Navigating the Future: Assessing the Impacts and Challenges of AI Integration in Higher Education Institutions in Zanzibar

Navigating the Future: Assessing the Impacts and Challenges of AI Integration in Higher Education Institutions in Zanzibar | 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 Research Article Navigating the Future: Assessing the Impacts and Challenges of AI Integration in Higher Education Institutions in Zanzibar Jecha Jecha, Hayfa Nassor This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9479399/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Generative artificial intelligence (GenAI) is entering higher education systems across Africa at a pace that outstrips institutional preparedness, yet empirical evidence from small, resource-constrained island settings remains absent from the literature. This study examined GenAI perceptions, readiness, and preparedness among higher education participants in Zanzibar, Tanzania, a small island system with no prior documentation in the technology adoption literature. A convergent mixed methods design was employed: 235 participants (192 students and 43 lecturers) from four institutions completed a structured online questionnaire, providing both Likert-scale responses analysed via partial least squares structural equation modelling (PLS-SEM) and open-ended responses analysed using reflexive thematic analysis. PLS-SEM results showed that perception strongly predicted readiness (β = 0.849, R² = 0.722) and that readiness was the primary mechanism linking perception to preparedness (indirect β = 0.822; total effect β = 0.355, p < .001). Thematic analysis of 235 open-ended response sets yielded ten themes spanning infrastructure deficits, skills and digital literacy gaps, absent institutional policy, financial barriers, cognitive risks, and broadly positive orientations toward AI’s educational potential. Integration of the two strands showed that structural factors–unreliable internet, insufficient ICT resources, and the absence of institutional governance frameworks–explain why positive perceptions and readiness do not automatically translate into preparedness. These findings challenge simplified adoption models that treat psychological variables as sufficient and demonstrate that in low-resource island contexts, awareness and motivation require concurrent investment in infrastructure, training, and institutional policy to produce genuine implementation capacity. The practical implications for Zanzibar higher education sector and comparable systems across the Global South are discussed. generative AI higher education PLS-SEM technology adoption readiness preparedness mixed methods Zanzibar Africa Figures Figure 1 Figure 2 1. INTRODUCTION Artificial intelligence is reshaping higher education in ways that are both exciting and uneven. Large language models and generative AI (GenAI) tools have entered classrooms, research workflows, and administrative systems at a pace that most institutions were not designed to absorb. The promise is genuine: more personalized learning, expanded access to information, and time freed from routine tasks for teaching and scholarship. However, realizing that promise requires infrastructure, institutional capacity, and policy frameworks that are far from uniformly available. While universities in North America and Western Europe debate the governance of GenAI use, many African institutions face a more immediate question: how to engage with transformative technologies when foundational infrastructure, trained personnel, and institutional governance are still developing. That gap is not simply a matter of being behind on the adoption curve; it reflects structural differences in resources, connectivity, and context that determine what adoption actually means and what it requires (Maluleke, 2025 ). The urgency of this question is not hypothetical. The African Union’s Agenda 2063 explicitly positions AI as critical infrastructure for economic development and educational equity, and digital transformation initiatives across East Africa have placed technology integration at the center of national education strategies. Universities on the continent are beginning to incorporate GenAI into teaching, learning, and research. However, empirical evidence on how this adoption unfolds in practice, what faculty and students think, how ready they feel, and what actually prevents or enables integration remains thin, particularly in smaller and island-based higher education systems. Most documented studies on AI adoption in African higher education have concentrated on anglophone mainland countries such as Nigeria, Ghana, and South Africa, and many focus on student populations rather than educators (Maluleke, 2025 ). Zanzibar, a small island higher education system within Tanzania, does not appear in this literature. This matters because island contexts carry distinct infrastructure vulnerabilities, unreliable internet, power instability, smaller talent pools, and limited institutional budgets that merit investigation in their own right, not as a footnote to mainland findings. A second gap concerns how the adoption process itself is understood. Established technology adoption frameworks, such as the technology acceptance model (Davis, 1989 ), theory of planned behavior (Ajzen, 1991 ), and unified theory of acceptance and use of technology (UTAUT; Venkatesh et al., 2003 ) have been productively applied to AI adoption in various African settings, often revealing that teacher attitudes and intentions are broadly positive, even when actual adoption lags (Patterson et al., 2025 ). These frameworks, however, primarily capture psychological variables, such as perceived usefulness and ease of use, as well as social influence. The pathway from a favorable attitude to genuine preparedness to teach with technology involves more than perception. It involves skill development, institutional support, reliable infrastructure, and policy clarity. Few studies examine this full pathway from perception through readiness to operational preparedness, and even fewer combine quantitative testing of psychological relationships with qualitative investigation of the structural and institutional conditions that enable or obstruct implementation. Consequently, we may know that teachers hold positive views of GenAI without understanding why they feel unprepared to deploy it in teaching. A third gap concerns generative AI specifically. Research on AI in education predates the current GenAI moment by many years; however, the concerns raised by large language model hallucinations and misinformation risks, academic integrity, and questions about cognitive offloading are substantively different from those raised by earlier AI technologies and are only beginning to attract systematic empirical investigation. Teacher educators in Ghana, for instance, have been found to express both enthusiasm and anxiety about GenAI, with anxiety often centering on implications for critical thinking and equity (Akanzire et al., 2025 ). Nevertheless, the factors that shape these orientations in smaller, lower-resource contexts and how they translate into preparedness for classroom implementation remain largely undocumented. Zanzibar provides an instructive case for this investigation. Its higher education system comprises a small number of institutions serving a growing student population in an environment where digital transformation is underway but constrained. Internet connectivity is unreliable, ICT resources are limited, and institutional capacity for managing rapid technological change is still developing. These are not peripheral challenges; they are central to explaining why positive attitudes toward AI do not automatically produce preparedness to integrate it into academic and professional practice (Muringa, 2025 ). By investigating Zanzibar, this study generates empirical evidence from a context entirely absent from the existing literature, while also offering findings of broader relevance to small, resource-constrained higher education systems across the Global South. These findings have direct implications for policy and practice. If preparedness depends not only on attitudes and intentions but also on infrastructure, training, institutional policy, and financial resources, then strategies for promoting GenAI adoption must address structural conditions alongside individual psychology. This study documents these conditions and identifies where institutional and governmental investments are most urgently needed. 2. LITERATURE REVIEW 2.1 Generative AI Adoption in Higher Education Generative AI has entered higher education institutions globally at a considerable speed, prompting substantial research attention. UTAUT has emerged as a dominant theoretical lens for understanding technology adoption in educational settings. Developed by integrating eight predecessor acceptance models, UTAUT proposes that performance expectancy, effort expectancy, social influence, and facilitating conditions predict both behavioural intention and actual technology use (Venkatesh et al., 2003 ). This framework has since been extended and applied extensively in AI adoption research. Al-Emran et al. ( 2024 ) applied UTAUT to examine university teachers’ views on generative AI tools for student assessment across the Middle East, drawing on 358 faculty members from multiple countries. Their findings showed that performance expectancy, effort expectancy, social influence, and hedonic motivation each shaped educators’ behavioural intentions and actual use patterns. Institutional policies for GenAI integration were a significant driver, pointing to the importance of organisational structures alongside individual psychology. Trust has attracted growing attention as a separate dimension of GenAI adoption. Schreurs et al. ( 2025 ), in a comparative study of 823 higher education participants, including students, teachers, and researchers, found that trust in tools such as ChatGPT, Microsoft Copilot, and Google Gemini was experience-driven: frequency of use, duration, and self-assessed proficiency predicted trust, while demographic variables showed minimal influence. Trust, in turn, strongly predicted behavioral intention to adopt. These findings suggest that exposure and positive experience generate adoption momentum, a dynamic with particular relevance in contexts where limited infrastructure constrains initial exposure. Thaldar et al. ( 2025 ) documented the institutional governance side of this challenge in African higher education, using the University of KwaZulu-Natal as a case study. They described it as one of the first African institutions to develop comprehensive academic guidelines for GenAI use and observed that institutional capacity building and formal policy frameworks remain nascent across most African universities; only South Africa, Rwanda, and Nigeria have begun to align institutional policies with national AI strategies. Research on pedagogical dimensions reveals concerns specific to GenAI’s transformative character. Akanzire et al. ( 2025 ), examining teacher educator perspectives in Ghana, found that while educators expressed enthusiasm about GenAI’s educational potential, significant anxieties surfaced around critical thinking, equity of access, and appropriate assessment practices. This pattern echoes broader evidence that positive attitudes toward technology do not automatically translate into readiness to integrate it into teaching. Modiba et al. ( 2025 ), in a systematic literature review of GenAI in African higher education, identified dual opportunities–improved student writing, research productivity, and greater academic autonomy–alongside substantial barriers: inadequate policy regulation, technical and structural challenges related to connectivity and device access, and knowledge deficits among educators and students. 2.2 Infrastructure Barriers, Teacher Readiness, and the Adoption Gap A persistent pattern in the adoption literature is the gap between positive attitudes and actual preparedness. Mwapwele et al. ( 2019 ), applying the technology readiness index to teacher ICT adoption in South African rural schools, found that most surveyed teachers expressed optimism about ICT for teaching and learning; however, significant financial, technical, and digital skills deficits persisted. This finding maps enthusiasm without capacity directly onto contemporary GenAI research. The UTAUT theoretical architecture helps explain this disconnect. While performance expectancy and effort expectancy predict behavioural intention, facilitating conditions predict actual usage behaviour. A systematic review by Feng et al. ( 2025 ), covering 39 studies from 2015 to 2024, confirmed this pattern: facilitating conditions infrastructure, technical support, organisational policy, and institutional resources emerged as critical enablers of implementation success; however, they are routinely under-addressed in practice. In developing countries, Tasnim et al. ( 2025 ) drew on 321 Bangladeshi university students across 22 institutions to show that facilitating conditions, not performance expectancy alone, predicted e-learning continuance. Infrastructure reliability, content accessibility, network stability, technical training, and stakeholder collaboration were pivotal. The conceptual distinction between readiness and preparedness is important. Guo et al. (2024) validated a teacher acceptance of artificial intelligence (TAAI) instrument that captures five dimensions: perceived usefulness, perceived ease of use, behavioural intention, self-efficacy, and anxiety. This instrument measures psychological readiness. However, preparedness requires additional elements that acceptance models do not capture. Macwan et al. (2025), reviewing the literature on teacher readiness for digital technology, identified six clusters: cognitive readiness (perception, self-efficacy, prior experience), pedagogical readiness (integrating content with technology), affective readiness (anxiety, emotional attitudes), institutional readiness (infrastructure, policy support, governance), technological readiness (tool-specific competence), and sociocultural factors. Only the first three are primarily psychological, whereas the latter three are organisational and contextual. Therefore, preparedness encompasses psychological acceptance as well as institutional, infrastructural, and pedagogical enabling conditions. African higher-education contexts present particular structural complexities. Aluko and Mkhize ( 2025 ) examined AI integration in South African universities through the lens of curriculum transformation and found that historically White universities had made greater progress owing to superior funding and international partnerships, while historically Black universities faced systemic barriers, including limited resources, inadequate infrastructure, and constrained technical capacity. This inequality stems not from differential attitudes but from structural resource disparities, a finding with clear parallels in lower income contexts across the continent. This pattern also appears in the broader EdTech adoption literature: teachers in resource-constrained settings tend to express positive attitudes toward technology; however, the translation of those attitudes into classroom practice is obstructed by poor infrastructure, insufficient training, and the absence of supportive national policies. 2.3 Strengths and Gaps in Existing Research Research on technology adoption in higher education demonstrates methodological maturity in several respects. UTAUT provides a comprehensive, empirically validated framework integrating psychological and organisational variables; its predictive validity across cultural contexts and technology types has been confirmed in multiple meta-analyses and systematic reviews (Feng et al., 2025 ; Macwan et al., 2025). Recent studies have shown growing sophistication in measuring both perceptions and enabling conditions, moving beyond the oversimplification of earlier attitudinal research. The validation of measurement instruments (Guo et al. 2024), such as the TAAI scale and Viberg ’s(2020) digital preparedness instrument, provides reliable tools for quantitative work across contexts. African-centred research is also expanding, with context-sensitive studies from South Africa (Thaldar et al., 2025 ; Mwapwele et al., 2019 ; Aluko & Mkhize, 2025 ), Ghana (Akanzire et al., 2025 ), Nigeria (Macwan et al., 2025), Zambia (Mtonga & Mbewe, 2024 ), and Bangladesh (Tasnim et al., 2025 ) providing alternatives to the uncritical transfer of Western models. Several critical gaps remain. Geographic coverage is uneven: substantial research exists for North American and Western European contexts, African higher education is underrepresented, and within Africa, the literature concentrates on resource-rich institutions in South Africa, Nigeria, and Ghana. Small island higher education systems have not been studied. Zanzibar, in particular, has no documented empirical research on technology adoption in its universities, despite facing infrastructure challenges, geographic isolation, unreliable internet, and constrained budgets, which differ substantially from mainland African contexts. Research on GenAI, particularly, is nascent. Most African studies on GenAI adoption take the form of theoretical analyses, policy case studies, or systematic reviews rather than empirical measurements of teacher preparedness in particular institutional settings (Thaldar et al., 2025 ; Modiba et al., 2025 ; Akanzire et al., 2025 ). Quantitative studies testing the full pathway from perceptions through readiness to measurable preparedness are rare. The conceptualisation of preparedness itself lacks consensus: few studies operationalise it as a multidimensional construct integrating psychological, pedagogical, technical, and institutional dimensions, and the conceptual separation between readiness (motivational and psychological orientation) and preparedness (demonstrated capacity to implement) has not been systematically examined. Most studies also use purely quantitative or purely qualitative designs, leaving the interplay between individual psychological factors and structural enablers incompletely understood. The role of policy and governance, distinct from attitudinal variables, has received relatively little attention. 2.4 Research Gap and Research Questions The literature reveals a gap at the intersection of geography, methodology, and construct specification: no empirical research documents teacher perceptions, readiness, and preparedness for GenAI integration in Zanzibar ’s higher education system or in comparable small island settings elsewhere in Africa. This study directly addresses this gap through a convergent mixed methods investigation. Four research questions were used to organise the enquiry. What are higher education participants’ perceptions of the benefits, risks, and relevance of generative AI to higher education in Zanzibar? To what extent are higher education participants psychologically ready and willing to adopt generative AI in their academic and professional work, and what factors predict their readiness? How prepared are the participants and their institutions to effectively integrate generative AI, and what are the primary barriers to that preparedness? How do institutional factors (infrastructure, policy, support, and training), pedagogical factors (content knowledge and teaching and learning approach), and individual factors (confidence, skills, and prior experience) interact to enable or obstruct the translation of readiness into preparedness? Answering these questions generates evidence that directly informs policy and practice in Zanzibar and contributes context-sensitive data on technology adoption in resource-constrained, island-based settings, which are currently lacking in the literature. 3. METHOD 3.1 Research Design This study used a convergent mixed methods design to examine higher education participants’ perceptions of GenAI, their readiness to adopt GenAI, and their preparedness to integrate it into academic and professional practice in Zanzibar higher education institutions. In a convergent design, quantitative and qualitative data are collected concurrently, analysed separately using appropriate methods, and integrated during interpretation to develop a more complete account of the phenomenon (Creswell & Plano Clark, 2018 ). Both strands were given equal priority. The quantitative strand estimated the relationships among perceptions, readiness, and preparedness, whereas the qualitative strand illuminated the contextual barriers and facilitators shaping those relationships. Integration focused on identifying the convergence, divergence, and complementarity between the strands. Three considerations justified this design. The research questions required both quantitative hypothesis testing and qualitative contextualisation of institutional conditions. GenAI adoption in Zanzibar is an emerging area with little prior empirical work, making it important to capture both the breadth available through survey data and the depth available through open-ended accounts. Integrating both strands was expected to yield stronger, more contextually grounded evidence than either strand alone. 