size analyses revealed no strong predictors of accessibility outcomes, suggesting that organizational practices, governance structures, and technical decisions matter more than geography or state scale.\u003c/p\u003e \u003c/div\u003e"},{"header":"6. Limitations","content":"\u003cp\u003eThis study has several important limitations that must be considered when interpreting the findings.\u003c/p\u003e \u003cp\u003eFirst, the evaluation relied exclusively on automated tools (AccessScan, WAVE, and Google PageSpeed Insights). While these tools enable scalable and reproducible detection of programmatically identifiable issues, they capture only a limited subset\u0026mdash;typically 13\u0026ndash;30%\u0026mdash;of WCAG 2.1 Level AA success criteria [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. They cannot assess issues requiring human judgment, such as the meaningfulness or relevance of alternative text, logical reading and focus order, effective error identification and recovery, or the usability of complex interactions. As a result, the reported issues and ratings likely underestimate the full range of barriers faced by users with disabilities.\u003c/p\u003e \u003cp\u003eSecond, the analysis was confined to each state\u0026rsquo;s primary government homepage. These entry-point pages often receive prioritized maintenance and greater visibility, but essential services (e.g., benefits applications, tax filing, unemployment insurance, licensing, voter registration) are typically located on deeper subpages, subdomains, or third-party platforms that may use different templates or legacy systems. Prior research indicates that accessibility tends to be weaker on such deeper pages [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. The findings therefore reflect only the \u0026ldquo;front door\u0026rdquo; of state digital services and cannot be generalized to broader web ecosystems, mobile applications, or documents.\u003c/p\u003e \u003cp\u003eThird, the data constitute a single snapshot collected in mid-December 2025. Government websites evolve continuously through content updates, redesigns, and technical changes, meaning accessibility and performance characteristics can shift rapidly. The results may not capture subsequent improvements or regressions.\u003c/p\u003e \u003cp\u003eFourth, no manual expert inspection or user testing with people with disabilities was performed. These methods are indispensable for identifying issues beyond automated detection and for evaluating real-world usability across diverse assistive technologies and user needs. Their absence limits the ecological validity of the findings.\u003c/p\u003e \u003cp\u003eFinally, differences across tools introduce additional caveats. AccessScan\u0026rsquo;s proprietary ratings and categorical scores, WAVE\u0026rsquo;s error taxonomy and AIM scoring, and PageSpeed Insights\u0026rsquo; performance thresholds employ distinct heuristics and rule sets. While multi-tool triangulation provides complementary perspectives, direct comparability is constrained, and no single tool represents ground truth.\u003c/p\u003e \u003cp\u003eThese limitations underscore that the study provides a preliminary screening of detectable issues on homepages using automated methods, not a comprehensive assessment of WCAG 2.1 Level AA conformance or overall digital accessibility. Fuller evaluation requires manual review, user testing, and broader scope.\u003c/p\u003e"},{"header":"7. Discussion","content":"\u003cp\u003eThis automated, homepage-only evaluation using AccessScan, WAVE, and Google PageSpeed Insights provides a preliminary descriptive snapshot of detectable accessibility issues and performance metrics on U.S. state government homepages in December 2025, ahead of the 2026\u0026ndash;2027 ADA Title II deadlines.\u003c/p\u003e \u003cdiv id=\"Sec30\" class=\"Section2\"\u003e \u003ch2\u003e7.1 What the Automated Results Reveal\u003c/h2\u003e \u003cp\u003eThe results indicate persistent detectable barriers across the population. AccessScan's proprietary algorithm rated 90% of homepages \"Non-compliant\" and 10% \"Accessible.\" The five \"Accessible\"-rated homepages (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) demonstrated consistently higher sub-scores in categories such as clickables, titles, graphics, and forms, and averaged fewer WAVE errors (2.4 vs. overall mean 3.9). Spanning diverse geographic regions and population sizes, these homepages suggest that stronger automated outcomes are achievable through more systematic implementation of basic programmatic practices\u0026mdash;e.g., proper heading hierarchies, labeled interactive elements, and sufficient contrast\u0026mdash;without necessarily requiring outsized resources or advanced technical infrastructure.\u003c/p\u003e \u003cp\u003eNevertheless, no homepage was free of detectable issues: even the higher-rated group had residual WAVE errors (range 1\u0026ndash;5), typically involving contrast failures, missing alternative text, or incomplete labeling. Across all 50 homepages, common patterns included highly skewed contrast errors (mean 8.4, with some extremes over 100), weak title/heading structure (AccessScan mean 68.1), and other fundamental violations that tools like WAVE and AccessScan reliably identify. These are not obscure or recently introduced criteria but core WCAG 2.1 Level AA requirements emphasizing perceivable and operable content.\u003c/p\u003e \u003cp\u003eIn contrast, performance metrics showed greater consistency and strength. Near-universal good ratings for INP (98%) and strong CLS (80%) on both mobile and desktop reflect effective adoption of contemporary web practices, such as minimized main-thread blocking, responsive design, and layout stability techniques. LCP and TTFB pass rates (74\u0026ndash;84%) were respectable but revealed room for improvement, with only 40% of homepages achieving good scores across all four metrics on both devices.\u003c/p\u003e \u003cp\u003eThe key novel observation is the very weak descriptive association between these dimensions (e.g., R\u0026sup2; \u0026asymp;0.045 for WAVE AIM score vs. mobile LCP; similar low correlations for errors/contrast vs. performance metrics). Homepages with superior loading speed, interactivity, and stability were not systematically associated with fewer detectable accessibility issues, and vice versa. This lack of overlap suggests that accessibility and performance optimizations operate as largely separate endeavors in state government web development. Performance gains often stem from infrastructure-focused efforts (e.g., caching, code minification, content delivery networks) aligned with general user-experience and SEO goals. Accessibility, however, requires distinct attention to semantic HTML, ARIA practices, and perceptual standards that may fall to different teams or processes. Modern development frameworks enhancing performance do not inherently enforce WCAG-compliant markup [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. This empirical finding extends limited prior work on their intersection, illustrating a practical decoupling in public-sector contexts.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003e7.2 Understanding the Independence of Accessibility and Performance Optimization\u003c/h2\u003e \u003cp\u003eThe very weak descriptive association between accessibility metrics and performance indicators (R\u0026sup2;\u0026asymp;0.045) represents a key empirical finding requiring theoretical interpretation. Why would states that have invested substantially in fast, responsive, stable websites\u0026mdash;demonstrating technical sophistication and user experience prioritization\u0026mdash;not equivalently invest in accessible markup and WCAG compliance?