Adaptive Financial Education and Addiction Recovery: Associations with Self-Efficacy and Treatment Engagement in a Quasi-Experimental Study

Adaptive Financial Education and Addiction Recovery: Associations with Self-Efficacy and Treatment Engagement in a Quasi-Experimental Study | 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 Adaptive Financial Education and Addiction Recovery: Associations with Self-Efficacy and Treatment Engagement in a Quasi-Experimental Study Mohammad Hannan Mia, Mahadi Mokbul Ali This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9324305/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 Financial stress is associated with substance use disorder severity and relapse. Among individuals entering addiction treatment, financial hardship is highly prevalent. Mechanisms linking financial interventions to addiction outcomes remain underspecified. This study examines whether an adaptive financial education intervention is associated with improvements in financial self-efficacy, delay discounting, and treatment-related outcomes among individuals in addiction treatment and at-risk university students. The study is descriptive; causal claims are not warranted given design limitations. Participants (N = 420; 240 in outpatient treatment, 180 at-risk university students) were assigned by site to 12 weeks of adaptive financial education delivered through an interactive platform with personalized feedback, or to traditional classroom instruction. Multilevel models accounting for site-level clustering tested condition associations with financial self-efficacy, delay discounting, perceived stress, treatment engagement, and relapse intention. With only five sites, between-site variance estimates are unstable and all effects require cautious interpretation. The adaptive condition was associated with larger gains in financial self-efficacy (γ = 8.2, 95% CI [3.1, 13.3], p = .004) and larger reductions in delay discounting (γ = 0.41, 95% CI [0.10, 0.72], p = .012). Financial self-efficacy gains correlated with treatment engagement (r = .34, 95% CI [.22, .45]) and relapse intention (r = − .29, 95% CI [-.41, − .16]). All effects trended uniformly in predicted directions; replication is required before confidence is warranted. Adaptive financial education is associated with psychological outcomes that may support addiction recovery. Findings require replication in randomized designs with active controls and larger numbers of sites. Financial education Self-efficacy Addiction recovery Treatment engagement Adaptive learning HIGHLIGHTS • Adaptive financial education was associated with improved financial self-efficacy • Self-efficacy gains correlated with treatment engagement and lower relapse intention • Financial hardship is prevalent among individuals entering addiction treatment • Results are descriptive; causal claims require stronger designs • The five-site design produces unstable between-site variance estimates INTRODUCTION 1.1 Financial Stress and Addiction: Financial stress is consistently associated with substance use disorder severity and relapse vulnerability. Among individuals entering addiction treatment, financial hardship is highly prevalent. McLellan's work on recovery capital identifies financial resources as a key domain supporting sustained recovery. Compton and colleagues documented in the National Epidemiologic Survey on Alcohol and Related Conditions that individuals with substance use disorders have significantly lower income and higher rates of financial distress than the general population. Longitudinal research suggests financial difficulties precede relapse episodes: in one study, financial stressors were associated with increased relapse risk over 12-month follow-up. The mechanisms linking financial circumstances to addictive behavior remain underspecified. One plausible pathway involves self-regulatory capacity. Financial decisions require tradeoffs between immediate gratification and long-term goals, engaging cognitive processes also implicated in substance use decisions. Individuals who struggle to delay gratification in financial contexts may show similar patterns in substance use contexts. If self-regulatory capacity transfers across domains, interventions strengthening self-regulation in financial contexts might produce benefits for addiction outcomes. 1.2 The Self-Regulation Transfer Debate: The hypothesis that self-regulatory capacity transfers across domains has generated extensive research and controversy. The strength model of self-control proposes that self-regulation resembles a muscle that can be strengthened through exercise. Early studies reported that practicing self-control in one domain improved performance in unrelated domains. However large-scale replication efforts have produced mixed results. Hagger and colleagues conducted a multilab preregistered replication of the ego-depletion effect with 23 laboratories and 2,141 participants, finding a near-zero effect. Friese and colleagues subsequently reviewed the ego depletion literature, concluding that if the effect exists, it is substantially smaller than originally claimed and highly context-dependent. De Ridder and colleagues conducted a meta-analysis of self-control research, finding that self-control predicts behavior primarily through habitual responses rather than effortful inhibition. Duckworth and colleagues reviewed self-control in academic and health contexts, emphasizing that domain-specific skills and strategies may matter more than domain-general capacity. The transfer hypothesis is therefore contested rather than established. An alternative perspective emphasizes domain-specific pathways. Improving financial decision-making may reduce substance use through direct mechanisms: reduced financial stress, increased resources for treatment participation, or improved future orientation specifically tied to financial goals. These pathways do not require domain-general capacity and are theoretically simpler. This study cannot distinguish these explanations. 