3.2 Quantitative Component 3.2.1 Design and Analytical Approach The quantitative strand employed a cross-sectional survey design. Partial least squares structural equation modelling (PLS-SEM) was chosen as the primary analytical method because it is well suited to prediction-oriented research, models with mediation, and moderate sample sizes. PLS-SEM maximises the explained variance in key constructs rather than optimising global model fit, making it appropriate for exploratory and theory-building work, both of which describe the present study’s orientation toward GenAI adoption in Zanzibar higher education system (Hair et al., 2019 ). 3.2.2 Context and Sampling Data were collected from four universities in Zanzibar, Tanzania, between October and December 2024. Convenience sampling was used to recruit lecturers and students at these institutions. Therefore, the sample represents the broader higher education community, including those who teach with, learn with, or are likely to encounter GenAI tools in academic settings. This approach introduces potential selection bias and limits generalisability; however, it was appropriate given the exploratory orientation of the study, the absence of a comprehensive sampling frame across Zanzibar higher education institutions, the practical constraints of field data collection, and PLS-SEM tolerance for non-probability samples, where the emphasis is on prediction and explanation rather than population-level inference. To partially mitigate sampling bias, data were collected from four distinct institutions. The sample characteristics are reported in detail below, and the limitations on generalisation are explicitly acknowledged. 3.2.3 Sample Size The sample size adequacy for PLS-SEM was evaluated using the inverse square root method (Kock & Hadaya, 2018 ). For a model with three constructs and a maximum of two arrows entering any single construct, this method suggested a minimum of approximately 160 participants to detect medium effect sizes (f² ≥ 0.15) at α = 0.05 and power = 0.80. The final sample of 235 participants exceeded the threshold. 3.3 Measurement Instrument 3.3.1 Survey Development A structured online questionnaire was developed using Google Form. The instrument comprised three sections corresponding to the three latent constructs, each measured using multi-item Likert scales. Perception of GenAI (five items) captured attitudes, beliefs, and evaluations of GenAI in educational settings, rated on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). Readiness for GenAI (five items) assessed self-reported confidence and willingness to adopt GenAI tools using the same response scale as above. Preparedness for GenAI integration (four items) measured the perceived competence and capability to integrate GenAI into academic, teaching, and professional practice. All items were adapted from established scales in the technology acceptance and educational technology literature, with minor modifications to reflect the GenAI domain and the Zanzibar institutional context. 3.3.2 Measurement Model Specification All three constructs were specified as reflective, latent variables. In reflective models, observed indicators are treated as manifestations of an underlying construct; changes in the latent construct are expected to be reflected across all indicators, and indicators are expected to be highly intercorrelated. The removal of any single item did not alter the construct’s conceptual domain. This specification aligns with the conceptualisation of perception, readiness, and preparedness as underlying psychological states, expressed through survey responses. 3.4 Data Collection Procedure The survey link was distributed via institutional email lists and internal communication channels at the four participating universities, reaching both lecturers and registered students. Participation was voluntary; informed consent was obtained electronically at the start of the survey. No personally identifiable information was collected. This study was conducted in accordance with the ethical principles of the Declaration of XX University. All participants provided informed electronic consent prior to participation. The survey remained open for three months and yielded 235 complete responses. Google Forms built-in validation features minimised data entry errors. Cases with more than 20% missing data on construct items were excluded, consistent with standard PLS-SEM practice; no cases were excluded on this basis, as missing data were minimal across the dataset (< 2%). 3.5 Quantitative Data Analysis 3.5.1 Software and Analytical Steps Quantitative data were analysed using SmartPLS 4.0 (Ringle et al., 2024 ). Following standard practice, the analysis proceeded in two stages: assessment of the measurement model (reliability and validity of the latent constructs) and evaluation of the structural model (relationships among the constructs). 3.5.2 PLS-SEM Algorithm and Bootstrapping The PLS algorithm used a path weighting scheme with a maximum of 300 iterations and a stop criterion of 10⁻⁷; all results were standardised. All constructs were estimated using Mode A, consistent with their reflective specifications. The statistical significance of the path coefficients, indirect effects, and total effects was assessed through nonparametric bootstrapping with 10,000 resamples, producing standard errors, t-values, p-values, and 95% bias-corrected confidence intervals. 3.5.3 Measurement Model Evaluation Internal consistency reliability was assessed using Cronbach’s alpha (α) and composite reliability (ρc and ρa), with values ≥ 0.70 indicating acceptable reliability. Convergent validity was evaluated using the average variance extracted (AVE; ≥ 0.50). Indicator reliability was examined through outer loadings, with values ≥ 0.70 as the primary criterion. Hair et al. ( 2019 ) note that indicators with loadings between 0.40 and 0.70 may be retained if their removal would not increase the AVE or composite reliability above the threshold. Discriminant validity was assessed using the heterotrait-monotrait ratio of correlations (HTMT), with 0.85 as the conservative threshold, and 0.90 as the more liberal criterion. To guard against multicollinearity in the measurement model, variance inflation factors (VIFs) were examined for all outer loadings, with VIF < 5.0 indicating acceptable levels (Hair et al., 2019 ). The inner model VIF values were similarly examined for the structural paths. 3.5.4 Structural Model Evaluation and Mediation The structural model was evaluated using the following criteria: coefficients of determination (R²; 0.25, 0.50, and 0.75 interpreted as weak, moderate, and substantial), effect sizes (f²; 0.02, 0.15, and 0.35 as small, medium, and large), and path coefficients (β), with significance assessed via bootstrapped t-values (|t| ≥ 1.96 at α = 0.05, two-tailed). Model fit was examined using the Standardised Root Mean Square Residual (SRMR), with 0.10 as an acceptable threshold for PLS-SEM in exploratory research (Henseler et al., 2015 ). Mediation was examined by decomposing the effect of perception on preparedness into direct, indirect, and total components through readiness. The significance of the indirect and total effects was tested using bootstrapped confidence intervals. The patterns of direct and indirect effects were used to classify the mediation type following Zhao et al. ( 2010 ). 3.6 Qualitative Component 3.6.1 Data Source Qualitative data were collected through open-ended questions in the same Google Forms questionnaire. After completing the Likert-scale items, respondents were asked to provide open-ended responses to three questions, which were set as required fields in Google Forms to ensure complete data collection. All 235 respondents provided written responses, yielding robust data for thematic analysis. This embedded design ensured a direct link between the quantitative and qualitative data from the same participant group. 3.6.2 Qualitative Data Analysis Open-ended responses were analysed using reflexive thematic analysis, following Braun and Clarke’s ( 2019 ) six-phase approach. All responses were read repeatedly to achieve familiarity and identify the initial patterns. Meaningful units of text were then coded inductively, generating codes from the data rather than imposing them a priori. Related codes were grouped into candidate themes that described the key aspects of GenAI integration. Candidate themes were reviewed against the full dataset for internal coherence and distinctiveness of the themes. Themes were then defined and refined, clarifying the scope, core meaning, and boundaries, and representative quotations were selected. Finally, the themes were written and integrated with the quantitative findings. Responses were organised in a spreadsheet environment to support the systematic retrieval of coded segments and the maintenance of an audit trail. A word cloud was generated from all responses as a descriptive overview of frequently occurring terms; this served as a supplementary visualisation only and not as a substitute for the interpretive analysis. 3.6.3 Qualitative Rigor and Trustworthiness Trustworthiness was assessed using Lincoln and Guba’s four criteria. Credibility was enhanced through multiple readings of the data, iterative coding and theme refinement, and grounding of interpretations in the participants’ own words. Inter-rater reliability was assessed by having a second independent coder analyse a random subsample of 47 responses (20% of total). The second coder independently applied the coding framework without knowledge of the primary coder’s assignments. Agreement was evaluated using Cohen’s kappa (κ), yielding κ = 0.72, indicating substantial agreement and exceeding the conventional κ ≥ 0.70 threshold (Landis & Koch, 1977 ). This level of agreement demonstrates that the coding scheme was consistently applied and that the identified themes reflected genuine data patterns rather than individual interpretive bias. Transferability is supported by providing a detailed description of Zanzibar higher education context, enabling readers to evaluate the applicability of the findings to other settings. Dependability was addressed by documenting the coding procedures, analytical decisions, and theme development in an audit trail. The inter-rater reliability assessment further supports dependability by demonstrating the consistency and reproducibility of the coding process. Confirmability was promoted by systematically linking all reported themes to multiple direct quotations, allowing readers to verify the relationship between the raw data and interpretations. 3.7 Data Integration Strategy Integration followed a convergence-at-interpretation strategy consistent with the mixed methods guidelines. Both analytical strands were completed independently before integration to maintain methodological rigor. The quantitative strand produced path coefficients, effect sizes, and R² values, whereas the qualitative strand produced themes with illustrative quotations. The integration proceeded in three steps. First, the findings were compared to identify convergence (agreement between strands), divergence (contradiction), and complementarity ( one strand elaborated on the other). Next, a joint display (Table 7 ) was constructed, aligning the research questions with the quantitative results and qualitative themes. Finally, integrated patterns are synthesised in Section 4.3 and elaborated in the Discussion, enabling statistical relationships to be contextualised by lived institutional conditions. 4. RESULTS 4.1 Quantitative Results 4.1.1 Sample Characteristics The final sample comprised 235 higher education participants, including lecturers and students from four institutions in Zanzibar, Tanzania. Students constituted the majority of respondents (n = 192, 81.7%), with lecturers comprising the remaining 18.3% (n = 43) of the sample. All participants voluntarily completed an online questionnaire. Missing data were minimal (< 2% across all items), and no cases were excluded due to excessive missing data. Table 1 summarises key demographic and professional characteristics. Female participants constituted 67.0% of the sample, and males constituted 33.0%. The mean age was 27.5 years (SD = 7.4, range = 19–63), consistent with a sample that includes both undergraduate and postgraduate students alongside lecturers. The mean academic/professional experience was 3.2 years (SD = 2.1, range = 0.5–7). Participants represented a range of disciplines, with science, education, nursing and midwifery, human resource management, and economics being the most frequently represented. Table 1 Sample Characteristics (N = 235) Variable Category n % Role Lecturer 43 18.3 Student 192 81.7 Gender Female 157 67.0 Male 78 33.0 Age (years) M (SD) 27.5 (7.4) — Range 19–63 — Experience (years) M (SD) 3.2 (2.1) — Range 0.5–7 — Faculty/Department Science with Education 35 14.9 Nursing and Midwifery 23 9.8 Human Resource Management 16 6.8 Economics 13 5.5 Arts with Education 12 5.1 Information Technology 11 4.7 Science in Information Technology 10 4.3 Telecommunications Engineering 10 4.3 Computer Science 9 3.8 Business Administration 9 3.8 Other disciplines* 77 32.8 Note. M = mean; SD = standard deviation. Percentages may not total 100 due to rounding errors. *Other disciplines include fields with fewer than 3% representation (e.g. social work, public administration, psychology, law, and environmental science). 4.1.2 Measurement Model Assessment Internal Consistency Reliability All three constructs demonstrated a high internal consistency. Cronbach’s alpha values ranged from 0.857 to 0.904, composite reliability (ρc) from 0.901 to 0.929, and rho_a (ρa) from 0.865 to 0.905. All values exceeded the threshold of 0.70 ( Table 2 ). Table 2 Construct Reliability and Convergent Validity Construct Cronbach’s α ρc ρa AVE Interpretation Perception 0.904 0.929 0.905 0.722 Excellent Preparedness 0.857 0.904 0.865 0.702 Good Readiness 0.859 0.901 0.874 0.647 Good Note. Thresholds: α, ρc, ρa ≥ 0.70; AVE ≥ 0.50. Convergent Validity The AVE values ranged from 0.647 to 0.722, all of which exceeded 0.50. Outer loadings for perception items ranged from 0.784 to 0.876, and for preparedness items, from 0.784 to 0.897. The readiness item loadings ranged from 0.641 to 0.901. One readiness indicator (Readiness4) produced an outer loading of 0.641, which fell below the conventional 0.70 threshold. Notably, this item’s cross-loading on preparedness (0.676) exceeded its loading on readiness, warranting caution in interpretation. Hair et al. ( 2019 ) recommend retaining such items only when composite reliability and AVE remain above the threshold after their inclusion; for readiness, both criteria were met (ρc = 0.901; AVE = 0.647). Nonetheless, this item’s ambiguous positioning across readiness and preparedness represents a limitation of the measurement model, which is revisited in Section 5.4 . Discriminant Validity HTMT values for Perception–Preparedness and Readiness–Preparedness were 0.403 and 0.677, respectively, both below the conservative threshold of 0.85. The HTMT value for Perception–Readiness was 0.941, exceeding both the conservative (0.85) and liberal (0.90) thresholds (Table 3 ). This result indicates that the two constructs are difficult to discriminate empirically, which is theoretically expected given that Perception is conceptualised as the proximal antecedent of Readiness in this model. The inner-model variance inflation factor (VIF) values for both paths predicting Preparedness were 3.591, below the 5.0 threshold, indicating that collinearity does not distort path coefficient estimates. Nevertheless, HTMT violation is a genuine methodological concern and is discussed as a limitation in Section 5.4 . Table 3 Discriminant Validity (HTMT) Construct Pair HTMT Threshold Assessment Perception ↔ Preparedness 0.403 < 0.85 Discriminant validity established Readiness ↔ Preparedness 0.677 < 0.85 Discriminant validity established Perception ↔ Readiness 0.941 < 0.85 / < 0.90 Exceeds both thresholds — see limitation Note. The inner model VIF for the paths predicting preparedness was 3.591 (both predictors), below the 5.0 acceptability threshold. The SRMR was 0.119, marginally above the 0.10 exploratory threshold; model parsimony (three constructs) and high inter-construct correlation likely inflated this statistic. 4.1.3 Structural Model and Mediation Explained Variance (R²) Perception explained 72.2% of the variance in readiness (R² = 0.722, R²adj = 0.720), indicating a significant effect. Together, perception and readiness explained 38.6% of the variance in preparedness (R² = 0.386, R²adj = 0.381), indicating a moderate explanatory power. Effect Sizes (f²) Perception had a very large effect on readiness (f² = 2.591). Readiness had a large effect on preparedness (f² = 0.425). The direct effect of Perception on Preparedness was small (f² = 0.099). Direct Paths All three hypothesised paths were statistically significant (p < .001). The path from Perception to Readiness was strong and positive (β = 0.849). Readiness exerted a strong positive effect on preparedness (β = 0.967). The direct path from Perception to Preparedness was negative and statistically significant (β = −0.467; Table 4 ). The confidence interval for the Readiness → Preparedness path extended slightly above 1.0 at its upper bound [0.786, 1.157], a pattern consistent with suppression effects in the presence of high predictor intercorrelation, as reflected in the HTMT of 0.941 between Perception and Readiness (see Section 5.4 ). Table 4 Structural Model Summary (Direct Effects, Effect Sizes, and Explained Variance) Path β SE t p f² Effect R² R²adj Perception → Readiness 0.849 0.018 48.320 < .001 2.591 Very large 0.722 0.720 Readiness → Preparedness 0.967 0.095 10.150 < .001 0.425 Large 0.386 0.381 Perception → Preparedness −0.467 0.123 3.803 < .001 0.099 Small Note. Note: β = standardised path coefficient; SE = standard error from 10,000 bootstrap resamples; 95% bias-corrected CIs: Perception → Readiness [0.816, 0.885]; Readiness → Preparedness [0.786, 1.157]; Perception → Preparedness [− 0.703, − 0.225]. Mediation Analysis Decomposing the effect of perception on preparedness revealed a competitive (inconsistent) mediation pattern (Zhao et al., 2010 ): the direct path from perception to preparedness was negative and significant (β = −0.467, p < .001), while the indirect path through readiness was positive and large (β = 0.822, p < .001; 95% CI [0.657, 1.004]). The total effect was positive and significant (β = 0.355, p < .001; 95% CI [0.177, 0.522]), indicating that the indirect pathway more than offset the negative direct path. Results are shown in Table 5 . Table 5 Mediation Analysis Effect Path β SE t p 95% CI Direct Perception → Preparedness −0.467 0.123 3.803 < .001 [− 0.703, − 0.225] Indirect Perception → Readiness → Preparedness 0.822 0.090 9.142 < .001 [0.657, 1.004] Total Perception → Preparedness 0.355 0.089 4.000 < .001 [0.177, 0.522] The pattern of competitive mediation reflects the dominance of the indirect pathway: readiness is the mechanism through which perception influences preparedness. The negative direct path (β = −0.467) should be interpreted cautiously, given the near-collinearity of perception and readiness (HTMT = 0.941), which introduces suppression effects that may distort the direct-path estimate. The interpretable summary is the positive total effect (β = 0.355), indicating that higher perceptions predict higher preparedness, primarily through building readiness. 