\u003c/p\u003e \u003cp\u003eOrganizational Structure and Governance. Web performance and accessibility optimizations likely fall to different organizational units with distinct mandates, expertise, and success metrics. Performance optimization typically resides with IT infrastructure teams, DevOps engineers, or technical architects focused on server configuration, caching strategies, content delivery networks, database optimization, and code efficiency. These efforts align with goals affecting all users: reducing page load times improves user satisfaction, reduces bounce rates, enhances search engine rankings, and decreases server costs. Performance metrics like Core Web Vitals receive organizational attention because Google incorporates them into search rankings, creating external incentives.\u003c/p\u003e \u003cp\u003eAccessibility, by contrast, often becomes the domain of legal/compliance offices, accessibility coordinators (where they exist), or content management teams responding to regulatory requirements or complaints. These units focus on WCAG conformance as a civil rights obligation rather than a general user experience enhancement. Without centralized digital governance integrating both priorities under unified leadership\u0026mdash;such as a Chief Digital Officer or integrated Digital Services team with authority over both performance and accessibility\u0026mdash;states may pursue these as parallel but uncoordinated workstreams. Different teams, different budgets, different timelines, and different success definitions create organizational siloing where one dimension can advance while the other stagnates.\u003c/p\u003e \u003cp\u003eTechnical Expertise and Professional Specialization. The skills required for performance optimization and accessibility remediation are substantially different, with limited overlap in professional training and expertise. Performance engineers and site reliability specialists develop deep knowledge of caching algorithms, load balancing, database query optimization, asset compression, lazy loading, code splitting, and browser rendering pipelines. Their work requires understanding of computer networks, server architecture, and computational efficiency\u0026mdash;skills typically acquired through computer science or systems engineering education.\u003c/p\u003e \u003cp\u003eAccessibility specialists, by contrast, require knowledge of assistive technologies (screen readers, magnification software, alternative input devices), disability studies, WCAG success criteria interpretation, semantic HTML, ARIA specification, keyboard interaction patterns, and cognitive load principles. Many accessibility professionals come from backgrounds in human-computer interaction, user experience design, occupational therapy, or disability advocacy rather than systems engineering. Few professionals develop deep expertise in both domains, and university curricula rarely integrate accessibility thoroughly into computer science or web development programs [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThis specialization means that even technically sophisticated states with strong performance teams may lack internal accessibility expertise. They may not recognize that their performant single-page applications create screen reader barriers, that their optimized image delivery omits meaningful alternative text, or that their efficient JavaScript frameworks introduce keyboard traps. Conversely, organizations with accessibility coordinators may lack the technical depth to evaluate whether vendor-provided \"accessible\" components actually meet WCAG criteria or to audit complex modern frameworks for conformance.\u003c/p\u003e \u003cp\u003eFramework and Tooling Limitations. Modern web development frameworks and performance optimization tools do not inherently promote or enforce accessibility. Popular frameworks like React, Vue, Angular, and Next.js enable exceptional performance through virtual DOM rendering, efficient state management, code splitting, and server-side rendering\u0026mdash;architectural patterns that minimize computational overhead and accelerate page delivery. However, these same frameworks permit (and sometimes encourage) implementation patterns that create accessibility barriers.\u003c/p\u003e \u003cp\u003eSingle-page applications, for instance, achieve performance gains by loading content dynamically without full page refreshes, but often fail to announce content changes to screen readers, manage focus appropriately during navigation, or update browser history for accessible back-button functionality. Developers can build applications with sub-second load times and perfect Lighthouse performance scores while completely breaking keyboard navigation, omitting semantic landmarks, or using generic ARIA labels that provide no meaningful information to assistive technology users.\u003c/p\u003e \u003cp\u003eSimilarly, performance optimization techniques like lazy loading images can inadvertently break accessibility if images lack alt text or if the lazy loading implementation doesn't properly handle focus management. Infinite scroll patterns optimize perceived performance but create keyboard navigation nightmares. CSS-based image replacement techniques reduce file size but hide content from screen readers if implemented incorrectly.\u003c/p\u003e \u003cp\u003eThe disconnect persists because performance and accessibility operate at different abstraction levels: performance concerns computational efficiency (how fast?), while accessibility concerns semantic meaning and interaction patterns (what does it mean? how do users interact?). A page can be computationally efficient but semantically meaningless; conversely, well-structured semantic HTML can perform poorly due to server or network issues. Tools like Lighthouse measure both, but developers often focus on performance scores (which affect search rankings and are easily quantified) while treating accessibility as a compliance checkbox.\u003c/p\u003e \u003cp\u003eResource Competition and Prioritization. States face competing priorities with limited budgets, staff capacity, and political attention. Performance optimization often receives priority because it benefits all users visibly and immediately: everyone notices when a page loads slowly. Performance problems generate user complaints, negative media coverage, and measurable impacts on service completion rates. Search engine algorithms penalize slow sites, creating external pressure for improvement.\u003c/p\u003e \u003cp\u003eAccessibility issues, by contrast, affect a subset of users (those with disabilities) who may not report problems\u0026mdash;either because they lack accessible channels for feedback, have learned to work around barriers, or simply abandon inaccessible sites without complaint. Accessibility failures are invisible to decision-makers who don't use assistive technologies. Without advocacy pressure, litigation, or federal enforcement, accessibility can remain a low priority despite legal requirements.\u003c/p\u003e \u003cp\u003eThis creates a problematic dynamic where states invest in performance because the benefits are universal and the incentives are strong, while treating accessibility as a specialized concern requiring separate attention only when legally mandated. The approaching 2026\u0026ndash;2027 ADA Title II deadlines may shift this calculus by creating enforcement risk, but the empirical data from December 2025 suggest that many states have not yet prioritized accessibility equivalently to performance despite the impending compliance obligations.\u003c/p\u003e \u003cp\u003eImplications for Digital Governance. The empirical disconnect between accessibility and performance outcomes suggests that states cannot assume general technical investments or \"digital modernization\" initiatives will automatically improve accessibility. Upgrading to modern frameworks, adopting cloud infrastructure, or implementing performance monitoring does not inherently address WCAG conformance unless accessibility is explicitly integrated into governance, procurement, development workflows, and quality assurance processes.