1.3 Adaptive Learning Technologies as Intervention Platforms: Adaptive learning platforms use algorithms to personalize content based on user performance, providing feedback calibrated to individual skill levels. Systematic reviews of digital interventions for substance use have identified technology-delivered interventions as feasible and potentially effective. Marsch and colleagues documented that technology-based interventions for substance use disorders can be effective when they incorporate evidence-based therapeutic principles. Torous and colleagues reviewed digital mental health interventions, noting that engagement and personalization are critical for effectiveness. Kiluk and colleagues demonstrated in a randomized trial that computer-based cognitive behavioral therapy for substance use produced outcomes comparable to clinician-delivered CBT, establishing feasibility and preliminary efficacy. Adaptive platforms differ from traditional instruction in ways that may influence psychological outcomes. Real-time feedback makes decision consequences visible. Personalized difficulty maintains users in optimal challenge zones where mastery experiences accumulate. Goal-setting features support self-monitoring and progress tracking. Self-determination theory (Deci & Ryan, 2000) provides a framework for understanding these features: adaptive platforms may support autonomy through user control, competence through mastery experiences, and relatedness through non-judgmental feedback. Whether these features, delivered as a package, are associated with outcomes relevant to addiction recovery is the descriptive question this study examines. 1.4 Research Gap and Study Objectives: Previous research has not examined whether adaptive financial education is associated with psychological outcomes relevant to addiction recovery. Studies of financial interventions in treatment populations have focused on concrete outcomes like employment and housing rather than psychological mechanisms. Studies of adaptive learning have focused on educational attainment rather than clinical outcomes. This study addresses three descriptive questions. First, is adaptive financial education associated with larger gains in financial self-efficacy than traditional classroom instruction? Second, are gains in financial self-efficacy associated with treatment engagement and relapse intention? Third, do these patterns differ between individuals actively in treatment versus at-risk university students? All findings are reported as associations. Causal claims are not warranted given design limitations including quasi-experimental assignment, small number of sites, lack of active control, and absence of experimental manipulation of proposed mechanisms. 1.5 Pre-Registration: This study was not pre-registered. The absence of pre-registration should be considered when interpreting results, as the analyses reported are exploratory rather than confirmatory. THEORETICAL PERSPECTIVES 2.1 Self-Efficacy as a Mechanism: Bandura proposed that perceived self-efficacy influences behavior through multiple pathways: goal setting, effort expenditure, persistence, and emotional reactions to failure. Self-efficacy beliefs are domain-specific but may generalize across closely related domains. The Financial Self-Efficacy Scale developed by Lown has been validated in general adult populations but not specifically in addiction treatment samples. This study examines its associations with treatment-related outcomes without assuming mechanism. 2.2 Delay Discounting as an Indicator: Delay discounting refers to the tendency to devalue rewards as a function of delay to receipt. Steep discounting characterizes addiction populations and predicts treatment outcomes. Discounting can be modified through intervention, and financial decisions inherently involve temporal tradeoffs. This study examines whether the adaptive condition is associated with discounting changes. 2.3 Self-Determination Theory and Adaptive Platforms: Self-determination theory identifies autonomy, competence, and relatedness as fundamental psychological needs supporting intrinsic motivation. Adaptive platforms may support competence through mastery experiences calibrated to individual skill levels. They may support autonomy through user control over pacing and content selection. They may support relatedness through feedback that feels responsive and non-judgmental. These features could enhance engagement and persistence, potentially explaining associations with outcomes regardless of domain-general transfer. 