4.2 Qualitative Results 4.2.1 Thematic Analysis The RTA of the 235 open-ended response sets yielded ten major themes: infrastructure gaps, skills and literacy gaps, cognitive risks, misinformation concerns, policy and leadership, education and curriculum, financial barriers, positive perceptions, support needs, and cultural considerations. Table 6 summarises these themes with representative codes and illustrative quotations. Table 6 Thematic Analysis Summary Theme Code Description Illustrative Quotation Infrastructure Gaps Poor Internet Unreliable or inaccessible network connectivity “Network access is very poor.” ICT Shortage Lack of computers, labs, or up-to-date technology “Shortages of ICT Infrastructure.” Electricity Issues Power instability limiting digital capacity “Infrastructure limitations.” Skills & Literacy Gaps Low Digital Literacy Lack of skills to use AI tools “Low level of literacy in Artificial Intelligence.” Lack of Training Absence of structured training programmes “Lack of proper training.” Poor Knowledge Low awareness and understanding of AI “Lack of proper knowledge concerning Artificial Intelligence.” Cognitive Risks Overreliance AI may replace critical thinking “Students based more on Artificial Intelligence rather than their own thinking.” Thinking Decline AI may lower intellectual engagement “AI hinders the thinking ability of students.” Laziness Fear AI promotes a shortcut culture “Lazy mindset to the users.” Misinformation Concerns Incorrect Answers AI tools may mislead learners “Sometimes it provides incorrect answer.” Outdated Content AI databases may not be current “Some data are out of update.” Policy & Leadership Weak Institutional Policy Absence of AI governance frameworks “My institution has no clear strategy for integrating AI.” Need for National Strategy Government role in investment and regulation “Government should provide financial support and strategic framework.” Education & Curriculum Curriculum Misalignment Curriculum unprepared for AI integration “Curriculum lagging behind current trends.” Content Relevance AI training needed across disciplines “AI should be trained to all students, not only ICT.” Financial Barriers High Internet Cost Cost of access as barrier “To reduce the internet cost.” Technology Cost Costly devices and infrastructure upgrades “Implementing AI can be expensive.” Positive Perceptions Helpful Tool AI seen as educationally valuable “AI is important for Zanzibar higher education.” Future-Oriented View Hope AI will advance local education “AI must be adopted in Zanzibar due to its importance.” Support Needs Training & Capacity Building Institutional training and workshops needed “To train the employer outside Zanzibar on how to use AI.” Strategic Partnerships External partnerships suggested “Strategic Partnership is very important for AI achievement.” Cultural Considerations Local Adaptation AI tools must align with Zanzibari context “Adopting AI in education must relate with our Zanzibar culture.” 4.2.2 Thematic Answers to Research Questions RQ1: Prevailing Perceptions of AI Integration Participants’ perceptions of AI were mixed but were broadly positive. Many viewed AI as a valuable and necessary tool for advancing education, describing it as “ important for Zanzibar higher education ” and asserting that “ AI must be adopted in Zanzibar due to its importance .” Respondents highlighted AI’s potential to enhance efficiency, support teaching, and save time for educators. Alongside these positive views, substantial concerns have emerged regarding cognitive risks. Several participants worried that students might depend on AI in ways that undermine their own reasoning abilities, such as, “ students based more on Artificial Intelligence rather than depending on their thinking ” and “ AI hinders the thinking ability of students .” Some linked this to a broader “ lazy mindset ” concern, fearing that AI would encourage shortcuts and reduce genuine intellectual efforts. Misinformation concerns were also prominent. Respondents noted that AI tools “ sometimes provide incorrect answers ” and that “ some data are out of date , ” raising questions about the reliability and currency of AI-generated information. On balance, teachers recognised AI’s potential but were acutely aware of the risks to student autonomy, critical thinking, and information quality. RQ2: Institutional Preparedness for AI Adoption Participants generally described their institutions as being insufficiently prepared for AI adoption. Infrastructure was the most frequently cited barrier. Respondents reported “ very poor ” network access, “ shortages of ICT infrastructure , ” and “ infrastructure limitations , ” pointing to unreliable Internet, limited device access, and unstable electricity as major constraints. Institutional policy and leadership emerged as additional weak points. Several respondents noted that their institutions “ have no clear strategy for integrating AI , ” indicating an absence of formal frameworks, guidelines, or strategic planning. This absence of policy direction compounded readiness problems at the institutional level. Pervasive skills and literacy gaps added a third layer of constraint. Participants described “ low levels of literacy in artificial intelligence ” and “ lack of proper training , ” noting that both staff and students lacked the competence needed to use AI tools effectively. These three categories–infrastructure, policy, and skills–collectively point to low institutional readiness, in which even genuine interest cannot overcome structural deficits. RQ3: Facilitators and Barriers to AI Acceptance and Integration The principal barriers were financial constraints (“implementing AI can be expensive,” “reduce the internet cost”), infrastructure challenges, and low digital literacy (“ lack of proper knowledge concerning artificial intelligence,” “low skills in computer technology ”). Cultural and pedagogical concerns also surfaced: excessive AI use was seen as potentially producing “lazy students” and weakening critical thinking. Several facilitating factors were identified. Many participants held fundamentally positive views of AI’s potential, describing it as “ very intelligent ” and expressing that it “ should be strongly promoted in Zanzibar higher education .” Respondents emphasised the need for government support, financial investment, and strategic frameworks, and called for training, capacity building, and strategic partnerships. Some suggested piloting AI initiatives before scaling them. The presence of positive attitudes and explicit calls for support indicates meaningful potential for progress, contingent on structural investments. 4.3 Integration of Quantitative and Qualitative Findings 4.3.1 Convergent Findings The integration of the two strands revealed consistent patterns across four areas. Importance of Perception. Perception strongly predicted readiness (β = 0.849, f² = 2.591), accounting for 72.2% of its variance. Teachers described AI as important and necessary for Zanzibar educational future, while articulating genuine concerns. Taken together, these findings show that teachers’ perceptions form a powerful starting point for adoption, broadly positive but not uncritical. Readiness as a Key Mechanism. Mediation analysis identified readiness as the principal pathway through which perception influences preparedness (indirect β = 0.822; readiness → preparedness β = 0.967). Qualitative data echoed this result: participants repeatedly emphasised that training and capacity building are required to translate favourable views of AI into actual preparedness. Readiness bridges attitudes and implementation, but only if institutional support is available. Infrastructure as a Structural Bottleneck. The model explained substantially less variance in Preparedness (R² = 0.386) than in Readiness (R² = 0.722), pointing to factors beyond perception and readiness that constrain preparedness. Qualitative data identified these factors explicitly: infrastructure gaps, poor internet connectivity, insufficient ICT resources, and electricity instability were the most frequently and forcefully mentioned barriers. Even strong readiness cannot produce preparedness in the absence of basic infrastructure. Skills and literacy as missing links. The moderate R² for preparedness also reflects the role of skills and literacy that were not quantitatively modelled. Qualitative themes identified low digital and AI literacy and inadequate training as core barriers, clarifying why perception and readiness alone leave a substantial portion of preparedness variance unexplained. Table 7 Convergent Mixed Methods Integration Summary Research Question Quantitative Finding Qualitative Finding Integrated Interpretation RQ1: Perceptions Perception strongly predicts Readiness (β = 0.849) Mixed but positive views; cognitive risk and misinformation concerns prominent Positive but critical perceptions create readiness potential; concerns about quality and cognition are well-founded and need pedagogical responses RQ2: Institutional Preparedness Preparedness R² = 38.6%; unexplained variance signals environmental constraints Infrastructure, skills gaps, weak policy, financial barriers Preparedness is constrained by specific, addressable structural factors, not solely by psychology. Targeted intervention at institutional level is needed RQ3: Facilitators and Barriers Readiness mediates Perception → Preparedness (indirect β = 0.822); total effect positive (β = 0.355) Barriers: infrastructure, skills, policy, cost; Facilitators: positive attitudes, calls for support The primary pathway to preparedness is readiness. Most critical barriers are structural; facilitators include training and institutional investment 5. DISCUSSION 5.1 Integrated Interpretation of Findings The quantitative and qualitative strands converge on a picture of GenAI adoption in Zanzibar that is more structurally constrained than psychologically resistant. Participants held broadly positive perceptions of GenAI, and these perceptions strongly predicted their readiness to adopt it (β = 0.849). Readiness drives preparedness (β = 0.967). However, this pathway from perception to readiness to preparedness is not self-perpetuating. It depends on training, institutional support, reliable infrastructure, and clear policies, all of which the qualitative data identify as absent or inadequate. The mediation analysis showed a competitive (inconsistent) mediation pattern: the direct path from Perception to Preparedness was negative (β = −0.467), whereas the indirect path through readiness was strongly positive (β = 0.822). The total effect was positive (β = 0.355), indicating that the indirect pathway dominated. This pattern is consistent with a model in which readiness is the operative mechanism linking Perception to Preparedness. Caution is required before substantively interpreting a negative direct path. The HTMT ratio for perception readiness (0.941) indicates near-collinearity between the two predictors, and the confidence interval for the readiness → preparedness path extends slightly above 1.0 at its upper bound for both suppression signs. When two predictors are highly intercorrelated, the direct path in a mediation model can become distorted and sign-reversed as a statistical consequence rather than a genuine substantive effect. Therefore, the negative direct effect is not interpreted as a meaningful psychological phenomenon. The interpretable finding is the total effect (β = 0.355, p < .001): positive perceptions, when they generate readiness, produce higher preparedness. 5.2 Theoretical Contributions First, they provide empirical evidence for readiness as a mediating construct positioned between perception and preparedness in the GenAI adoption pathway. Existing frameworks, including UTAUT, typically treat facilitating conditions as direct predictors of use behaviour. This study suggests that readiness, capturing the motivational and practical orientation toward adoption, operates as an intervening mechanism that must be actively built rather than assumed. In low-resource contexts, positive perceptions will not automatically generate readiness without deliberate investment in training and capacity development. Second, the competitive mediation pattern challenges simplified models that treat perceptions as uniformly beneficial. The conditions under which perception generates preparedness versus frustration appear to depend heavily on structural factors that are typically absent from psychological models of adoption. This highlights the need to adopt frameworks that explicitly incorporate infrastructure, policy, and training as moderating conditions rather than treating them as afterthoughts or limitations. Third, this study demonstrates the value of convergent mixed-methods designs in this research domain. Quantitative modelling identified the strength and direction of relationships among psychological constructs. Qualitative analysis revealed that infrastructural and institutional factors omitted from the model are decisive for real-world implementation. Neither strand alone would have produced this account of the process. 5.3 Practical Implications for Zanzibar Higher Education System The findings indicate five institutional priorities. Infrastructure investment is a prerequisite, not a preference. Reliable Internet access, adequate ICT facilities, and stable electricity are not supplementary support; they are prerequisites for meaningful AI integration. Institutions and government bodies that invest in AI awareness and training without addressing infrastructure are likely to generate awareness-capacity gaps, as suggested by the mediation model. Systematic capacity building is the mechanism through which readiness is developed. The strong path from readiness to preparedness (β = 0.967) indicates that training investments yield substantial dividends. Programs should extend beyond basic tool literacy to include pedagogical integration, ethical use, and critical evaluation of AI output skills that address the cognitive and misinformation concerns raised by the teachers. Institutional policies and governance frameworks should be developed proactively. Participants described the absence of clear institutional strategies for AI integration. Without policy direction, even well-resourced and willing teachers lack the guidance needed to implement AI in assessments and teaching in principled ways. Universities should develop explicit guidelines, ethical standards, and support mechanisms rather than leaving individual teachers to navigate these questions alone. Positive perceptions, although insufficient, are a resource. The very strong effect of perception on readiness (β = 0.849, f² = 2.591) indicates that the participants’ existing enthusiasm is a genuine asset. Structured showcases of AI use cases, peer-sharing activities, and well-designed awareness campaigns can translate enthusiasm into readiness, but only when paired with real training and infrastructure support. Local adaptation matters. Participants emphasised the need to align AI implementation with Zanzibar cultural context and educational values. Implementation strategies should be locally designed and piloted, avoiding the uncritical adoption of frameworks developed in resource-rich contexts, where the constraints and cultural dynamics are substantively different. 5.4 Limitations The sample composition ensures transparency. The participants included lecturers and students from four institutions, with students constituting the majority. The mean age of 27.5 years and mean experience of 3.2 years reflect this mixed composition of the sample. Generalisation is constrained accordingly: the findings apply to the broader higher-education community at these institutions rather than to faculty alone. The convenience sampling approach further limits inferences to specific institutions and the period of data collection. Self-reported measures introduce a potential common method bias, as all Likert-scale items and open-ended responses were collected from the same respondents at the same time. A common method bias test (e.g. Harman’s single-factor test) was not conducted; this is acknowledged as a limitation, and future studies should include this check to address this limitation. Self-reported measures may also be influenced by social desirability or inaccurate self-assessment of skills and preparedness of the respondents. The full texts of the 14 survey items are available from the corresponding author upon reasonable request. Future publications should include item texts in an appendix to enable the replication and assessment of face validity. The measurement model has several notable weaknesses. Readiness4’s outer loading of 0.641 fell below the recommended threshold of 0.70, and its cross-loading on Preparedness (0.676) exceeded its loading on its own construct, a cross-loading violation indicating ambiguity in how this item discriminated between the two constructs. Although the composite reliability and AVE for Readiness remained acceptable with the item retained, future studies should revise or replace this item. The HTMT ratio for perception readiness (0.941) exceeded both the conservative and liberal discriminant validity thresholds. The inter-model VIF of 3.591 is below the 5.0 threshold, and there are theoretical grounds for expecting high intercorrelations when perception is the direct antecedent of readiness. However, the confidence interval for the readiness → preparedness path extending slightly above 1.0, combined with the HTMT violation, indicates suppression effects that distort the direct path estimates. Accordingly, the negative direct path from perception to preparedness should not be interpreted as a substantive phenomenon; the interpretable finding is the total positive effect (β = 0.355). Future research with larger samples, refined items, and broader structural models should revisit these pathways. The SRMR of 0.119 marginally exceeds the 0.10 exploratory threshold (Henseler et al., 2015 ), suggesting that the three-construct model is somewhat underspecified. The qualitative strand confirms what is missing: infrastructure, policy, and digital literacy are decisive for preparedness but were not modelled quantitatively. Incorporating these as observed covariates in future work would likely improve model fit and explanatory power. The quantitative model’s R² for preparedness (0.386) leaves substantial variance unexplained, consistent with the above. Infrastructure, policy, financial resources, and digital literacy were not quantitatively modelled, which limits the structural results to the psychological pathway only. On the qualitative side, open-ended survey responses yielded less depth than semi-structured interviews, and variations in written expression may have influenced the richness of individual accounts. Thematic analysis is inherently interpretive; while inter-rater reliability was assessed (Cohen’s κ = 0.72, computed on 47 randomly selected responses representing approximately 200 individual coding decisions), alternative coding frameworks are possible. The unit of analysis for kappa was the individual coded segment rather than the respondent. 