\u003c/p\u003e \u003cp\u003eTo achieve comprehensive digital quality, states need governance structures that treat accessibility and performance as complementary dimensions of a unified digital services strategy. This might involve: establishing Chief Digital Officer positions with authority over both domains; creating cross-functional teams combining infrastructure engineers, front-end developers, and accessibility specialists; implementing quality gates requiring both performance benchmarks and accessibility audits before launch; adopting accessible design systems as organizational standards; training all web professionals in both domains; and structuring procurement to require vendor compliance across both dimensions. The weak empirical relationship suggests most states have not achieved this integration, instead managing accessibility and performance as separate technical concerns.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003e7.3 What the Results Do Not Tell Us\u003c/h2\u003e \u003cp\u003eInterpretation must account for methodological boundaries. Automated tools capture only 13\u0026ndash;30% of WCAG 2.1 Level AA success criteria, focusing on programmatic patterns while overlooking context-dependent issues like the relevance of alternative text, logical content sequence, or effective keyboard/focus management [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. The exclusive focus on homepages\u0026mdash;high-visibility entry points\u0026mdash;precludes generalization to deeper service pages, subdomains, PDFs, or mobile apps, where prior research suggests barriers are often more pronounced [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. The data reflect a single temporal snapshot (December 2025), subject to ongoing site updates.\u003c/p\u003e \u003cp\u003eConsequently, these results offer no definitive evidence on full WCAG conformance, real-world usability for disabled users, or preparedness for ADA Title II requirements. Favorable automated ratings indicate relative strengths in detectable areas but may mask undiscovered problems; conversely, lower ratings do not preclude positive attributes missed by the tools.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec33\" class=\"Section2\"\u003e \u003ch2\u003e7.4 Learning from Higher-Performing States\u003c/h2\u003e \u003cp\u003eWhile all 50 homepages exhibited detectable issues, the five rated \"Accessible\" by AccessScan\u0026mdash;Arkansas, Indiana, Mississippi, North Carolina, and Pennsylvania\u0026mdash;provide valuable insights into achievable standards and organizational practices that may enable relatively stronger automated outcomes.\u003c/p\u003e \u003cp\u003eWhat They Achieved. All five achieved perfect scores (100) in AccessScan's Titles category, indicating consistent, properly structured heading hierarchies with no skipped levels, appropriate nesting (H1\u0026rarr;H2\u0026rarr;H3 without gaps), and semantic use of headings for document structure rather than purely visual styling. This accomplishment, while technically straightforward (it requires systematic application of heading elements in logical order), proved elusive for 90% of states. The consistency across all five suggests not isolated successes but systematic organizational practices\u0026mdash;perhaps style guides mandating heading structures, content management system configurations enforcing hierarchy, accessible component libraries with proper heading patterns, or quality assurance processes that catch heading errors before publication.\u003c/p\u003e \u003cp\u003eSimilarly, four of five states (all except Arkansas at 75) scored\u0026thinsp;\u0026ge;\u0026thinsp;80 in Forms, with four achieving perfect 100 scores. This indicates accessible form implementations: proper label/input associations via for/id attributes, fieldset/legend groupings for related inputs, clear error messaging, and appropriate input types. Again, the consistency suggests organizational capacity rather than individual developer heroics\u0026mdash;possibly accessible form templates, validation requirements, or systematic testing procedures.\u003c/p\u003e \u003cp\u003eThe five also averaged fewer WAVE errors (2.4 vs. population mean 3.9), demonstrating that AccessScan's categorical ratings partially converge with WAVE's violation counting, though neither tool represents comprehensive conformance. Their stronger performance across multiple tool dimensions suggests genuine relative advantages in programmatically detectable accessibility characteristics.\u003c/p\u003e \u003cp\u003eWhat They Did Not Achieve. Despite \"Accessible\" ratings, all five showed residual WAVE errors (range 1\u0026ndash;5), demonstrating that no automated tool captures all issues and that even relatively stronger sites have remaining barriers requiring attention. Specific issues included missing alternative text (Arkansas, Mississippi), form labeling gaps (Indiana), contrast violations (Arkansas, Mississippi with 30 errors), and various ARIA or structural issues across all five.\u003c/p\u003e \u003cp\u003eThe Mississippi anomaly particularly illustrates automated evaluation's limitations and the importance of multi-tool triangulation. AccessScan rated Mississippi \"Accessible\" with perfect sub-scores (100/100/100/100 across Clickables, Titles, Graphics, Forms), yet WAVE detected 30 contrast errors\u0026mdash;a substantial number indicating pervasive color contrast problems\u0026mdash;and assigned a relatively low AIM score of 5.2 (vs. population mean 7.7). This dramatic tool divergence reveals either: (1) AccessScan and WAVE use different contrast calculation methods or thresholds; (2) they evaluate different page elements (e.g., AccessScan may skip certain dynamic content that WAVE captures); (3) one tool has false positives or false negatives; or (4) temporal variation if the state updated its homepage between tool runs (though data collection occurred within one week, making this unlikely).\u003c/p\u003e \u003cp\u003eThis divergence underscores that \"Accessible\" ratings reflect tool-specific heuristics and proprietary algorithms rather than verified WCAG conformance. Users should interpret such ratings as indicating relative performance on specific automated checks, not comprehensive accessibility quality. The Mississippi case particularly demonstrates the value of multi-tool approaches: relying solely on AccessScan would suggest strong accessibility, while WAVE reveals significant contrast issues requiring remediation.\u003c/p\u003e \u003cp\u003eGeographic and Demographic Diversity. The five \"Accessible\"-rated states span diverse contexts: three Southern states (Arkansas, Mississippi, North Carolina), one Midwestern (Indiana), and one Mid-Atlantic (Pennsylvania); populations ranging from small (Arkansas\u0026thinsp;~\u0026thinsp;3M, Mississippi\u0026thinsp;~\u0026thinsp;3M) to mid-size (Indiana\u0026thinsp;~\u0026thinsp;7M, North Carolina\u0026thinsp;~\u0026thinsp;11M) to large (Pennsylvania\u0026thinsp;~\u0026thinsp;13M); different political contexts, governance structures, and economic profiles. This diversity suggests that relatively stronger automated outcomes are not limited to large, wealthy, technologically advanced, or politically progressive states, but are achievable across different state characteristics.\u003c/p\u003e \u003cp\u003eThe absence of any Western states among the five may reflect random variation in a small subset (5 of 50) rather than meaningful regional patterns, particularly given the weak regional clustering observed in the broader analysis. The inclusion of smaller states like Arkansas and Mississippi challenges assumptions that accessibility requires large budgets or extensive technical staff\u0026mdash;suggesting that organizational practices, governance structures, leadership prioritization, and systematic attention to fundamentals may matter more than scale.