5.5 Conclusion The adoption of GenAI in Zanzibar higher education system is shaped by a combination of psychological orientation and structural constraints. Lecturer and student participants alike hold broadly positive views of AI, and these views translate into readiness through mechanisms that are theoretically sound and statistically robust. However, readiness does not automatically become preparedness; it requires infrastructure, training, institutional support, and policy clarity that are currently absent or underdeveloped in Zanzibar institutions. Pattern of competitive mediation. where positive perceptions can coexist with lower preparedness when structural conditions are inadequate captures the central tension facing higher education in this context. Strategies for promoting GenAI adoption that focus solely on attitudes and awareness will not resolve this tension. What is needed is a multilevel approach: building realistic and critical perceptions, developing readiness through deliberate capacity building, and removing structural constraints through investment in infrastructure and the development of clear institutional frameworks. Without this structural foundation, enthusiasm and willingness to adopt AI remain aspirational rather than operational. Declarations Conflict of Interest The authors declare no conflict of interest. Ethics This study was approved by the XX University Research Ethics Committee. Informed consent was obtained from all the participants. Funding This study did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors. Acknowledgments The authors would like to thank all survey respondents who contributed their time and insights to this study and XX University for facilitating access to key stakeholders. Data Availability The data supporting the findings of this study are available from the corresponding author upon reasonable requests. References Ajzen I (1991) The theory of planned behavior. Organ Behav Hum Decis Process 50(2):179–211. https://doi.org/10.1016/0749-5978(91)90020-T Akanzire BN, Nyaaba M, Nabang M (2025) Generative AI in teacher education: Teacher educators’ perception and preparedness. J Digit Educational Technol 5(1):ep2508. https://doi.org/10.30935/jdet/15887 Al-Emran M, Malik SI, Al-Kabi MN (2024) University teachers’ views on the adoption and integration of generative AI tools for student assessment in higher education. Educ Sci 14(10):1090. https://doi.org/10.3390/educsci14101090 Aluko T, Mkhize S (2025) Artificial intelligence in South African universities: Curriculum transformation and decolonisation—aid or obstacle? Front Sociol 10:1543471. https://doi.org/10.3389/fsoc.2025.1543471 Braun V, Clarke V (2019) Reflecting on reflexive thematic analysis. Qualitative Res Sport Exerc Health 11(4):589–597. https://doi.org/10.1080/2159676X.2019.1628806 Creswell JW, Plano Clark VL (2018) Designing and conducting mixed methods research, 3rd edn. SAGE Davis FD (1989) Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Q 13(3):319–340. https://doi.org/10.2307/249008 Feng J, Gao Y, Chen L, Liu X (2025) Key factors influencing educational technology adoption in higher education: A systematic review. PLOS Digit Health 4(4):e0000764. https://doi.org/10.1371/journal.pdig.0000764 Guo S, Shi L, Zhai X (2025) Developing and validating an instrument for teachers’ acceptance of artificial intelligence in education. Educ Inform Technol. https://doi.org/10.1007/s10639-025-13338-6 Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of PLS-SEM. Eur Bus Rev 31(1):2–24. https://doi.org/10.1108/EBR-11-2018-0203 Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135. https://doi.org/10.1007/s11747-014-0403-8 Kock N, Hadaya P (2018) Minimum sample size estimation in PLS-SEM: The inverse square root and gamma-exponential methods. Inform Syst J 28(1):227–261. https://doi.org/10.1111/isj.12131 Landis JR, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33(1):159–174. https://doi.org/10.2307/2529310 Macwan A, Barot B (2025) Exploring Factors Influencing Teacher Readiness and Perceptions Towards Digital Learning Technology: A Systematic Literature Review. Int J Multidisciplinary Res 7(6). https://doi.org/10.36948/ijfmr.2025.v07i06.59978 Maluleke AF (2025) AI adoption in African higher education: A systematic review of benefits and ethical implications. Interdisciplinary J Educ Res 7(2):a05. https://doi.org/10.38140/ijer-2025.vol7.2.05 Modiba FS, Van den Berg A, Mago S (2025) Opportunities and challenges of generative artificial intelligence supporting research in African classrooms. South Afr J High Educ 39(3):173–193. https://doi.org/10.20853/39-3-6272 Mtonga M, Mbewe O (2024) Profiling mathematics teacher educators’ readiness for digital technology integration: Evidence from Zambia. J Math Teacher Educ. https://doi.org/10.1007/s10857-024-09657-z Muringa TP (2025) Exploring ethical dilemmas and institutional challenges in AI adoption: A study of South African universities. Front Educ 10:1628019. https://doi.org/10.3389/feduc.2025.1628019 Mwapwele SD, Marais M, Dlamini S, Van Biljon J (2019) Teachers’ ICT Adoption in South African Rural Schools: A Study of Technology Readiness and Implications for the South Africa Connect Broadband Policy. The African Journal of Information and Communication (AJIC), (24). https://doi.org/10.23962/10539/28658 Patterson ME, Breytenbach J, Coffman I (2025) AI-adoption attitudes in Southern Africa’s higher education sector: A pilot survey using the COM-B model. Afr J Inform Communication 35:1–13. https://doi.org/10.23962/ajic.i35.21607 Ringle CM, Wende S, Becker J-M (2024) SmartPLS 4 . SmartPLS. https://www.smartpls.com Schreurs B, Schott T, Köppl M (2025) Trust in generative AI tools: A comparative study of higher education students, teachers, and researchers. Information 16(7):622. https://doi.org/10.3390/info16070622 Tasnim N, Chatterjee R, Sanda AS, Rashid F (2025) Facilitating conditions and e-learning persistence in Bangladeshi higher education: Navigating the post-pandemic landscape. Int J Social Media Mental Wellbeing 8(4). https://doi.org/10.37502/IJSMR.2025.8407 Thaldar D, Botes M, Badru A, Chenia H, Duma S, Dlamini SB, Chuturgoon A (2025) Generative AI governance in higher education: A case study from Africa. Front Political Sci 7:1666661. https://doi.org/10.3389/fpos.2025.1666661 Venkatesh V, Morris MG, Davis GB, Davis FD (2003) User acceptance of information technology: Toward a unified view. MIS Q 27(3):425–478. https://doi.org/10.2307/30036540 Viberg O, Mavroudi A, Khalil M, Bälter O (2020) Validating an Instrument to Measure Teachers’Preparedness to Use Digital Technology in their Teaching. Nordic J Digit Lit 15(1):38–54. https://doi.org/10.18261/issn.1891-943x-2020-01-04 Zhao X, Lynch JG Jr., Chen Q (2010) Reconsidering Baron and Kenny: Myths and truths about mediation analysis. J Consum Res 37(2):197–206. https://doi.org/10.1086/651257 Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted 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-9479399","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":626771436,"identity":"cd0bc80f-891a-4c3b-9007-7c44cd7afd96","order_by":0,"name":"Jecha Jecha","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+0lEQVRIiWNgGAWjYBACxgYGNhAtAyYZKoCYmbmBKC08EC1nQFoY8WsBAogWiAFtUGPwAeb25mcPfvw5zMPH3p34uXBebTR/O1DLj4ptuB3Wc8zcsLftMA8bz9nN0jO3Hc+dcZixgbHnzG3cWmbksEnwNgC1SORukObddiy3AaiFmbENj5b5b9gk/wAdxib/dvNv3jnHcucT1DKDh02ahw1kC+82ad6GmtwNBLX0pJlJy7alA/2Su82a59iB3I1ALQfx+cWw/fAzyTd/rOXk289uvs1TU5c77/zhgw9+VODR0oDKPwwmD+BUDwTyaPw6fIpHwSgYBaNghAIANU9WwHok/9QAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0002-7784-0994","institution":"Zanzibar University","correspondingAuthor":true,"prefix":"","firstName":"Jecha","middleName":"","lastName":"Jecha","suffix":""},{"id":626771437,"identity":"3fc14f88-98cc-4045-a6c3-7100ed10ea67","order_by":1,"name":"Hayfa Nassor","email":"","orcid":"","institution":"Zanzibar University","correspondingAuthor":false,"prefix":"","firstName":"Hayfa","middleName":"","lastName":"Nassor","suffix":""}],"badges":[],"createdAt":"2026-04-21 06:07:21","currentVersionCode":1,"declarations":{"humanSubjects":true,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":true,"humanSubjectConsent":true,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9479399/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9479399/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107520060,"identity":"d11572da-1a41-41cf-b468-b45c83455850","added_by":"auto","created_at":"2026-04-22 08:58:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":44347,"visible":true,"origin":"","legend":"\u003cp\u003eHypothesised Structural Model\u003c/p\u003e","description":"","filename":"Pathfindog.png","url":"https://assets-eu.researchsquare.com/files/rs-9479399/v1/12b7b9b4cdfc8573d1fb0df7.png"},{"id":107520059,"identity":"538c6bf0-d109-4ddd-a30b-49086c71fea3","added_by":"auto","created_at":"2026-04-22 08:58:34","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":50698,"visible":true,"origin":"","legend":"\u003cp\u003eFinal Structural Model with Standardised Path Coefficients and R² Values\u003c/p\u003e","description":"","filename":"Pathfind2.png","url":"https://assets-eu.researchsquare.com/files/rs-9479399/v1/47cc652b1a3b3d732acc4d00.png"},{"id":107520225,"identity":"85dfe762-7da8-4299-b02b-9c5b0a499146","added_by":"auto","created_at":"2026-04-22 08:59:29","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":809973,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9479399/v1/f96b5462-038b-463e-8157-071655a38aa7.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eNavigating the Future: Assessing the Impacts and Challenges of AI Integration in Higher Education Institutions in Zanzibar\u003c/p\u003e","fulltext":[{"header":"1. INTRODUCTION","content":"\u003cp\u003eArtificial intelligence is reshaping higher education in ways that are both exciting and uneven. Large language models and generative AI (GenAI) tools have entered classrooms, research workflows, and administrative systems at a pace that most institutions were not designed to absorb. The promise is genuine: more personalized learning, expanded access to information, and time freed from routine tasks for teaching and scholarship. However, realizing that promise requires infrastructure, institutional capacity, and policy frameworks that are far from uniformly available. While universities in North America and Western Europe debate the governance of GenAI use, many African institutions face a more immediate question: how to engage with transformative technologies when foundational infrastructure, trained personnel, and institutional governance are still developing. That gap is not simply a matter of being behind on the adoption curve; it reflects structural differences in resources, connectivity, and context that determine what adoption actually means and what it requires (Maluleke, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe urgency of this question is not hypothetical. The African Union\u0026rsquo;s Agenda 2063 explicitly positions AI as critical infrastructure for economic development and educational equity, and digital transformation initiatives across East Africa have placed technology integration at the center of national education strategies. Universities on the continent are beginning to incorporate GenAI into teaching, learning, and research. However, empirical evidence on how this adoption unfolds in practice, what faculty and students think, how ready they feel, and what actually prevents or enables integration remains thin, particularly in smaller and island-based higher education systems. Most documented studies on AI adoption in African higher education have concentrated on anglophone mainland countries such as Nigeria, Ghana, and South Africa, and many focus on student populations rather than educators (Maluleke, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Zanzibar, a small island higher education system within Tanzania, does not appear in this literature. This matters because island contexts carry distinct infrastructure vulnerabilities, unreliable internet, power instability, smaller talent pools, and limited institutional budgets that merit investigation in their own right, not as a footnote to mainland findings.\u003c/p\u003e \u003cp\u003eA second gap concerns how the adoption process itself is understood. Established technology adoption frameworks, such as the technology acceptance model (Davis, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e1989\u003c/span\u003e), theory of planned behavior (Ajzen, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1991\u003c/span\u003e), and unified theory of acceptance and use of technology (UTAUT; Venkatesh et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2003\u003c/span\u003e) have been productively applied to AI adoption in various African settings, often revealing that teacher attitudes and intentions are broadly positive, even when actual adoption lags (Patterson et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). These frameworks, however, primarily capture psychological variables, such as perceived usefulness and ease of use, as well as social influence. The pathway from a favorable attitude to genuine preparedness to teach with technology involves more than perception. It involves skill development, institutional support, reliable infrastructure, and policy clarity. Few studies examine this full pathway from perception through readiness to operational preparedness, and even fewer combine quantitative testing of psychological relationships with qualitative investigation of the structural and institutional conditions that enable or obstruct implementation. Consequently, we may know that teachers hold positive views of GenAI without understanding why they feel unprepared to deploy it in teaching.\u003c/p\u003e \u003cp\u003eA third gap concerns generative AI specifically. Research on AI in education predates the current GenAI moment by many years; however, the concerns raised by large language model hallucinations and misinformation risks, academic integrity, and questions about cognitive offloading are substantively different from those raised by earlier AI technologies and are only beginning to attract systematic empirical investigation. Teacher educators in Ghana, for instance, have been found to express both enthusiasm and anxiety about GenAI, with anxiety often centering on implications for critical thinking and equity (Akanzire et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Nevertheless, the factors that shape these orientations in smaller, lower-resource contexts and how they translate into preparedness for classroom implementation remain largely undocumented.\u003c/p\u003e \u003cp\u003eZanzibar provides an instructive case for this investigation. Its higher education system comprises a small number of institutions serving a growing student population in an environment where digital transformation is underway but constrained. Internet connectivity is unreliable, ICT resources are limited, and institutional capacity for managing rapid technological change is still developing. These are not peripheral challenges; they are central to explaining why positive attitudes toward AI do not automatically produce preparedness to integrate it into academic and professional practice (Muringa, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). By investigating Zanzibar, this study generates empirical evidence from a context entirely absent from the existing literature, while also offering findings of broader relevance to small, resource-constrained higher education systems across the Global South.\u003c/p\u003e \u003cp\u003eThese findings have direct implications for policy and practice. If preparedness depends not only on attitudes and intentions but also on infrastructure, training, institutional policy, and financial resources, then strategies for promoting GenAI adoption must address structural conditions alongside individual psychology. This study documents these conditions and identifies where institutional and governmental investments are most urgently needed.\u003c/p\u003e"},{"header":"2. LITERATURE REVIEW","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Generative AI Adoption in Higher Education\u003c/h2\u003e \u003cp\u003eGenerative AI has entered higher education institutions globally at a considerable speed, prompting substantial research attention. UTAUT has emerged as a dominant theoretical lens for understanding technology adoption in educational settings. Developed by integrating eight predecessor acceptance models, UTAUT proposes that performance expectancy, effort expectancy, social influence, and facilitating conditions predict both behavioural intention and actual technology use (Venkatesh et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). This framework has since been extended and applied extensively in AI adoption research. Al-Emran et al. (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) applied UTAUT to examine university teachers\u0026rsquo; views on generative AI tools for student assessment across the Middle East, drawing on 358 faculty members from multiple countries. Their findings showed that performance expectancy, effort expectancy, social influence, and hedonic motivation each shaped educators\u0026rsquo; behavioural intentions and actual use patterns. Institutional policies for GenAI integration were a significant driver, pointing to the importance of organisational structures alongside individual psychology.\u003c/p\u003e \u003cp\u003eTrust has attracted growing attention as a separate dimension of GenAI adoption. Schreurs et al. (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), in a comparative study of 823 higher education participants, including students, teachers, and researchers, found that trust in tools such as ChatGPT, Microsoft Copilot, and Google Gemini was experience-driven: frequency of use, duration, and self-assessed proficiency predicted trust, while demographic variables showed minimal influence. Trust, in turn, strongly predicted behavioral intention to adopt. These findings suggest that exposure and positive experience generate adoption momentum, a dynamic with particular relevance in contexts where limited infrastructure constrains initial exposure. Thaldar et al. (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) documented the institutional governance side of this challenge in African higher education, using the University of KwaZulu-Natal as a case study. They described it as one of the first African institutions to develop comprehensive academic guidelines for GenAI use and observed that institutional capacity building and formal policy frameworks remain nascent across most African universities; only South Africa, Rwanda, and Nigeria have begun to align institutional policies with national AI strategies.\u003c/p\u003e \u003cp\u003eResearch on pedagogical dimensions reveals concerns specific to GenAI\u0026rsquo;s transformative character. Akanzire et al. (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), examining teacher educator perspectives in Ghana, found that while educators expressed enthusiasm about GenAI\u0026rsquo;s educational potential, significant anxieties surfaced around critical thinking, equity of access, and appropriate assessment practices. This pattern echoes broader evidence that positive attitudes toward technology do not automatically translate into readiness to integrate it into teaching. Modiba et al. (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), in a systematic literature review of GenAI in African higher education, identified dual opportunities\u0026ndash;improved student writing, research productivity, and greater academic autonomy\u0026ndash;alongside substantial barriers: inadequate policy regulation, technical and structural challenges related to connectivity and device access, and knowledge deficits among educators and students.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Infrastructure Barriers, Teacher Readiness, and the Adoption Gap\u003c/h2\u003e \u003cp\u003eA persistent pattern in the adoption literature is the gap between positive attitudes and actual preparedness. Mwapwele et al. (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), applying the technology readiness index to teacher ICT adoption in South African rural schools, found that most surveyed teachers expressed optimism about ICT for teaching and learning; however, significant financial, technical, and digital skills deficits persisted. This finding maps enthusiasm without capacity directly onto contemporary GenAI research.\u003c/p\u003e \u003cp\u003eThe UTAUT theoretical architecture helps explain this disconnect. While performance expectancy and effort expectancy predict behavioural intention, facilitating conditions predict actual usage behaviour. A systematic review by Feng et al. (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), covering 39 studies from 2015 to 2024, confirmed this pattern: facilitating conditions infrastructure, technical support, organisational policy, and institutional resources emerged as critical enablers of implementation success; however, they are routinely under-addressed in practice. In developing countries, Tasnim et al. (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) drew on 321 Bangladeshi university students across 22 institutions to show that facilitating conditions, not performance expectancy alone, predicted e-learning continuance. Infrastructure reliability, content accessibility, network stability, technical training, and stakeholder collaboration were pivotal.\u003c/p\u003e \u003cp\u003eThe conceptual distinction between readiness and preparedness is important. Guo et al. (2024) validated a teacher acceptance of artificial intelligence (TAAI) instrument that captures five dimensions: perceived usefulness, perceived ease of use, behavioural intention, self-efficacy, and anxiety. This instrument measures psychological readiness. However, preparedness requires additional elements that acceptance models do not capture. Macwan et al. (2025), reviewing the literature on teacher readiness for digital technology, identified six clusters: cognitive readiness (perception, self-efficacy, prior experience), pedagogical readiness (integrating content with technology), affective readiness (anxiety, emotional attitudes), institutional readiness (infrastructure, policy support, governance), technological readiness (tool-specific competence), and sociocultural factors. Only the first three are primarily psychological, whereas the latter three are organisational and contextual. Therefore, preparedness encompasses psychological acceptance as well as institutional, infrastructural, and pedagogical enabling conditions.\u003c/p\u003e \u003cp\u003eAfrican higher-education contexts present particular structural complexities. Aluko and Mkhize (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) examined AI integration in South African universities through the lens of curriculum transformation and found that historically White universities had made greater progress owing to superior funding and international partnerships, while historically Black universities faced systemic barriers, including limited resources, inadequate infrastructure, and constrained technical capacity. This inequality stems not from differential attitudes but from structural resource disparities, a finding with clear parallels in lower income contexts across the continent. This pattern also appears in the broader EdTech adoption literature: teachers in resource-constrained settings tend to express positive attitudes toward technology; however, the translation of those attitudes into classroom practice is obstructed by poor infrastructure, insufficient training, and the absence of supportive national policies.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Strengths and Gaps in Existing Research\u003c/h2\u003e \u003cp\u003eResearch on technology adoption in higher education demonstrates methodological maturity in several respects. UTAUT provides a comprehensive, empirically validated framework integrating psychological and organisational variables; its predictive validity across cultural contexts and technology types has been confirmed in multiple meta-analyses and systematic reviews (Feng et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Macwan et al., 2025). Recent studies have shown growing sophistication in measuring both perceptions and enabling conditions, moving beyond the oversimplification of earlier attitudinal research. The validation of measurement instruments (Guo et al. 2024), such as the TAAI scale and Viberg \u0026rsquo;s(2020) digital preparedness instrument, provides reliable tools for quantitative work across contexts. African-centred research is also expanding, with context-sensitive studies from South Africa (Thaldar et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mwapwele et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Aluko \u0026amp; Mkhize, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), Ghana (Akanzire et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), Nigeria (Macwan et al., 2025), Zambia (Mtonga \u0026amp; Mbewe, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), and Bangladesh (Tasnim et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) providing alternatives to the uncritical transfer of Western models.\u003c/p\u003e \u003cp\u003eSeveral critical gaps remain. Geographic coverage is uneven: substantial research exists for North American and Western European contexts, African higher education is underrepresented, and within Africa, the literature concentrates on resource-rich institutions in South Africa, Nigeria, and Ghana. Small island higher education systems have not been studied. Zanzibar, in particular, has no documented empirical research on technology adoption in its universities, despite facing infrastructure challenges, geographic isolation, unreliable internet, and constrained budgets, which differ substantially from mainland African contexts.\u003c/p\u003e \u003cp\u003eResearch on GenAI, particularly, is nascent. Most African studies on GenAI adoption take the form of theoretical analyses, policy case studies, or systematic reviews rather than empirical measurements of teacher preparedness in particular institutional settings (Thaldar et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Modiba et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Akanzire et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Quantitative studies testing the full pathway from perceptions through readiness to measurable preparedness are rare. The conceptualisation of preparedness itself lacks consensus: few studies operationalise it as a multidimensional construct integrating psychological, pedagogical, technical, and institutional dimensions, and the conceptual separation between readiness (motivational and psychological orientation) and preparedness (demonstrated capacity to implement) has not been systematically examined. Most studies also use purely quantitative or purely qualitative designs, leaving the interplay between individual psychological factors and structural enablers incompletely understood. The role of policy and governance, distinct from attitudinal variables, has received relatively little attention.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Research Gap and Research Questions\u003c/h2\u003e \u003cp\u003eThe literature reveals a gap at the intersection of geography, methodology, and construct specification: no empirical research documents teacher perceptions, readiness, and preparedness for GenAI integration in Zanzibar \u0026rsquo;s higher education system or in comparable small island settings elsewhere in Africa. This study directly addresses this gap through a convergent mixed methods investigation.\u003c/p\u003e \u003cp\u003eFour research questions were used to organise the enquiry.\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eWhat are higher education participants\u0026rsquo; perceptions of the benefits, risks, and relevance of generative AI to higher education in Zanzibar?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTo what extent are higher education participants psychologically ready and willing to adopt generative AI in their academic and professional work, and what factors predict their readiness?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHow prepared are the participants and their institutions to effectively integrate generative AI, and what are the primary barriers to that preparedness?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eHow do institutional factors (infrastructure, policy, support, and training), pedagogical factors (content knowledge and teaching and learning approach), and individual factors (confidence, skills, and prior experience) interact to enable or obstruct the translation of readiness into preparedness?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eAnswering these questions generates evidence that directly informs policy and practice in Zanzibar and contributes context-sensitive data on technology adoption in resource-constrained, island-based settings, which are currently lacking in the literature.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. METHOD","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Research Design\u003c/h2\u003e \u003cp\u003eThis study used a convergent mixed methods design to examine higher education participants\u0026rsquo; perceptions of GenAI, their readiness to adopt GenAI, and their preparedness to integrate it into academic and professional practice in Zanzibar higher education institutions. In a convergent design, quantitative and qualitative data are collected concurrently, analysed separately using appropriate methods, and integrated during interpretation to develop a more complete account of the phenomenon (Creswell \u0026amp; Plano Clark, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Both strands were given equal priority. The quantitative strand estimated the relationships among perceptions, readiness, and preparedness, whereas the qualitative strand illuminated the contextual barriers and facilitators shaping those relationships. Integration focused on identifying the convergence, divergence, and complementarity between the strands.\u003c/p\u003e \u003cp\u003eThree considerations justified this design. The research questions required both quantitative hypothesis testing and qualitative contextualisation of institutional conditions. GenAI adoption in Zanzibar is an emerging area with little prior empirical work, making it important to capture both the breadth available through survey data and the depth available through open-ended accounts. Integrating both strands was expected to yield stronger, more contextually grounded evidence than either strand alone.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Quantitative Component\u003c/h2\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003e3.2.1 Design and Analytical Approach\u003c/h2\u003e \u003cp\u003eThe quantitative strand employed a cross-sectional survey design. Partial least squares structural equation modelling (PLS-SEM) was chosen as the primary analytical method because it is well suited to prediction-oriented research, models with mediation, and moderate sample sizes. PLS-SEM maximises the explained variance in key constructs rather than optimising global model fit, making it appropriate for exploratory and theory-building work, both of which describe the present study\u0026rsquo;s orientation toward GenAI adoption in Zanzibar higher education system (Hair et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e \u003ch2\u003e3.2.2 Context and Sampling\u003c/h2\u003e \u003cp\u003eData were collected from four universities in Zanzibar, Tanzania, between October and December 2024. Convenience sampling was used to recruit lecturers and students at these institutions. Therefore, the sample represents the broader higher education community, including those who teach with, learn with, or are likely to encounter GenAI tools in academic settings. This approach introduces potential selection bias and limits generalisability; however, it was appropriate given the exploratory orientation of the study, the absence of a comprehensive sampling frame across Zanzibar higher education institutions, the practical constraints of field data collection, and PLS-SEM tolerance for non-probability samples, where the emphasis is on prediction and explanation rather than population-level inference. To partially mitigate sampling bias, data were collected from four distinct institutions. The sample characteristics are reported in detail below, and the limitations on generalisation are explicitly acknowledged.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e3.2.3 Sample Size\u003c/h2\u003e \u003cp\u003eThe sample size adequacy for PLS-SEM was evaluated using the inverse square root method (Kock \u0026amp; Hadaya, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). For a model with three constructs and a maximum of two arrows entering any single construct, this method suggested a minimum of approximately 160 participants to detect medium effect sizes (f\u0026sup2; \u0026ge; 0.15) at α\u0026thinsp;=\u0026thinsp;0.05 and power\u0026thinsp;=\u0026thinsp;0.80. The final sample of 235 participants exceeded the threshold.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Measurement Instrument\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e3.3.1 Survey Development\u003c/h2\u003e \u003cp\u003eA structured online questionnaire was developed using Google Form. The instrument comprised three sections corresponding to the three latent constructs, each measured using multi-item Likert scales.\u003c/p\u003e \u003cp\u003ePerception of GenAI (five items) captured attitudes, beliefs, and evaluations of GenAI in educational settings, rated on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). Readiness for GenAI (five items) assessed self-reported confidence and willingness to adopt GenAI tools using the same response scale as above. Preparedness for GenAI integration (four items) measured the perceived competence and capability to integrate GenAI into academic, teaching, and professional practice. All items were adapted from established scales in the technology acceptance and educational technology literature, with minor modifications to reflect the GenAI domain and the Zanzibar institutional context.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e3.3.2 Measurement Model Specification\u003c/h2\u003e \u003cp\u003eAll three constructs were specified as reflective, latent variables. In reflective models, observed indicators are treated as manifestations of an underlying construct; changes in the latent construct are expected to be reflected across all indicators, and indicators are expected to be highly intercorrelated. The removal of any single item did not alter the construct\u0026rsquo;s conceptual domain. This specification aligns with the conceptualisation of perception, readiness, and preparedness as underlying psychological states, expressed through survey responses.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Data Collection Procedure\u003c/h2\u003e \u003cp\u003eThe survey link was distributed via institutional email lists and internal communication channels at the four participating universities, reaching both lecturers and registered students. Participation was voluntary; informed consent was obtained electronically at the start of the survey. No personally identifiable information was collected. This study was conducted in accordance with the ethical principles of the Declaration of XX University. All participants provided informed electronic consent prior to participation. The survey remained open for three months and yielded 235 complete responses. Google Forms built-in validation features minimised data entry errors. Cases with more than 20% missing data on construct items were excluded, consistent with standard PLS-SEM practice; no cases were excluded on this basis, as missing data were minimal across the dataset (\u0026lt;\u0026thinsp;2%).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Quantitative Data Analysis\u003c/h2\u003e \u003cdiv id=\"Sec18\" class=\"Section3\"\u003e \u003ch2\u003e3.5.1 Software and Analytical Steps\u003c/h2\u003e \u003cp\u003eQuantitative data were analysed using SmartPLS 4.0 (Ringle et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Following standard practice, the analysis proceeded in two stages: assessment of the measurement model (reliability and validity of the latent constructs) and evaluation of the structural model (relationships among the constructs).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section3\"\u003e \u003ch2\u003e3.5.2 PLS-SEM Algorithm and Bootstrapping\u003c/h2\u003e \u003cp\u003eThe PLS algorithm used a path weighting scheme with a maximum of 300 iterations and a stop criterion of 10⁻⁷; all results were standardised. All constructs were estimated using Mode A, consistent with their reflective specifications. The statistical significance of the path coefficients, indirect effects, and total effects was assessed through nonparametric bootstrapping with 10,000 resamples, producing standard errors, t-values, p-values, and 95% bias-corrected confidence intervals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section3\"\u003e \u003ch2\u003e3.5.3 Measurement Model Evaluation\u003c/h2\u003e \u003cp\u003eInternal consistency reliability was assessed using Cronbach\u0026rsquo;s alpha (α) and composite reliability (ρc and ρa), with values\u0026thinsp;\u0026ge;\u0026thinsp;0.70 indicating acceptable reliability. Convergent validity was evaluated using the average variance extracted (AVE; \u0026ge; 0.50). Indicator reliability was examined through outer loadings, with values\u0026thinsp;\u0026ge;\u0026thinsp;0.70 as the primary criterion. Hair et al. (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) note that indicators with loadings between 0.40 and 0.70 may be retained if their removal would not increase the AVE or composite reliability above the threshold. Discriminant validity was assessed using the heterotrait-monotrait ratio of correlations (HTMT), with 0.85 as the conservative threshold, and 0.90 as the more liberal criterion. To guard against multicollinearity in the measurement model, variance inflation factors (VIFs) were examined for all outer loadings, with VIF\u0026thinsp;\u0026lt;\u0026thinsp;5.0 indicating acceptable levels (Hair et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The inner model VIF values were similarly examined for the structural paths.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section3\"\u003e \u003ch2\u003e3.5.4 Structural Model Evaluation and Mediation\u003c/h2\u003e \u003cp\u003eThe structural model was evaluated using the following criteria: coefficients of determination (R\u0026sup2;; 0.25, 0.50, and 0.75 interpreted as weak, moderate, and substantial), effect sizes (f\u0026sup2;; 0.02, 0.15, and 0.35 as small, medium, and large), and path coefficients (β), with significance assessed via bootstrapped t-values (|t| \u0026ge; 1.96 at α\u0026thinsp;=\u0026thinsp;0.05, two-tailed). Model fit was examined using the Standardised Root Mean Square Residual (SRMR), with 0.10 as an acceptable threshold for PLS-SEM in exploratory research (Henseler et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMediation was examined by decomposing the effect of perception on preparedness into direct, indirect, and total components through readiness. The significance of the indirect and total effects was tested using bootstrapped confidence intervals. The patterns of direct and indirect effects were used to classify the mediation type following Zhao et al. (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2010\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Qualitative Component\u003c/h2\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003e3.6.1 Data Source\u003c/h2\u003e \u003cp\u003eQualitative data were collected through open-ended questions in the same Google Forms questionnaire. After completing the Likert-scale items, respondents were asked to provide open-ended responses to three questions, which were set as required fields in Google Forms to ensure complete data collection. All 235 respondents provided written responses, yielding robust data for thematic analysis. This embedded design ensured a direct link between the quantitative and qualitative data from the same participant group.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section3\"\u003e \u003ch2\u003e3.6.2 Qualitative Data Analysis\u003c/h2\u003e \u003cp\u003eOpen-ended responses were analysed using reflexive thematic analysis, following Braun and Clarke\u0026rsquo;s (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) six-phase approach. All responses were read repeatedly to achieve familiarity and identify the initial patterns. Meaningful units of text were then coded inductively, generating codes from the data rather than imposing them a priori. Related codes were grouped into candidate themes that described the key aspects of GenAI integration. Candidate themes were reviewed against the full dataset for internal coherence and distinctiveness of the themes. Themes were then defined and refined, clarifying the scope, core meaning, and boundaries, and representative quotations were selected. Finally, the themes were written and integrated with the quantitative findings.\u003c/p\u003e \u003cp\u003eResponses were organised in a spreadsheet environment to support the systematic retrieval of coded segments and the maintenance of an audit trail. A word cloud was generated from all responses as a descriptive overview of frequently occurring terms; this served as a supplementary visualisation only and not as a substitute for the interpretive analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003e3.6.3 Qualitative Rigor and Trustworthiness\u003c/h2\u003e \u003cp\u003eTrustworthiness was assessed using Lincoln and Guba\u0026rsquo;s four criteria. Credibility was enhanced through multiple readings of the data, iterative coding and theme refinement, and grounding of interpretations in the participants\u0026rsquo; own words. Inter-rater reliability was assessed by having a second independent coder analyse a random subsample of 47 responses (20% of total). The second coder independently applied the coding framework without knowledge of the primary coder\u0026rsquo;s assignments. Agreement was evaluated using Cohen\u0026rsquo;s kappa (κ), yielding κ\u0026thinsp;=\u0026thinsp;0.72, indicating substantial agreement and exceeding the conventional κ\u0026thinsp;\u0026ge;\u0026thinsp;0.70 threshold (Landis \u0026amp; Koch, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e1977\u003c/span\u003e). This level of agreement demonstrates that the coding scheme was consistently applied and that the identified themes reflected genuine data patterns rather than individual interpretive bias.