\u003c/p\u003e \u003cp\u003ePotential Organizational Factors. Without case studies or interviews with these states' web teams, we can only infer possible organizational factors from technical patterns and public information. Likely contributing factors might include:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eAccessible design systems or component libraries: If states adopted accessible-by-default component libraries (e.g., U.S. Web Design System, which provides accessible components) or developed internal design systems with accessibility baked in, developers would inherit proper heading structures, form patterns, and interactive elements without requiring deep accessibility expertise for every implementation.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eContent management system configurations: Modern CMS platforms (e.g., Drupal, WordPress) can be configured to enforce accessibility requirements\u0026mdash;blocking publication of content missing alt text, providing accessible WYSIWYG editors, or validating heading hierarchies before save. States that properly configured CMS accessibility features would systematically prevent common errors.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eStaff training and capacity building: States that invested in training developers, designers, and content authors in accessibility fundamentals would build organizational capacity for consistent implementation. Even basic training covering heading hierarchies, alt text principles, and form labeling can dramatically improve automated audit results.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eDedicated accessibility coordinators with authority: The ADA Title II rule requires designated accessibility coordinators; states that appointed coordinators with genuine authority, resources, and technical expertise (rather than assigning accessibility as an afterthought to an already-overloaded staff member) would be better positioned to drive systematic improvements.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eQuality assurance integration: States that integrated automated accessibility scanning into pre-publication quality assurance workflows\u0026mdash;making accessibility checks a requirement before new features launch\u0026mdash;would systematically catch issues before they reach production.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eRecent major redesigns with accessibility prioritized: States that recently completed major homepage redesigns with accessibility as an explicit design requirement would naturally show stronger results. The timing of redesigns relative to the data collection period could partially explain stronger outcomes.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eResponse to previous advocacy or enforcement: States that had faced previous accessibility complaints, settlements, or DOJ investigations may have undertaken remediation efforts that improved homepage accessibility, though deeper pages might retain legacy barriers.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eTransferable Lessons for Other States. Other states seeking to improve accessibility outcomes could potentially benefit from examining these five states' approaches more closely. Relevant knowledge-sharing mechanisms might include: state CIO consortia or associations conducting peer learning sessions; documentation and open-sourcing of accessible component libraries and CMS configurations; case study development through academic-practitioner partnerships; federal technical assistance programs highlighting successful state practices; and regional collaborative networks where neighboring states share resources and expertise.\u003c/p\u003e \u003cp\u003eHowever, the presence of residual errors on all five homepages, tool divergence (especially Mississippi), and homepage-only scope mean these states represent progress toward, not achievement of, comprehensive WCAG 2.1 Level AA conformance. They demonstrate that systematic attention to fundamentals produces measurably better automated outcomes, but they also illustrate that even relatively stronger sites require ongoing attention, manual expert review, user testing with disabled participants, and extension of accessibility practices beyond high-visibility homepages to complete digital ecosystems.\u003c/p\u003e \u003cp\u003ePerformance Variability Among the Five. Notably, the five \"Accessible\"-rated states showed varied Core Web Vitals performance, with mobile LCP ranging from excellent (Indiana 1.6s, North Carolina 1.8s) to approaching the \"needs improvement\" threshold (Arkansas 2.6s). This reinforces the broader finding that accessibility and performance optimizations operate independently: states can achieve strong automated accessibility without equivalent performance optimization (Arkansas), or combine both (Indiana, North Carolina). This variability suggests that different organizational units or contractors may handle accessibility versus performance, with no systematic coordination ensuring excellence across both dimensions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec34\" class=\"Section2\"\u003e \u003ch2\u003e7.5 Broader Context and Implications\u003c/h2\u003e \u003cp\u003eThe patterns of recurring detectable issues mirror findings from other e-government evaluations, where programmatic barriers persist despite regulatory frameworks [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. The observed performance-accessibility disconnect underscores a broader challenge: technical investments benefiting overall usability do not reliably extend to WCAG-specific needs.\u003c/p\u003e \u003cp\u003eFor users relying on assistive technologies, these detectable barriers on gateway homepages can significantly impede access. Skipped or improperly structured headings hinder screen-reader navigation and outline generation; insufficient contrast reduces readability for low-vision users; missing or generic labels confuse form interactions or link purposes when announced out of visual context. While strong performance reduces general latency and frustration\u0026mdash;potentially aiding task completion for all users\u0026mdash;it cannot remedy these structural deficiencies.\u003c/p\u003e \u003cp\u003eThe combination of persistent accessibility barriers with strong performance metrics creates a paradoxical user experience: state homepages load quickly and respond smoothly for users without disabilities, providing an efficient, modern digital experience. Yet for users relying on screen readers, keyboard navigation, or high contrast, those same performant pages present fundamental barriers that prevent effective navigation or interaction\u0026mdash;fast but unusable. This divergence highlights how \"digital modernization\" focused primarily on performance, mobile responsiveness, and visual design can inadvertently widen accessibility gaps if WCAG compliance is not equally prioritized. States may perceive themselves as digitally sophisticated because their sites perform well in visible, easily measured dimensions (speed, visual design), while remaining unaware of barriers invisible to decision-makers who don't use assistive technologies.\u003c/p\u003e \u003cp\u003eAutomated screening proves efficient for population-level identification of common issues and relative outperformers (e.g., the five higher-rated homepages as potential reference points). It provides a practical starting point for addressing programmatically fixable problems. However, it reinforces the essential role of manual expert inspection and user testing with diverse disabled participants to evaluate the full range of WCAG criteria and actual usability impacts. In the context of approaching ADA Title II deadlines, descriptive baselines like this can support initial diagnostic efforts and highlight the value of multi-method approaches for more robust assessments.