\u003c/p\u003e \u003cp\u003eTransferability is supported by providing a detailed description of Zanzibar higher education context, enabling readers to evaluate the applicability of the findings to other settings. Dependability was addressed by documenting the coding procedures, analytical decisions, and theme development in an audit trail. The inter-rater reliability assessment further supports dependability by demonstrating the consistency and reproducibility of the coding process. Confirmability was promoted by systematically linking all reported themes to multiple direct quotations, allowing readers to verify the relationship between the raw data and interpretations.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e3.7 Data Integration Strategy\u003c/h2\u003e \u003cp\u003e Integration followed a convergence-at-interpretation strategy consistent with the mixed methods guidelines. Both analytical strands were completed independently before integration to maintain methodological rigor. The quantitative strand produced path coefficients, effect sizes, and R\u0026sup2; values, whereas the qualitative strand produced themes with illustrative quotations.\u003c/p\u003e \u003cp\u003eThe integration proceeded in three steps. First, the findings were compared to identify convergence (agreement between strands), divergence (contradiction), and complementarity ( one strand elaborated on the other). Next, a joint display (Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e) was constructed, aligning the research questions with the quantitative results and qualitative themes. Finally, integrated patterns are synthesised in Section \u003cspan refid=\"Sec35\" class=\"InternalRef\"\u003e4.3\u003c/span\u003e and elaborated in the Discussion, enabling statistical relationships to be contextualised by lived institutional conditions.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. RESULTS","content":"\u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Quantitative Results\u003c/h2\u003e \u003cdiv id=\"Sec29\" class=\"Section3\"\u003e \u003ch2\u003e4.1.1 Sample Characteristics\u003c/h2\u003e \u003cp\u003eThe final sample comprised 235 higher education participants, including lecturers and students from four institutions in Zanzibar, Tanzania. Students constituted the majority of respondents (n\u0026thinsp;=\u0026thinsp;192, 81.7%), with lecturers comprising the remaining 18.3% (n\u0026thinsp;=\u0026thinsp;43) of the sample. All participants voluntarily completed an online questionnaire. Missing data were minimal (\u0026lt;\u0026thinsp;2% across all items), and no cases were excluded due to excessive missing data.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarises key demographic and professional characteristics. Female participants constituted 67.0% of the sample, and males constituted 33.0%. The mean age was 27.5 years (SD\u0026thinsp;=\u0026thinsp;7.4, range\u0026thinsp;=\u0026thinsp;19\u0026ndash;63), consistent with a sample that includes both undergraduate and postgraduate students alongside lecturers. The mean academic/professional experience was 3.2 years (SD\u0026thinsp;=\u0026thinsp;2.1, range\u0026thinsp;=\u0026thinsp;0.5\u0026ndash;7). Participants represented a range of disciplines, with science, education, nursing and midwifery, human resource management, and economics being the most frequently represented.\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\u003eSample Characteristics (N\u0026thinsp;=\u0026thinsp;235)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003en\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e%\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRole\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLecturer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e18.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStudent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e192\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e81.7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e157\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e67.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e33.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eM (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27.5 (7.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19\u0026ndash;63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExperience (years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eM (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.2 (2.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.5\u0026ndash;7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFaculty/Department\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eScience with Education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e14.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNursing and Midwifery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHuman Resource Management\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEconomics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eArts with Education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInformation Technology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eScience in Information Technology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTelecommunications Engineering\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eComputer Science\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBusiness Administration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOther disciplines*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e32.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003cem\u003eNote.\u003c/em\u003e M\u0026thinsp;=\u0026thinsp;mean; SD\u0026thinsp;=\u0026thinsp;standard deviation. Percentages may not total 100 due to rounding errors. *Other disciplines include fields with fewer than 3% representation (e.g. social work, public administration, psychology, law, and environmental science).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section3\"\u003e \u003ch2\u003e4.1.2 Measurement Model Assessment\u003c/h2\u003e \u003cp\u003e \u003cb\u003eInternal Consistency Reliability\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAll three constructs demonstrated a high internal consistency. Cronbach\u0026rsquo;s alpha values ranged from 0.857 to 0.904, composite reliability (ρc) from 0.901 to 0.929, and rho_a (ρa) from 0.865 to 0.905. All values exceeded the threshold of 0.70 ( Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\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\u003eConstruct Reliability and Convergent Validity\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=\"char\" char=\".\" 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=\"char\" char=\".\" 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\u003eConstruct\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCronbach\u0026rsquo;s α\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eρc\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eρa\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAVE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eInterpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerception\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.904\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.929\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.905\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.722\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eExcellent\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePreparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.857\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.904\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.865\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.702\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eGood\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReadiness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.859\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.901\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.874\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.647\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eGood\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cem\u003eNote.\u003c/em\u003e Thresholds: α, ρc, ρa\u0026thinsp;\u0026ge;\u0026thinsp;0.70; AVE\u0026thinsp;\u0026ge;\u0026thinsp;0.50.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\u003cp\u003e\u003cb\u003eConvergent Validity\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThe AVE values ranged from 0.647 to 0.722, all of which exceeded 0.50. Outer loadings for perception items ranged from 0.784 to 0.876, and for preparedness items, from 0.784 to 0.897. The readiness item loadings ranged from 0.641 to 0.901. One readiness indicator (Readiness4) produced an outer loading of 0.641, which fell below the conventional 0.70 threshold. Notably, this item\u0026rsquo;s cross-loading on preparedness (0.676) exceeded its loading on readiness, warranting caution in interpretation. Hair et al. (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) recommend retaining such items only when composite reliability and AVE remain above the threshold after their inclusion; for readiness, both criteria were met (ρc\u0026thinsp;=\u0026thinsp;0.901; AVE\u0026thinsp;=\u0026thinsp;0.647). Nonetheless, this item\u0026rsquo;s ambiguous positioning across readiness and preparedness represents a limitation of the measurement model, which is revisited in Section \u003cspan refid=\"Sec41\" class=\"InternalRef\"\u003e5.4\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003eDiscriminant Validity\u003c/b\u003e \u003c/p\u003e \u003cp\u003eHTMT values for Perception\u0026ndash;Preparedness and Readiness\u0026ndash;Preparedness were 0.403 and 0.677, respectively, both below the conservative threshold of 0.85. The HTMT value for Perception\u0026ndash;Readiness was 0.941, exceeding both the conservative (0.85) and liberal (0.90) thresholds (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). This result indicates that the two constructs are difficult to discriminate empirically, which is theoretically expected given that Perception is conceptualised as the proximal antecedent of Readiness in this model. The inner-model variance inflation factor (VIF) values for both paths predicting Preparedness were 3.591, below the 5.0 threshold, indicating that collinearity does not distort path coefficient estimates. Nevertheless, HTMT violation is a genuine methodological concern and is discussed as a limitation in Section \u003cspan refid=\"Sec41\" class=\"InternalRef\"\u003e5.4\u003c/span\u003e.\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\u003eDiscriminant Validity (HTMT)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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=\"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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConstruct Pair\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHTMT\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThreshold\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAssessment\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerception \u0026harr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.403\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDiscriminant validity established\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReadiness \u0026harr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.677\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDiscriminant validity established\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerception \u0026harr; Readiness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.85 / \u0026lt; 0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eExceeds both thresholds \u0026mdash; see limitation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003cem\u003eNote.\u003c/em\u003e The inner model VIF for the paths predicting preparedness was 3.591 (both predictors), below the 5.0 acceptability threshold. The SRMR was 0.119, marginally above the 0.10 exploratory threshold; model parsimony (three constructs) and high inter-construct correlation likely inflated this statistic.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec31\" class=\"Section3\"\u003e \u003ch2\u003e4.1.3 Structural Model and Mediation\u003c/h2\u003e \u003cp\u003e \u003cb\u003eExplained Variance (R\u0026sup2;)\u003c/b\u003e \u003c/p\u003e \u003cp\u003ePerception explained 72.2% of the variance in readiness (R\u0026sup2; = 0.722, R\u0026sup2;adj\u0026thinsp;=\u0026thinsp;0.720), indicating a significant effect. Together, perception and readiness explained 38.6% of the variance in preparedness (R\u0026sup2; = 0.386, R\u0026sup2;adj\u0026thinsp;=\u0026thinsp;0.381), indicating a moderate explanatory power.\u003c/p\u003e \u003cp\u003e \u003cb\u003eEffect Sizes (f\u0026sup2;)\u003c/b\u003e \u003c/p\u003e \u003cp\u003ePerception had a very large effect on readiness (f\u0026sup2; = 2.591). Readiness had a large effect on preparedness (f\u0026sup2; = 0.425). The direct effect of Perception on Preparedness was small (f\u0026sup2; = 0.099).\u003c/p\u003e \u003cp\u003e \u003cb\u003eDirect Paths\u003c/b\u003e \u003c/p\u003e \u003cp\u003eAll three hypothesised paths were statistically significant (p \u0026lt; .001). The path from Perception to Readiness was strong and positive (β\u0026thinsp;=\u0026thinsp;0.849). Readiness exerted a strong positive effect on preparedness (β\u0026thinsp;=\u0026thinsp;0.967). The direct path from Perception to Preparedness was negative and statistically significant (β = \u0026minus;0.467; Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). The confidence interval for the Readiness \u0026rarr; Preparedness path extended slightly above 1.0 at its upper bound [0.786, 1.157], a pattern consistent with suppression effects in the presence of high predictor intercorrelation, as reflected in the HTMT of 0.941 between Perception and Readiness (see Section \u003cspan refid=\"Sec41\" class=\"InternalRef\"\u003e5.4\u003c/span\u003e).\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\u003eStructural Model Summary (Direct Effects, Effect Sizes, and Explained Variance)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\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=\"char\" char=\".\" 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=\"char\" char=\".\" 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=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePath\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eβ\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003et\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ep\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ef\u0026sup2;\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eEffect\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eR\u0026sup2;\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eR\u0026sup2;adj\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerception \u0026rarr; Readiness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.849\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.018\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e48.320\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.591\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eVery large\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.722\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.720\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReadiness \u0026rarr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.967\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.095\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e10.150\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.425\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eLarge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.386\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.381\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerception \u0026rarr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u0026minus;0.467\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.123\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.803\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.099\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSmall\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"9\"\u003e\u003cem\u003eNote.\u003c/em\u003e Note: β\u0026thinsp;=\u0026thinsp;standardised path coefficient; SE\u0026thinsp;=\u0026thinsp;standard error from 10,000 bootstrap resamples; 95% bias-corrected CIs: Perception \u0026rarr; Readiness [0.816, 0.885]; Readiness \u0026rarr; Preparedness [0.786, 1.157]; Perception \u0026rarr; Preparedness [\u0026minus;\u0026thinsp;0.703, \u0026minus;\u0026thinsp;0.225].\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e\u003cb\u003eMediation Analysis\u003c/b\u003e\u003c/p\u003e \u003cp\u003eDecomposing the effect of perception on preparedness revealed a competitive (inconsistent) mediation pattern (Zhao et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2010\u003c/span\u003e): the direct path from perception to preparedness was negative and significant (β = \u0026minus;0.467, p \u0026lt; .001), while the indirect path through readiness was positive and large (β\u0026thinsp;=\u0026thinsp;0.822, p \u0026lt; .001; 95% CI [0.657, 1.004]). The total effect was positive and significant (β\u0026thinsp;=\u0026thinsp;0.355, p \u0026lt; .001; 95% CI [0.177, 0.522]), indicating that the indirect pathway more than offset the negative direct path. Results are shown in Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\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\u003eMediation Analysis\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" 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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEffect\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePath\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eβ\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003et\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ep\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDirect\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePerception \u0026rarr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.467\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.123\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e3.803\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;0.703, \u0026minus;\u0026thinsp;0.225]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIndirect\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePerception \u0026rarr; Readiness \u0026rarr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.822\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.090\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e9.142\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[0.657, 1.004]\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=\"left\" colname=\"c2\"\u003e \u003cp\u003ePerception \u0026rarr; Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.355\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.089\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e[0.177, 0.522]\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\u003eThe pattern of competitive mediation reflects the dominance of the indirect pathway: readiness is the mechanism through which perception influences preparedness. The negative direct path (β = \u0026minus;0.467) should be interpreted cautiously, given the near-collinearity of perception and readiness (HTMT\u0026thinsp;=\u0026thinsp;0.941), which introduces suppression effects that may distort the direct-path estimate. The interpretable summary is the positive total effect (β\u0026thinsp;=\u0026thinsp;0.355), indicating that higher perceptions predict higher preparedness, primarily through building readiness.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Qualitative Results\u003c/h2\u003e \u003cdiv id=\"Sec33\" class=\"Section3\"\u003e \u003ch2\u003e4.2.1 Thematic Analysis\u003c/h2\u003e \u003cp\u003eThe RTA of the 235 open-ended response sets yielded ten major themes: infrastructure gaps, skills and literacy gaps, cognitive risks, misinformation concerns, policy and leadership, education and curriculum, financial barriers, positive perceptions, support needs, and cultural considerations. Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e summarises these themes with representative codes and illustrative quotations.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eThematic Analysis Summary\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTheme\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCode\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIllustrative Quotation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInfrastructure Gaps\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoor Internet\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUnreliable or inaccessible network connectivity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Network access is very poor.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eICT Shortage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLack of computers, labs, or up-to-date technology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Shortages of ICT Infrastructure.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eElectricity Issues\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePower instability limiting digital capacity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Infrastructure limitations.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSkills \u0026amp; Literacy Gaps\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLow Digital Literacy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLack of skills to use AI tools\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Low level of literacy in Artificial Intelligence.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLack of Training\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAbsence of structured training programmes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Lack of proper training.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoor Knowledge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLow awareness and understanding of AI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Lack of proper knowledge concerning Artificial Intelligence.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCognitive Risks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverreliance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI may replace critical thinking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Students based more on Artificial Intelligence rather than their own thinking.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eThinking Decline\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI may lower intellectual engagement\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;AI hinders the thinking ability of students.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLaziness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFear AI promotes a shortcut culture\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Lazy mindset to the users.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMisinformation Concerns\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIncorrect Answers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI tools may mislead learners\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Sometimes it provides incorrect answer.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOutdated Content\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI databases may not be current\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Some data are out of update.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePolicy \u0026amp; Leadership\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eWeak Institutional Policy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAbsence of AI governance frameworks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;My institution has no clear strategy for integrating AI.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNeed for National Strategy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGovernment role in investment and regulation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Government should provide financial support and strategic framework.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation \u0026amp; Curriculum\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCurriculum Misalignment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCurriculum unprepared for AI integration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Curriculum lagging behind current trends.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eContent Relevance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI training needed across disciplines\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;AI should be trained to all students, not only ICT.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFinancial Barriers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHigh Internet Cost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCost of access as barrier\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;To reduce the internet cost.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTechnology Cost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCostly devices and infrastructure upgrades\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Implementing AI can be expensive.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePositive Perceptions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHelpful Tool\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI seen as educationally valuable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;AI is important for Zanzibar higher education.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFuture-Oriented View\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHope AI will advance local education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;AI must be adopted in Zanzibar due to its importance.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSupport Needs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTraining \u0026amp; Capacity Building\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInstitutional training and workshops needed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;To train the employer outside Zanzibar on how to use AI.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStrategic Partnerships\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eExternal partnerships suggested\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Strategic Partnership is very important for AI achievement.\u0026rdquo;\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCultural Considerations\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLocal Adaptation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI tools must align with Zanzibari context\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003e\u0026ldquo;Adopting AI in education must relate with our Zanzibar culture.\u0026rdquo;\u003c/em\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 \u003c/div\u003e \u003cdiv id=\"Sec34\" class=\"Section3\"\u003e \u003ch2\u003e4.2.2 Thematic Answers to Research Questions\u003c/h2\u003e \u003cp\u003e \u003cb\u003eRQ1: Prevailing Perceptions of AI Integration\u003c/b\u003e \u003c/p\u003e \u003cp\u003eParticipants\u0026rsquo; perceptions of AI were mixed but were broadly positive. Many viewed AI as a valuable and necessary tool for advancing education, describing it as \u0026ldquo;\u003cem\u003eimportant for Zanzibar higher education\u003c/em\u003e\u0026rdquo; and asserting that \u0026ldquo;\u003cem\u003eAI must be adopted in Zanzibar due to its importance\u003c/em\u003e.\u0026rdquo; Respondents highlighted AI\u0026rsquo;s potential to enhance efficiency, support teaching, and save time for educators.\u003c/p\u003e \u003cp\u003eAlongside these positive views, substantial concerns have emerged regarding cognitive risks. Several participants worried that students might depend on AI in ways that undermine their own reasoning abilities, such as, \u0026ldquo;\u003cem\u003estudents based more on Artificial Intelligence rather than depending on their thinking\u003c/em\u003e\u0026rdquo; and \u0026ldquo;\u003cem\u003eAI hinders the thinking ability of students\u003c/em\u003e.\u0026rdquo; Some linked this to a broader \u0026ldquo;\u003cem\u003elazy mindset\u003c/em\u003e\u0026rdquo; concern, fearing that AI would encourage shortcuts and reduce genuine intellectual efforts.\u003c/p\u003e \u003cp\u003eMisinformation concerns were also prominent. Respondents noted that AI tools \u0026ldquo;\u003cem\u003esometimes provide incorrect answers\u003c/em\u003e\u0026rdquo; and that \u0026ldquo;\u003cem\u003esome data are out of date\u003c/em\u003e, \u0026rdquo; raising questions about the reliability and currency of AI-generated information. On balance, teachers recognised AI\u0026rsquo;s potential but were acutely aware of the risks to student autonomy, critical thinking, and information quality.\u003c/p\u003e \u003cp\u003e \u003cb\u003eRQ2: Institutional Preparedness for AI Adoption\u003c/b\u003e \u003c/p\u003e \u003cp\u003eParticipants generally described their institutions as being insufficiently prepared for AI adoption. Infrastructure was the most frequently cited barrier. Respondents reported \u0026ldquo;\u003cem\u003every poor\u003c/em\u003e\u0026rdquo; network access, \u0026ldquo;\u003cem\u003eshortages of ICT infrastructure\u003c/em\u003e, \u0026rdquo; and \u0026ldquo;\u003cem\u003einfrastructure limitations\u003c/em\u003e, \u0026rdquo; pointing to unreliable Internet, limited device access, and unstable electricity as major constraints.\u003c/p\u003e \u003cp\u003eInstitutional policy and leadership emerged as additional weak points. Several respondents noted that their institutions \u0026ldquo;\u003cem\u003ehave no clear strategy for integrating AI\u003c/em\u003e, \u0026rdquo; indicating an absence of formal frameworks, guidelines, or strategic planning. This absence of policy direction compounded readiness problems at the institutional level.\u003c/p\u003e \u003cp\u003ePervasive skills and literacy gaps added a third layer of constraint. Participants described \u0026ldquo;\u003cem\u003elow levels of literacy in artificial intelligence\u003c/em\u003e\u0026rdquo; and \u0026ldquo;\u003cem\u003elack of proper training\u003c/em\u003e, \u0026rdquo; noting that both staff and students lacked the competence needed to use AI tools effectively. These three categories\u0026ndash;infrastructure, policy, and skills\u0026ndash;collectively point to low institutional readiness, in which even genuine interest cannot overcome structural deficits.\u003c/p\u003e \u003cp\u003e \u003cb\u003eRQ3: Facilitators and Barriers to AI Acceptance and Integration\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe principal barriers were financial constraints (\u0026ldquo;implementing AI can be expensive,\u0026rdquo; \u0026ldquo;reduce the internet cost\u0026rdquo;), infrastructure challenges, and low digital literacy (\u0026ldquo;\u003cem\u003elack of proper knowledge concerning artificial intelligence,\u0026rdquo; \u0026ldquo;low skills in computer technology\u003c/em\u003e\u0026rdquo;). Cultural and pedagogical concerns also surfaced: excessive AI use was seen as potentially producing \u0026ldquo;lazy students\u0026rdquo; and weakening critical thinking.\u003c/p\u003e \u003cp\u003eSeveral facilitating factors were identified. Many participants held fundamentally positive views of AI\u0026rsquo;s potential, describing it as \u0026ldquo;\u003cem\u003every intelligent\u003c/em\u003e\u0026rdquo; and expressing that it \u0026ldquo;\u003cem\u003eshould be strongly promoted in Zanzibar higher education\u003c/em\u003e.\u0026rdquo; Respondents emphasised the need for government support, financial investment, and strategic frameworks, and called for training, capacity building, and strategic partnerships. Some suggested piloting AI initiatives before scaling them. The presence of positive attitudes and explicit calls for support indicates meaningful potential for progress, contingent on structural investments.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec35\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Integration of Quantitative and Qualitative Findings\u003c/h2\u003e \u003cdiv id=\"Sec36\" class=\"Section3\"\u003e \u003ch2\u003e4.3.1 Convergent Findings\u003c/h2\u003e \u003cp\u003eThe integration of the two strands revealed consistent patterns across four areas.\u003c/p\u003e \u003cp\u003e \u003cb\u003eImportance of Perception.\u003c/b\u003e Perception strongly predicted readiness (β\u0026thinsp;=\u0026thinsp;0.849, f\u0026sup2; = 2.591), accounting for 72.2% of its variance. Teachers described AI as important and necessary for Zanzibar educational future, while articulating genuine concerns. Taken together, these findings show that teachers\u0026rsquo; perceptions form a powerful starting point for adoption, broadly positive but not uncritical.\u003c/p\u003e \u003cp\u003e \u003cb\u003eReadiness as a Key Mechanism.\u003c/b\u003e Mediation analysis identified readiness as the principal pathway through which perception influences preparedness (indirect β\u0026thinsp;=\u0026thinsp;0.822; readiness \u0026rarr; preparedness β\u0026thinsp;=\u0026thinsp;0.967). Qualitative data echoed this result: participants repeatedly emphasised that training and capacity building are required to translate favourable views of AI into actual preparedness. Readiness bridges attitudes and implementation, but only if institutional support is available.\u003c/p\u003e \u003cp\u003e \u003cb\u003eInfrastructure as a Structural Bottleneck.\u003c/b\u003e The model explained substantially less variance in Preparedness (R\u0026sup2; = 0.386) than in Readiness (R\u0026sup2; = 0.722), pointing to factors beyond perception and readiness that constrain preparedness. Qualitative data identified these factors explicitly: infrastructure gaps, poor internet connectivity, insufficient ICT resources, and electricity instability were the most frequently and forcefully mentioned barriers. Even strong readiness cannot produce preparedness in the absence of basic infrastructure.\u003c/p\u003e \u003cp\u003e \u003cb\u003eSkills and literacy as missing links.\u003c/b\u003e The moderate R\u0026sup2; for preparedness also reflects the role of skills and literacy that were not quantitatively modelled. Qualitative themes identified low digital and AI literacy and inadequate training as core barriers, clarifying why perception and readiness alone leave a substantial portion of preparedness variance unexplained.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eConvergent Mixed Methods Integration Summary\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResearch Question\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eQuantitative Finding\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQualitative Finding\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIntegrated Interpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRQ1: Perceptions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePerception strongly predicts Readiness (β\u0026thinsp;=\u0026thinsp;0.849)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMixed but positive views; cognitive risk and misinformation concerns prominent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePositive but critical perceptions create readiness potential; concerns about quality and cognition are well-founded and need pedagogical responses\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRQ2: Institutional Preparedness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePreparedness R\u0026sup2; = 38.6%; unexplained variance signals environmental constraints\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInfrastructure, skills gaps, weak policy, financial barriers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePreparedness is constrained by specific, addressable structural factors, not solely by psychology. Targeted intervention at institutional level is needed\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRQ3: Facilitators and Barriers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eReadiness mediates Perception \u0026rarr; Preparedness (indirect β\u0026thinsp;=\u0026thinsp;0.822); total effect positive (β\u0026thinsp;=\u0026thinsp;0.355)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBarriers: infrastructure, skills, policy, cost; Facilitators: positive attitudes, calls for support\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eThe primary pathway to preparedness is readiness. Most critical barriers are structural; facilitators include training and institutional investment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"5. DISCUSSION","content":"\u003cdiv id=\"Sec38\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Integrated Interpretation of Findings\u003c/h2\u003e \u003cp\u003eThe quantitative and qualitative strands converge on a picture of GenAI adoption in Zanzibar that is more structurally constrained than psychologically resistant. Participants held broadly positive perceptions of GenAI, and these perceptions strongly predicted their readiness to adopt it (β\u0026thinsp;=\u0026thinsp;0.849). Readiness drives preparedness (β\u0026thinsp;=\u0026thinsp;0.967). However, this pathway from perception to readiness to preparedness is not self-perpetuating. It depends on training, institutional support, reliable infrastructure, and clear policies, all of which the qualitative data identify as absent or inadequate.\u003c/p\u003e \u003cp\u003eThe mediation analysis showed a competitive (inconsistent) mediation pattern: the direct path from Perception to Preparedness was negative (β = \u0026minus;0.467), whereas the indirect path through readiness was strongly positive (β\u0026thinsp;=\u0026thinsp;0.822). The total effect was positive (β\u0026thinsp;=\u0026thinsp;0.355), indicating that the indirect pathway dominated. This pattern is consistent with a model in which readiness is the operative mechanism linking Perception to Preparedness.\u003c/p\u003e \u003cp\u003eCaution is required before substantively interpreting a negative direct path. The HTMT ratio for perception readiness (0.941) indicates near-collinearity between the two predictors, and the confidence interval for the readiness \u0026rarr; preparedness path extends slightly above 1.0 at its upper bound for both suppression signs. When two predictors are highly intercorrelated, the direct path in a mediation model can become distorted and sign-reversed as a statistical consequence rather than a genuine substantive effect. Therefore, the negative direct effect is not interpreted as a meaningful psychological phenomenon. The interpretable finding is the total effect (β\u0026thinsp;=\u0026thinsp;0.355, p \u0026lt; .001): positive perceptions, when they generate readiness, produce higher preparedness.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec39\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Theoretical Contributions\u003c/h2\u003e \u003cp\u003eFirst, they provide empirical evidence for readiness as a mediating construct positioned between perception and preparedness in the GenAI adoption pathway. Existing frameworks, including UTAUT, typically treat facilitating conditions as direct predictors of use behaviour. This study suggests that readiness, capturing the motivational and practical orientation toward adoption, operates as an intervening mechanism that must be actively built rather than assumed. In low-resource contexts, positive perceptions will not automatically generate readiness without deliberate investment in training and capacity development.\u003c/p\u003e \u003cp\u003eSecond, the competitive mediation pattern challenges simplified models that treat perceptions as uniformly beneficial. The conditions under which perception generates preparedness versus frustration appear to depend heavily on structural factors that are typically absent from psychological models of adoption. This highlights the need to adopt frameworks that explicitly incorporate infrastructure, policy, and training as moderating conditions rather than treating them as afterthoughts or limitations.\u003c/p\u003e \u003cp\u003eThird, this study demonstrates the value of convergent mixed-methods designs in this research domain. Quantitative modelling identified the strength and direction of relationships among psychological constructs. Qualitative analysis revealed that infrastructural and institutional factors omitted from the model are decisive for real-world implementation. Neither strand alone would have produced this account of the process.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec40\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Practical Implications for Zanzibar Higher Education System\u003c/h2\u003e \u003cp\u003eThe findings indicate five institutional priorities.\u003c/p\u003e \u003cp\u003eInfrastructure investment is a prerequisite, not a preference. Reliable Internet access, adequate ICT facilities, and stable electricity are not supplementary support; they are prerequisites for meaningful AI integration. Institutions and government bodies that invest in AI awareness and training without addressing infrastructure are likely to generate awareness-capacity gaps, as suggested by the mediation model.