\u003c/p\u003e \u003c/div\u003e"},{"header":"8. Recommendations","content":"\u003cp\u003eThe findings from this automated screening suggest targeted areas for improvement based on the detectable issues observed on state government homepages. Recommendations are derived directly from common patterns (e.g., contrast failures, heading deficiencies, residual errors even on higher-rated pages) and the observed disconnect between performance and accessibility metrics.\u003c/p\u003e \u003cdiv id=\"Sec36\" class=\"Section2\"\u003e \u003ch2\u003e8.1 Immediate Priorities: Addressing High-Impact Detectable Barriers\u003c/h2\u003e \u003cp\u003eStates can achieve measurable progress by systematically addressing issues reliably flagged by automated tools, prioritized by prevalence and user impact.\u003c/p\u003e \u003cp\u003ePriority 1: Color Contrast Remediation (affects 68% of homepages). With mean 8.4 WAVE contrast errors and some states exceeding 100 violations, contrast failures represent the most prevalent detectable barrier. Remediation steps: (1) Conduct systematic audit of all text-background color combinations in design systems and style guides using automated contrast checkers (WebAIM Contrast Checker, Colour Contrast Analyser); (2) Update CSS variables, design tokens, or style guide specifications to meet WCAG minimum ratios (4.5:1 for normal text, 3:1 for large text and UI components); (3) Pay particular attention to common problem areas identified in this study\u0026mdash;navigation elements, footers, disclaimers, and interactive components; (4) Establish design system governance preventing future contrast violations through approved color palettes. Expected effort: Low to moderate (primarily designer time for palette review and CSS updates). Expected impact: High (benefits users with low vision, color blindness, and anyone viewing content in bright ambient light or on lower-quality displays).\u003c/p\u003e \u003cp\u003ePriority 2: Heading Structure Systematization (mean AccessScan titles score 68.1, minimum 33). Heading hierarchy problems appeared across 90% of homepages. Remediation steps: (1) Review all heading usage to ensure single H1 per page (main page title), logical H2-H6 structure following content hierarchy without skipped levels; (2) Update page templates and component libraries to enforce proper heading patterns; (3) Configure CMS to validate heading structures before publication; (4) Train content authors on heading purposes (semantic structure, not visual styling) and proper usage. Expected effort: Moderate (requires template updates, CMS configuration, training). Expected impact: High (screen reader users depend on headings for navigation, document comprehension, and efficient information location).\u003c/p\u003e \u003cp\u003ePriority 3: Alternative Text and Form Labeling. Missing or inadequate alternative text and form labels were common across error reports. Remediation steps: (1) Establish clear alt text guidelines distinguishing decorative (alt=\"\"), functional (describing action), and informative (conveying information) images; (2) Implement CMS workflows requiring alt text before image upload; (3) Review all form inputs for proper label associations via for/id attributes; (4) Avoid placeholder-as-label patterns; (5) Ensure fieldset/legend groupings for related inputs. Expected effort: Moderate (workflow changes, CMS configuration, content review). Expected impact: High (essential for screen reader users to access visual information and complete forms).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec37\" class=\"Section2\"\u003e \u003ch2\u003e8.2 Systematic Integration: Beyond Quick Fixes\u003c/h2\u003e \u003cp\u003eIntegrate Automated Screening into Development Workflows. Regular use of multiple automated tools (e.g., AccessScan, WAVE, axe DevTools) should become standard practice, not one-time audits. Implement automated accessibility testing in continuous integration/continuous deployment (CI/CD) pipelines to catch regressions before production. The partial cross-tool consistency observed\u0026mdash;higher-rated AccessScan homepages averaging fewer WAVE errors\u0026mdash;demonstrates that multi-tool screening identifies relative strengths and weaknesses. However, tools' limited coverage (13\u0026ndash;30% of criteria) means they should serve as an initial quality gate, not a complete solution.\u003c/p\u003e \u003cp\u003eAdopt Accessible-by-Default Design Systems. The five \"Accessible\"-rated states' consistent heading structures and well-labeled elements suggest systematic approaches rather than ad-hoc fixes. States should adopt or develop accessible design systems (e.g., U.S. Web Design System) providing pre-tested, WCAG-compliant components. This shifts accessibility from individual developer responsibility to organizational infrastructure, reducing expertise requirements and ensuring consistency.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec38\" class=\"Section2\"\u003e \u003ch2\u003e8.3 Complementary Evaluation Methods\u003c/h2\u003e \u003cp\u003eSupplement with Manual Expert Review and User Testing. Automated tools detect only 13\u0026ndash;30% of WCAG criteria. To address the majority of success criteria requiring human judgment\u0026mdash;meaningful alternative text, logical content organization, effective error recovery, keyboard interaction quality\u0026mdash;states must incorporate manual expert review by accessibility specialists and usability testing with disabled participants using diverse assistive technologies. The presence of residual errors even on the five higher-rated homepages underscores that favorable automated ratings do not guarantee comprehensive quality or real-world usability.\u003c/p\u003e \u003cp\u003eManual review should prioritize: (1) evaluating whether alternative text is meaningful and contextually appropriate, not just present; (2) testing keyboard navigation for logical tab order, visible focus indicators, and absence of keyboard traps; (3) assessing form error messages for clarity and recovery guidance; (4) verifying dynamic content updates are announced to screen readers; (5) testing complex interactive components (modals, accordions, carousels) for full assistive technology compatibility.\u003c/p\u003e \u003cp\u003eUser testing with disabled participants provides irreplaceable insights into real-world usability that neither automated tools nor expert review can fully capture. States should recruit diverse participants\u0026mdash;screen reader users (JAWS, NVDA, VoiceOver), keyboard-only navigators, low-vision users with magnification, users with cognitive disabilities\u0026mdash;to attempt common tasks (finding contact information, locating services, downloading forms) and document barriers, task completion rates, and satisfaction.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec39\" class=\"Section2\"\u003e \u003ch2\u003e8.4 Addressing the Accessibility-Performance Disconnect\u003c/h2\u003e \u003cp\u003eTreat Accessibility and Performance as Complementary Priorities. The weak association (R\u0026sup2;\u0026asymp;0.045) between accessibility metrics and Core Web Vitals indicates these dimensions are typically pursued separately, likely by different organizational units with distinct expertise and mandates. States should integrate efforts through: (1) unified digital governance structures (Chief Digital Officer or integrated Digital Services teams with authority over both domains); (2) cross-functional teams combining infrastructure engineers, front-end developers, and accessibility specialists; (3) quality assurance processes requiring both performance benchmarks and accessibility audits before launch; (4) procurement standards mandating vendor compliance across both dimensions; (5) training programs building dual competencies in modern web development and accessibility implementation.\u003c/p\u003e \u003cp\u003ePerformance-optimized frameworks (React, Vue, Angular) do not inherently enforce accessible patterns. States using modern frameworks should ensure developers understand accessible implementation patterns: focus management in single-page applications, ARIA live regions for dynamic updates, semantic landmarks, and keyboard interaction standards. This prevents the creation of fast but unusable applications.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec40\" class=\"Section2\"\u003e \u003ch2\u003e8.5 Differentiated Strategies by Current Performance Level\u003c/h2\u003e \u003cp\u003eFor the 45 AccessScan \"Non-compliant\" States: Start with automated screening using multiple tools to identify programmatically detectable barriers. Prioritize the high-impact issues identified above (contrast, headings, alt text, form labels). Focus initial efforts on homepages and highest-traffic pages. Build internal capacity through training and accessible design system adoption. Set incremental goals: reduce detectable errors by 50% within 6 months, 80% within 12 months. Track progress through quarterly automated audits.\u003c/p\u003e \u003cp\u003eFor the 5 AccessScan \"Accessible\" States (AR, IN, MS, NC, PA): Shift focus from automated screening to manual expert review and user testing. Address residual errors identified by WAVE (range 1\u0026ndash;5). Investigate tool divergence issues (particularly Mississippi's 30 contrast errors despite perfect AccessScan scores). Expand accessibility practices beyond homepages to complete digital ecosystems. Document and share organizational practices enabling relatively stronger outcomes to assist other states.\u003c/p\u003e \u003cp\u003eFor States with Strong Performance but Weak Accessibility (or vice versa): Examine organizational structures to understand why one dimension advanced while the other lagged. Consider whether separate teams manage these concerns and whether integration mechanisms exist. Implement cross-functional governance ensuring both dimensions receive equivalent attention. Leverage existing strengths: states excelling in performance likely have strong technical capacity that can be redirected toward accessibility with appropriate training and prioritization.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec41\" class=\"Section2\"\u003e \u003ch2\u003e8.6 Governance and Capacity Building\u003c/h2\u003e \u003cp\u003eEstablish Dedicated Accessibility Roles with Authority. The ADA Title II rule requires designated accessibility coordinators. States should ensure these coordinators have: (1) genuine authority and organizational influence, not just titular responsibility; (2) adequate resources (budget, staff, tools); (3) technical expertise or access to technical expertise; (4) reporting lines to senior leadership; (5) clear mandates to review and approve digital properties before launch. Accessibility should not be an afterthought assigned to already-overloaded staff but a funded, empowered organizational function.\u003c/p\u003e \u003cp\u003eReform Procurement Practices. Many states rely on third-party vendors and contractors for website development and maintenance. Procurement contracts should: (1) explicitly require WCAG 2.1 Level AA conformance with specific deliverables and testing requirements; (2) mandate Voluntary Product Accessibility Templates (VPATs) or Accessibility Conformance Reports (ACRs) with third-party verification; (3) include accessibility evaluation in vendor selection criteria; (4) establish ongoing monitoring and remediation obligations; (5) create consequences for non-conformance. Federal procurement guidelines (Section508.gov) provide models for incorporating accessibility into solicitations and contracts.\u003c/p\u003e \u003cp\u003eInvest in Training and Professional Development. Build internal capacity through comprehensive training programs for designers (accessible design principles, color contrast, cognitive load), developers (semantic HTML, ARIA, keyboard interaction patterns, assistive technology compatibility), content authors (alt text, heading usage, plain language), and procurement officers (accessibility requirements, VPAT evaluation). Consider certification programs, accessibility champions networks, and communities of practice enabling knowledge sharing across agencies and states.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec42\" class=\"Section2\"\u003e \u003ch2\u003e8.7 Leveraging the Present Baseline\u003c/h2\u003e \u003cp\u003eThis study provides a December 2025 baseline against which future progress can be measured. States should: (1) identify their position relative to peers; (2) examine specific detectable barriers on their homepages using the same tools for internal audits; (3) learn from higher-performing states' technical patterns; (4) set measurable goals for improvement; (5) track progress through regular automated screening. Advocacy organizations and researchers can use these findings to hold states accountable, prioritize engagement, and document whether compliance rates improve as deadlines approach.\u003c/p\u003e \u003cp\u003eThese recommendations, derived directly from the empirical findings, provide actionable guidance for states at different starting points. However, addressing detectable barriers on homepages represents only the beginning of ADA Title II compliance. Comprehensive conformance requires extending these practices to complete digital ecosystems, integrating manual and user-centered evaluation methods, and treating accessibility as an ongoing organizational commitment rather than a one-time remediation project.\u003c/p\u003e \u003c/div\u003e"},{"header":"9. Conclusion","content":"\u003cp\u003eThis study provides an automated, homepage-only evaluation of detectable accessibility issues and performance metrics across all 50 U.S. state government homepages in December 2025, offering a preliminary descriptive baseline ahead of the 2026\u0026ndash;2027 ADA Title II deadlines.\u003c/p\u003e \u003cp\u003eAutomated tools revealed persistent detectable barriers on all homepages. AccessScan rated 90% \"Non-compliant\" by its proprietary algorithm, with common issues including contrast failures (mean 8.4 in WAVE), heading deficiencies (AccessScan titles mean 68.1), missing alternative text, and unlabeled elements (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Tables\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Even the five homepages rated \"Accessible\" by AccessScan showed residual WAVE errors. Performance metrics were stronger, with high pass rates for interactivity (INP 98%) and layout stability (CLS 80%), though only 40% passed all Core Web Vitals on both mobile and desktop (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe very weak descriptive association between accessibility and performance indicators (e.g., R\u0026sup2; \u0026asymp;0.045; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) suggests these dimensions are largely independent in current implementations, reflecting organizational structures where different teams manage performance optimization and accessibility compliance with limited coordination.