\u003c/p\u003e \u003cp\u003eSystematic capacity building is the mechanism through which readiness is developed. The strong path from readiness to preparedness (β\u0026thinsp;=\u0026thinsp;0.967) indicates that training investments yield substantial dividends. Programs should extend beyond basic tool literacy to include pedagogical integration, ethical use, and critical evaluation of AI output skills that address the cognitive and misinformation concerns raised by the teachers.\u003c/p\u003e \u003cp\u003eInstitutional policies and governance frameworks should be developed proactively. Participants described the absence of clear institutional strategies for AI integration. Without policy direction, even well-resourced and willing teachers lack the guidance needed to implement AI in assessments and teaching in principled ways. Universities should develop explicit guidelines, ethical standards, and support mechanisms rather than leaving individual teachers to navigate these questions alone.\u003c/p\u003e \u003cp\u003ePositive perceptions, although insufficient, are a resource. The very strong effect of perception on readiness (β\u0026thinsp;=\u0026thinsp;0.849, f\u0026sup2; = 2.591) indicates that the participants\u0026rsquo; existing enthusiasm is a genuine asset. Structured showcases of AI use cases, peer-sharing activities, and well-designed awareness campaigns can translate enthusiasm into readiness, but only when paired with real training and infrastructure support.\u003c/p\u003e \u003cp\u003eLocal adaptation matters. Participants emphasised the need to align AI implementation with Zanzibar cultural context and educational values. Implementation strategies should be locally designed and piloted, avoiding the uncritical adoption of frameworks developed in resource-rich contexts, where the constraints and cultural dynamics are substantively different.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec41\" class=\"Section2\"\u003e \u003ch2\u003e5.4 Limitations\u003c/h2\u003e \u003cp\u003eThe sample composition ensures transparency. The participants included lecturers and students from four institutions, with students constituting the majority. The mean age of 27.5 years and mean experience of 3.2 years reflect this mixed composition of the sample. Generalisation is constrained accordingly: the findings apply to the broader higher-education community at these institutions rather than to faculty alone. The convenience sampling approach further limits inferences to specific institutions and the period of data collection.\u003c/p\u003e \u003cp\u003eSelf-reported measures introduce a potential common method bias, as all Likert-scale items and open-ended responses were collected from the same respondents at the same time. A common method bias test (e.g. Harman\u0026rsquo;s single-factor test) was not conducted; this is acknowledged as a limitation, and future studies should include this check to address this limitation. Self-reported measures may also be influenced by social desirability or inaccurate self-assessment of skills and preparedness of the respondents.\u003c/p\u003e \u003cp\u003eThe full texts of the 14 survey items are available from the corresponding author upon reasonable request. Future publications should include item texts in an appendix to enable the replication and assessment of face validity.\u003c/p\u003e \u003cp\u003eThe measurement model has several notable weaknesses. Readiness4\u0026rsquo;s outer loading of 0.641 fell below the recommended threshold of 0.70, and its cross-loading on Preparedness (0.676) exceeded its loading on its own construct, a cross-loading violation indicating ambiguity in how this item discriminated between the two constructs. Although the composite reliability and AVE for Readiness remained acceptable with the item retained, future studies should revise or replace this item.\u003c/p\u003e \u003cp\u003eThe HTMT ratio for perception readiness (0.941) exceeded both the conservative and liberal discriminant validity thresholds. The inter-model VIF of 3.591 is below the 5.0 threshold, and there are theoretical grounds for expecting high intercorrelations when perception is the direct antecedent of readiness. However, the confidence interval for the readiness \u0026rarr; preparedness path extending slightly above 1.0, combined with the HTMT violation, indicates suppression effects that distort the direct path estimates. Accordingly, the negative direct path from perception to preparedness should not be interpreted as a substantive phenomenon; the interpretable finding is the total positive effect (β\u0026thinsp;=\u0026thinsp;0.355). Future research with larger samples, refined items, and broader structural models should revisit these pathways.\u003c/p\u003e \u003cp\u003eThe SRMR of 0.119 marginally exceeds the 0.10 exploratory threshold (Henseler et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), suggesting that the three-construct model is somewhat underspecified. The qualitative strand confirms what is missing: infrastructure, policy, and digital literacy are decisive for preparedness but were not modelled quantitatively. Incorporating these as observed covariates in future work would likely improve model fit and explanatory power.\u003c/p\u003e \u003cp\u003eThe quantitative model\u0026rsquo;s R\u0026sup2; for preparedness (0.386) leaves substantial variance unexplained, consistent with the above. Infrastructure, policy, financial resources, and digital literacy were not quantitatively modelled, which limits the structural results to the psychological pathway only.\u003c/p\u003e \u003cp\u003eOn the qualitative side, open-ended survey responses yielded less depth than semi-structured interviews, and variations in written expression may have influenced the richness of individual accounts. Thematic analysis is inherently interpretive; while inter-rater reliability was assessed (Cohen\u0026rsquo;s κ\u0026thinsp;=\u0026thinsp;0.72, computed on 47 randomly selected responses representing approximately 200 individual coding decisions), alternative coding frameworks are possible. The unit of analysis for kappa was the individual coded segment rather than the respondent.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec42\" class=\"Section2\"\u003e \u003ch2\u003e5.5 Conclusion\u003c/h2\u003e \u003cp\u003eThe adoption of GenAI in Zanzibar higher education system is shaped by a combination of psychological orientation and structural constraints. Lecturer and student participants alike hold broadly positive views of AI, and these views translate into readiness through mechanisms that are theoretically sound and statistically robust. However, readiness does not automatically become preparedness; it requires infrastructure, training, institutional support, and policy clarity that are currently absent or underdeveloped in Zanzibar institutions.\u003c/p\u003e \u003cp\u003ePattern of competitive mediation. where positive perceptions can coexist with lower preparedness when structural conditions are inadequate captures the central tension facing higher education in this context. Strategies for promoting GenAI adoption that focus solely on attitudes and awareness will not resolve this tension. What is needed is a multilevel approach: building realistic and critical perceptions, developing readiness through deliberate capacity building, and removing structural constraints through investment in infrastructure and the development of clear institutional frameworks. Without this structural foundation, enthusiasm and willingness to adopt AI remain aspirational rather than operational.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eConflict of Interest\u003c/h2\u003e \u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e \u003ch2\u003eEthics\u003c/h2\u003e \u003cp\u003eThis study was approved by the XX University Research Ethics Committee. Informed consent was obtained from all the participants.\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThis study did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\u003ch2\u003eAcknowledgments\u003c/h2\u003e \u003cp\u003eThe authors would like to thank all survey respondents who contributed their time and insights to this study and XX University for facilitating access to key stakeholders.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e \u003cp\u003eThe data supporting the findings of this study are available from the corresponding author upon reasonable requests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAjzen I (1991) The theory of planned behavior. Organ Behav Hum Decis Process 50(2):179\u0026ndash;211. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0749-5978(91)90020-T\u003c/span\u003e\u003cspan address=\"10.1016/0749-5978(91)90020-T\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAkanzire BN, Nyaaba M, Nabang M (2025) Generative AI in teacher education: Teacher educators\u0026rsquo; perception and preparedness. J Digit Educational Technol 5(1):ep2508. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.30935/jdet/15887\u003c/span\u003e\u003cspan address=\"10.30935/jdet/15887\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAl-Emran M, Malik SI, Al-Kabi MN (2024) University teachers\u0026rsquo; views on the adoption and integration of generative AI tools for student assessment in higher education. Educ Sci 14(10):1090. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/educsci14101090\u003c/span\u003e\u003cspan address=\"10.3390/educsci14101090\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAluko T, Mkhize S (2025) Artificial intelligence in South African universities: Curriculum transformation and decolonisation\u0026mdash;aid or obstacle? Front Sociol 10:1543471. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fsoc.2025.1543471\u003c/span\u003e\u003cspan address=\"10.3389/fsoc.2025.1543471\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraun V, Clarke V (2019) Reflecting on reflexive thematic analysis. Qualitative Res Sport Exerc Health 11(4):589\u0026ndash;597. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/2159676X.2019.1628806\u003c/span\u003e\u003cspan address=\"10.1080/2159676X.2019.1628806\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCreswell JW, Plano Clark VL (2018) Designing and conducting mixed methods research, 3rd edn. SAGE\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis FD (1989) Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Q 13(3):319\u0026ndash;340. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/249008\u003c/span\u003e\u003cspan address=\"10.2307/249008\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng J, Gao Y, Chen L, Liu X (2025) Key factors influencing educational technology adoption in higher education: A systematic review. PLOS Digit Health 4(4):e0000764. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pdig.0000764\u003c/span\u003e\u003cspan address=\"10.1371/journal.pdig.0000764\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo S, Shi L, Zhai X (2025) Developing and validating an instrument for teachers\u0026rsquo; acceptance of artificial intelligence in education. Educ Inform Technol. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-025-13338-6\u003c/span\u003e\u003cspan address=\"10.1007/s10639-025-13338-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of PLS-SEM. Eur Bus Rev 31(1):2\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1108/EBR-11-2018-0203\u003c/span\u003e\u003cspan address=\"10.1108/EBR-11-2018-0203\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHenseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in variance-based structural equation modeling. J Acad Mark Sci 43(1):115\u0026ndash;135. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11747-014-0403-8\u003c/span\u003e\u003cspan address=\"10.1007/s11747-014-0403-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKock N, Hadaya P (2018) Minimum sample size estimation in PLS-SEM: The inverse square root and gamma-exponential methods. Inform Syst J 28(1):227\u0026ndash;261. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/isj.12131\u003c/span\u003e\u003cspan address=\"10.1111/isj.12131\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLandis JR, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33(1):159\u0026ndash;174. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/2529310\u003c/span\u003e\u003cspan address=\"10.2307/2529310\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMacwan A, Barot B (2025) Exploring Factors Influencing Teacher Readiness and Perceptions Towards Digital Learning Technology: A Systematic Literature Review. Int J Multidisciplinary Res 7(6). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.36948/ijfmr.2025.v07i06.59978\u003c/span\u003e\u003cspan address=\"10.36948/ijfmr.2025.v07i06.59978\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaluleke AF (2025) AI adoption in African higher education: A systematic review of benefits and ethical implications. Interdisciplinary J Educ Res 7(2):a05. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.38140/ijer-2025.vol7.2.05\u003c/span\u003e\u003cspan address=\"10.38140/ijer-2025.vol7.2.05\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eModiba FS, Van den Berg A, Mago S (2025) Opportunities and challenges of generative artificial intelligence supporting research in African classrooms. South Afr J High Educ 39(3):173\u0026ndash;193. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.20853/39-3-6272\u003c/span\u003e\u003cspan address=\"10.20853/39-3-6272\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMtonga M, Mbewe O (2024) Profiling mathematics teacher educators\u0026rsquo; readiness for digital technology integration: Evidence from Zambia. J Math Teacher Educ. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10857-024-09657-z\u003c/span\u003e\u003cspan address=\"10.1007/s10857-024-09657-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuringa TP (2025) Exploring ethical dilemmas and institutional challenges in AI adoption: A study of South African universities. Front Educ 10:1628019. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2025.1628019\u003c/span\u003e\u003cspan address=\"10.3389/feduc.2025.1628019\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMwapwele SD, Marais M, Dlamini S, Van Biljon J (2019) Teachers\u0026rsquo; ICT Adoption in South African Rural Schools: A Study of Technology Readiness and Implications for the South Africa Connect Broadband Policy. The African Journal of Information and Communication (AJIC), (24). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.23962/10539/28658\u003c/span\u003e\u003cspan address=\"10.23962/10539/28658\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatterson ME, Breytenbach J, Coffman I (2025) AI-adoption attitudes in Southern Africa\u0026rsquo;s higher education sector: A pilot survey using the COM-B model. Afr J Inform Communication 35:1\u0026ndash;13. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.23962/ajic.i35.21607\u003c/span\u003e\u003cspan address=\"10.23962/ajic.i35.21607\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRingle CM, Wende S, Becker J-M (2024) \u003cem\u003eSmartPLS 4\u003c/em\u003e. SmartPLS. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.smartpls.com\u003c/span\u003e\u003cspan address=\"https://www.smartpls.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchreurs B, Schott T, K\u0026ouml;ppl M (2025) Trust in generative AI tools: A comparative study of higher education students, teachers, and researchers. Information 16(7):622. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/info16070622\u003c/span\u003e\u003cspan address=\"10.3390/info16070622\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTasnim N, Chatterjee R, Sanda AS, Rashid F (2025) Facilitating conditions and e-learning persistence in Bangladeshi higher education: Navigating the post-pandemic landscape. Int J Social Media Mental Wellbeing 8(4). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.37502/IJSMR.2025.8407\u003c/span\u003e\u003cspan address=\"10.37502/IJSMR.2025.8407\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThaldar D, Botes M, Badru A, Chenia H, Duma S, Dlamini SB, Chuturgoon A (2025) Generative AI governance in higher education: A case study from Africa. Front Political Sci 7:1666661. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpos.2025.1666661\u003c/span\u003e\u003cspan address=\"10.3389/fpos.2025.1666661\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVenkatesh V, Morris MG, Davis GB, Davis FD (2003) User acceptance of information technology: Toward a unified view. MIS Q 27(3):425\u0026ndash;478. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/30036540\u003c/span\u003e\u003cspan address=\"10.2307/30036540\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eViberg O, Mavroudi A, Khalil M, B\u0026auml;lter O (2020) Validating an Instrument to Measure Teachers\u0026rsquo;Preparedness to Use Digital Technology in their Teaching. Nordic J Digit Lit 15(1):38\u0026ndash;54. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18261/issn.1891-943x-2020-01-04\u003c/span\u003e\u003cspan address=\"10.18261/issn.1891-943x-2020-01-04\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao X, Lynch JG Jr., Chen Q (2010) Reconsidering Baron and Kenny: Myths and truths about mediation analysis. J Consum Res 37(2):197\u0026ndash;206. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1086/651257\u003c/span\u003e\u003cspan address=\"10.1086/651257\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Zanzibar University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"generative AI, higher education, PLS-SEM, technology adoption, readiness, preparedness, mixed methods, Zanzibar, Africa","lastPublishedDoi":"10.21203/rs.3.rs-9479399/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9479399/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eGenerative artificial intelligence (GenAI) is entering higher education systems across Africa at a pace that outstrips institutional preparedness, yet empirical evidence from small, resource-constrained island settings remains absent from the literature. This study examined GenAI perceptions, readiness, and preparedness among higher education participants in Zanzibar, Tanzania, a small island system with no prior documentation in the technology adoption literature. A convergent mixed methods design was employed: 235 participants (192 students and 43 lecturers) from four institutions completed a structured online questionnaire, providing both Likert-scale responses analysed via partial least squares structural equation modelling (PLS-SEM) and open-ended responses analysed using reflexive thematic analysis. PLS-SEM results showed that perception strongly predicted readiness (β\u0026thinsp;=\u0026thinsp;0.849, R\u0026sup2; = 0.722) and that readiness was the primary mechanism linking perception to preparedness (indirect β\u0026thinsp;=\u0026thinsp;0.822; total effect β\u0026thinsp;=\u0026thinsp;0.355, p \u0026lt; .001). Thematic analysis of 235 open-ended response sets yielded ten themes spanning infrastructure deficits, skills and digital literacy gaps, absent institutional policy, financial barriers, cognitive risks, and broadly positive orientations toward AI\u0026rsquo;s educational potential. Integration of the two strands showed that structural factors\u0026ndash;unreliable internet, insufficient ICT resources, and the absence of institutional governance frameworks\u0026ndash;explain why positive perceptions and readiness do not automatically translate into preparedness. These findings challenge simplified adoption models that treat psychological variables as sufficient and demonstrate that in low-resource island contexts, awareness and motivation require concurrent investment in infrastructure, training, and institutional policy to produce genuine implementation capacity. The practical implications for Zanzibar higher education sector and comparable systems across the Global South are discussed.\u003c/p\u003e","manuscriptTitle":"Navigating the Future: Assessing the Impacts and Challenges of AI Integration in Higher Education Institutions in Zanzibar","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-22 08:56:49","doi":"10.21203/rs.3.rs-9479399/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"27e7a922-bbdb-44c9-8b3f-04a30e48cb2b","owner":[],"postedDate":"April 22nd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-22T08:56:50+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-22 08:56:49","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9479399","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9479399","identity":"rs-9479399","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}