\u003c/p\u003e \u003cp\u003eFive states (Arkansas, Indiana, Mississippi, North Carolina, Pennsylvania) demonstrated that relatively stronger automated accessibility outcomes are achievable across diverse geographic and demographic contexts through systematic attention to fundamental practices like proper heading hierarchies and well-labeled interactive elements. However, residual errors on even these higher-rated homepages, combined with tool divergence (particularly Mississippi's 30 contrast errors despite perfect AccessScan scores), underscore that automated \"Accessible\" ratings reflect tool-specific heuristics rather than comprehensive WCAG conformance. The findings suggest that common detectable barriers\u0026mdash;contrast failures, heading deficiencies, missing alternative text\u0026mdash;are addressable through systematic organizational practices including accessible design systems, CMS configurations, staff training, and integrated quality assurance, but that comprehensive ADA Title II compliance requires extending these practices beyond homepages to complete digital ecosystems and supplementing automated screening with manual expert review and user testing with disabled participants.\u003c/p\u003e \u003cp\u003eThese results highlight common programmatically detectable issues on high-visibility entry points and relative strengths in performance optimization. The findings also underscore automated tools' limitations\u0026mdash;they detect only a subset of WCAG 2.1 Level AA criteria and evaluate homepages alone. Complementary manual review and user testing are necessary for fuller assessment.\u003c/p\u003e \u003cp\u003eAs a descriptive snapshot, this baseline illustrates patterns in tool-detected issues and the empirical separation of accessibility and performance priorities. It can serve as a reference for future automated screenings and longitudinal tracking as the 2026\u0026ndash;2027 deadlines approach, while emphasizing the need for broader, multi-method evaluations to better understand usability for disabled users in state digital services.\u003c/p\u003e"},{"header":"Statements and Declarations","content":"\u003cp\u003eCompeting Interests\u003c/p\u003e\n\u003cp\u003eThe author declares that they have no known competing financial or non-financial interests that could have appeared to influence the work reported in this paper.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAccessible.org. (2025). \u003cem\u003eAccessibility Scans Reliably Flag 13% of WCAG Criteria\u003c/em\u003e. Retrieved from https://accessible.org/automated-scans-wcag/\u003c/li\u003e\n\u003cli\u003eAdepoju, S. A., Shehu, I. S., \u0026amp; Bake, P. (2016). Accessibility evaluation and performance analysis of e-government websites in Nigeria. \u003cem\u003eJournal of Advances in Information Technology, 7\u003c/em\u003e(1), 49\u0026ndash;53. https://doi.org/10.12720/jait.7.1.49-53\u003c/li\u003e\n\u003cli\u003eAl-Sakran, H. O., \u0026amp; Alsudairi, M. A. (2021). Usability and accessibility assessment of Saudi Arabia mobile e-government websites. IEEE Access, 9, 48254\u0026ndash;48275. https://doi.org/10.1109/ACCESS.2021.3068917\u003c/li\u003e\n\u003cli\u003eAljedaani, W.(2025). Evaluating the accessibility of E-government websites in the United States: empirical evidence and insights. Universal Access in the Information Society, 24, 2845\u0026ndash;2866. https://doi.org/10.1007/s10209-025-01229-z\u003c/li\u003e\n\u003cli\u003eChrome Developers. (2023). \u003cem\u003eWeb Vitals Documentation (LCP, INP, CLS, TTFB).\u003c/em\u003e\u003cbr\u003ehttps://web.dev/vitals/\u003c/li\u003e\n\u003cli\u003eCisneros, D. A., Huam\u0026aacute;n Monz\u0026oacute;n, F., \u0026amp; Paz, F. (2021). Accessibility evaluation of E-government web applications: A systematic review. In M. M. Soares, E. Rosenzweig, \u0026amp; A. Marcus (Eds.), Design, user experience, and usability: Design for diversity, well-being, and social development (pp. 210\u0026ndash;223). Springer. https://doi.org/10.1007/978-3-030-78224-5_15\u003c/li\u003e\n\u003cli\u003eDjatmiko, G. H., Sinaga, O., \u0026amp; Pawirosumarto, S. (2025). Digital transformation and social inclusion in public services: A qualitative analysis of e-government adoption for marginalized communities in sustainable governance. Sustainability, 17(7), Article 2908. https://doi.org/10.3390/su17072908\u003c/li\u003e\n\u003cli\u003eDangol, P.(2025). Website accessibility evaluation of the federal government of Nepal. Universal Access in the Information Society, 24, 3125\u0026ndash;3141. https://doi.org/10.1007/s10209-023-01076-w\u003c/li\u003e\n\u003cli\u003eDeque Systems. (2023). Automated Testing Identifies 57% of Digital Accessibility Issues. Retrieved fromhttps://www.deque.com/blog/automated-testing-study-identifies-57-percent-of-digital-accessibility-issues/\u003c/li\u003e\n\u003cli\u003eDroutsas, N., Spyridonis, F., Daylamani-Zad, D., \u0026amp; Ghinea, G. (2025). Web accessibility barriers and their cross-disability impact in eSystems: A scoping review. Computer Standards \u0026amp; Interfaces, 92, Article 103923. https://doi.org/10.1016/j.csi.2024.103923\u003c/li\u003e\n\u003cli\u003eEuropean Commission. (2022).\u003cem\u003eReport on the implementation of Directive (EU) 2016/2102 on the accessibility of the websites and mobile applications of public sector bodies.\u003c/em\u003e\u003cbr\u003ehttps://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022SC0410\u003c/li\u003e\n\u003cli\u003eEuropean Commission. (2023). \u003cem\u003eeGovernment Benchmark 2023. \u003cbr\u003e \u003c/em\u003ehttps://digital-strategy.ec.europa.eu/en/library/egovernment-benchmark-2023\u003c/li\u003e\n\u003cli\u003eGoogle. (2023). \u003cem\u003eCore Web Vitals Documentation.\u003c/em\u003e Chrome Developers / web.dev.\u003cbr\u003e Includes LCP, INP, CLS, and TTFB performance thresholds.\u003cbr\u003ehttps://web.dev/vitals/\u003c/li\u003e\n\u003cli\u003eHerres, G. (n.d.). \u003cem\u003eHow Much Can Automated Accessibility Tools Do?\u003c/em\u003e Easy A11y Guide. Retrieved from https://easya11yguide.com/tips/how-much-can-automated-accessibility-tools-do/\u003c/li\u003e\n\u003cli\u003eIntelligence Community Design System. (n.d.). \u003cem\u003eLimitations of Automated Testing\u003c/em\u003e. Retrieved from https://design.sis.gov.uk/accessibility/testing/automated-testing-limitation/\u003c/li\u003e\n\u003cli\u003eKolotouchkina, O., Barrientos-B\u0026aacute;ez, A., S\u0026aacute;daba-Chalezquer, C., \u0026amp; S\u0026aacute;nchez-Amboage, E. (2022). Smart cities and the digital divide: Myths and realities regarding the \u0026quot;smartness\u0026quot; of cities. Cities, 124, 103613. https://doi.org/10.1016/j.cities.2022.103613\u003c/li\u003e\n\u003cli\u003eLazar, J., Goldstein, D. F., \u0026amp; Taylor, A. (2015). \u003cem\u003eEnsuring Digital Accessibility Through Process and Policy.\u003c/em\u003e Morgan Kaufmann. ISBN: 978-0128006467\u003cbr\u003ePublisher page: https://www.elsevier.com/books/ensuring-digital-accessibility-through-process-and-policy/lazar/978-0-12-800646-7\u003c/li\u003e\n\u003cli\u003eDroutsas et al, A., \u0026amp; Lazar, J. (2011).Accessibility of U.S. federal government home pages: Section 508 compliance and site accessibility statements. \u003cem\u003eGovernment Information Quarterly, 28\u003c/em\u003e(3), 303\u0026ndash;309. https://doi.org/10.1016/j.giq.2011.02.002.\u003c/li\u003e\n\u003cli\u003ePaul, S. (2023). Accessibility analysis using WCAG 2.1: evidence from Indian e-government websites. Universal Access in the Information Society, 22(2), 663\u0026ndash;669. https://doi.org/10.1007/s10209-021-00861-9\u003c/li\u003e\n\u003cli\u003eSarrab, M., Al-Saleem, S., \u0026amp; Aljaser, E. (2021). Evaluating e-government websites accessibility: Challenges and recommendations. Universal Access in the Information Society, 20, 855\u0026ndash;867. https://doi.org/10.1007/s10209-020-00745-4\u003c/li\u003e\n\u003cli\u003eSection508.gov. (n.d.). \u003cem\u003eAccessibility Conformance Report (ACR)\u003c/em\u003e. U.S. General Services Administration. Retrieved from https://www.section508.gov/sell/acr/\u003c/li\u003e\n\u003cli\u003eSection508.gov. (n.d.). \u003cem\u003eBuy Accessible Products and Services\u003c/em\u003e. U.S. General Services Administration. Retrieved from https://www.section508.gov/buy/\u003c/li\u003e\n\u003cli\u003eTPGi. (2025). \u003cem\u003eVPAT 101: A Guide for Federal Contractors and Subcontractors\u003c/em\u003e. Retrieved from https://www.tpgi.com/vpat-101-a-guide-for-federal-contractors-and-subcontractors/\u003c/li\u003e\n\u003cli\u003eUnited Nations. (2021). \u003cem\u003eICT and Digital Accessibility: Disability Inclusion Practice Note.\u003c/em\u003e\u003cbr\u003eUnited Nations Sustainable Development Group. https://unsdg.un.org/sites/default/files/2021-04/ICT-Digital%20Accessbility-BOS-Disability%20Inclusion-Practice%20Note-20210303.pdf\u003c/li\u003e\n\u003cli\u003eU.S. Department of Justice. (2024). \u003cem\u003eNondiscrimination on the Basis of Disability; Accessibility of Web Information and Services of State and Local Government Entities.\u003c/em\u003e Federal Register, 89(80). https://www.federalregister.gov/documents/2024/04/24/2024-07758/nondiscrimination-on-the-basis-of-disability-accessibility-of-web-information-and-services-of-state\u003c/li\u003e\n\u003cli\u003eVassilakopoulou, P., \u0026amp; Hustad, E. (2023). Bridging digital divides: A literature review and research agenda for information systems research. \u003cem\u003eInformation Systems Frontiers\u003c/em\u003e, \u003cem\u003e25\u003c/em\u003e(3), 955\u0026ndash;969. https://doi.org/10.1007/s10796-020-10096-3\u003c/li\u003e\n\u003cli\u003eWebAIM. (2023). \u003cem\u003eThe WebAIM Million 2023: An accessibility analysis of the top 1,000,000 home pages. \u003c/em\u003ehttps://webaim.org/projects/million/2023/\u003c/li\u003e\n\u003cli\u003eWebAIM. (2024). \u003cem\u003eThe WebAIM Million 2024: An accessibility analysis of the top 1,000,000 home pages.\u003c/em\u003e https://webaim.org/projects/million/\u003c/li\u003e\n\u003cli\u003eW3C Web Accessibility Initiative. (2018). \u003cem\u003eWeb Content Accessibility Guidelines (WCAG) 2.1. \u003c/em\u003ehttps://www.w3.org/TR/WCAG21/\u003c/li\u003e\n\u003cli\u003eW3C Web Accessibility Initiative. (2024). \u003cem\u003eAccessibility Principles. \u003c/em\u003ehttps://www.w3.org/WAI/fundamentals/accessibility-principles/\u003c/li\u003e\n\u003cli\u003eW3C. (n.d.). \u003cem\u003eInvolving Users in Evaluating Web Accessibility\u003c/em\u003e. Web Accessibility Initiative. Retrieved from https://www.w3.org/WAI/test-evaluate/involving-users/\u003c/li\u003e\n\u003cli\u003eW3C. (2018).\u003cem\u003eWeb Content Accessibility Guidelines (WCAG) 2.1.\u003c/em\u003e World Wide Web Consortium Recommendation. https://www.w3.org/TR/WCAG21/\u003c/li\u003e\n\u003cli\u003eZubair, M. (2025). Evaluating the accessibility and performance of government websites in Nigeria. \u003cem\u003eUniversal Access in the Information Society\u003c/em\u003e, \u003cem\u003e24\u003c/em\u003e, 3639\u0026ndash;3648. https://doi.org/10.1007/s10209-025-01244-0\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"universal-access-in-the-information-society","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"uais","sideBox":"Learn more about [Universal Access in the Information Society](http://link.springer.com/journal/10209)","snPcode":"10209","submissionUrl":"https://submission.nature.com/new-submission/10209/3","title":"Universal Access in the Information Society","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"digital accessibility, e-government, WCAG 2.1, ADA Title II, Core Web Vitals, automated evaluation","lastPublishedDoi":"10.21203/rs.3.rs-8663556/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8663556/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003eThe U.S. Department of Justice\u0026rsquo;s 2024 ADA Title II rule adopts Web Content and Accessibility Guidelines (WCAG) 2.1 Level AA as the technical standard for state and local government websites, with enforceable deadlines in 2026\u0026ndash;2027. This study establishes a population-level baseline of detectable accessibility issues and performance characteristics on U.S. state government homepages ahead of these deadlines.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThe primary homepages of all 50 U.S. state government websites were evaluated using three automated tools: AccessScan for structural, navigational, semantic, and perceptual issues; WAVE for WCAG-aligned error detection; and Google PageSpeed Insights for Core Web Vitals (loading, interactivity, and visual stability) on mobile and desktop. Automated tools detect only a limited subset (approximately 13\u0026ndash;30%) of WCAG 2.1 Level AA success criteria; therefore, results reflect detectable issues rather than comprehensive conformance.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eAccessScan rated 45 homepages (90%) as \u0026ldquo;Non-compliant\u0026rdquo; and 5 (10%) as \u0026ldquo;Accessible\u0026rdquo; according to its proprietary heuristics; however, all five exhibited residual WAVE errors (mean 2.4 errors). Common detectable barriers included color contrast failures (mean 8.4 per page), heading structure deficiencies, missing alternative text, and incomplete ARIA labeling. No homepage was free of detectable issues. Performance was generally strong, with 74\u0026ndash;98% of homepages meeting individual Core Web Vitals \u0026ldquo;good\u0026rdquo; thresholds, though only 40% passed all metrics on both mobile and desktop. Associations between accessibility and performance metrics were weak.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eDetectable accessibility barriers remain widespread on state government homepages despite generally strong web performance, suggesting accessibility and performance optimizations are often pursued independently. These findings provide a descriptive benchmark ahead of ADA Title II enforcement deadlines and underscore the necessity of complementary manual review and user testing to assess full WCAG 2.1 Level AA conformance across public-sector digital ecosystems.\u003c/p\u003e","manuscriptTitle":"Automated Evaluation of Detectable Accessibility Issues on U.S. State Government Homepages: A Baseline Assessment Ahead of the 2026–2027 ADA Title II Deadlines","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-10 12:11:41","doi":"10.21203/rs.3.rs-8663556/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-02-26T14:44:41+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-13T08:36:34+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"71511166775036580820151885786336714296","date":"2026-02-06T08:23:49+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-05T15:29:57+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-23T14:40:28+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-23T05:38:58+00:00","index":"","fulltext":""},{"type":"submitted","content":"Universal Access in the Information Society","date":"2026-01-21T21:32:09+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"universal-access-in-the-information-society","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"uais","sideBox":"Learn more about [Universal Access in the Information Society](http://link.springer.com/journal/10209)","snPcode":"10209","submissionUrl":"https://submission.nature.com/new-submission/10209/3","title":"Universal Access in the Information Society","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"d69091d3-f000-4e14-bb74-0409e2196dda","owner":[],"postedDate":"February 10th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-04-13T16:09:15+00:00","versionOfRecord":{"articleIdentity":"rs-8663556","link":"https://doi.org/10.1007/s10209-026-01328-5","journal":{"identity":"universal-access-in-the-information-society","isVorOnly":false,"title":"Universal Access in the Information Society"},"publishedOn":"2026-04-07 15:58:06","publishedOnDateReadable":"April 7th, 2026"},"versionCreatedAt":"2026-02-10 12:11:41","video":"","vorDoi":"10.1007/s10209-026-01328-5","vorDoiUrl":"https://doi.org/10.1007/s10209-026-01328-5","workflowStages":[]},"version":"v1","identity":"rs-8663556","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8663556","identity":"rs-8663556","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}