2.4 Contested Theories and Replication Failures: Key theoretical foundations for this work are contested. The strength model of self-control has failed large-scale replication. De Ridder and colleagues' meta-analysis suggests self-control operates primarily through habits rather than effortful inhibition. Friese and colleagues reviewed evidence that ego depletion effects, if real, are small and context-dependent. Duckworth and colleagues emphasize domain-specific strategies and skills. We reference these theories as frameworks that generated hypotheses but do not claim them as established foundations. METHODS 3.1 Participants: Participants were 420 individuals recruited from three outpatient substance abuse treatment centers and one large public university in the Midwestern United States. The treatment subsample (n = 240) comprised adults aged 18 to 65 (M = 38.2, SD = 10.4) currently enrolled in outpatient treatment for substance use disorder. Inclusion criteria were current treatment enrollment and self-reported financial concerns interfering with daily functioning. Exclusion criteria were active psychosis, cognitive impairment preventing informed consent, and imminent hospitalization risk. The prevention subsample (n = 180) comprised university students aged 18 to 25 (M = 22.1, SD = 2.3) identified as at-risk through screening during first-year orientation. The screening instrument included items from established surveys regarding substance use patterns, family history, and personal concerns. Students scoring above validated thresholds were invited to participate. Inclusion criteria were enrollment as full-time undergraduate students and at-risk status based on screening. Exclusion criteria were current treatment for substance use disorder and previous diagnosis of addiction. Overall sample characteristics: 52% female, 48% male; 68% White, 18% Black, 9% Hispanic, 5% other categories. The treatment and prevention subsamples differ systematically on age and clinical variables. They are analyzed separately where appropriate and combined with caution. 3.2 Design: A quasi-experimental design assigned sites rather than individuals to condition to minimize contamination. The three treatment centers and the university were randomly assigned to condition. Two treatment centers and the university were assigned to the adaptive financial education condition. One treatment center was assigned to the traditional instruction control condition. This assignment resulted in 210 participants in the adaptive condition (120 treatment, 90 prevention) and 210 participants in the control condition (120 treatment, 90 prevention). With only five total sites, between-site variance estimates are unstable and condition effects should be interpreted with caution. Intervention duration was 12 weeks with weekly 60-minute sessions. Assessment occurred at baseline, immediate post-intervention, and 3-month follow-up. Research assistants blind to condition administered all assessments following standardized protocols. 3.3 Adaptive Financial Education Intervention: The adaptive financial education platform was developed specifically for this study. Technical specifications are provided to enable replication. Platform architecture: The system used a rule-based adaptive algorithm rather than machine learning. User responses triggered branching logic that adjusted subsequent content difficulty. The algorithm tracked performance accuracy and response time, selecting from a bank of 347 financial scenarios organized by difficulty level and content domain. Content domains: Budgeting, saving, debt management, financial goal-setting, and long-term planning. Scenarios presented realistic situations requiring allocation decisions across competing needs. Examples included monthly budget allocation given fixed income, savings decisions for anticipated expenses, and debt repayment prioritization. Adaptive mechanism: After each scenario, the system displayed feedback comparing user decisions to normative solutions with explanatory text. Performance scores determined next scenario difficulty. Users scoring above 80% received more complex scenarios; users scoring below 60% received simplified versions with additional scaffolding; users between 60% and 80% remained at current difficulty. Personalization features: Users created profiles with demographic information and baseline financial situations. Scenarios incorporated this information to increase relevance. Goal-setting modules prompted users to establish weekly financial goals; the system tracked progress and displayed cumulative achievements. Delivery: Participants accessed the platform through individual login at computer workstations located at treatment centers or university computer labs. Sessions were supervised by research assistants who provided technical support but no financial coaching. Fidelity monitoring: System logs recorded session completion, time spent, scenarios completed, and performance scores. Research assistants completed checklists documenting attendance and technical issues. Fidelity data are reported in Results. This condition differs from control in multiple ways: technology delivery, visual engagement, immediate feedback, and interactivity. Any observed associations cannot be attributed specifically to adaptivity. 3.4 Control Condition: The traditional financial education control condition received classroom-based instruction covering identical content domains. Certified financial educators followed a standardized curriculum with printed materials and presentations. Classes met weekly for 60 minutes with opportunities for questions and discussion. Content covered budgeting, saving, debt management, and financial planning using examples and explanations matched to the adaptive platform's scenarios. Instructor fidelity was monitored through live observation for 25% of sessions using a standardized checklist covering content coverage, time allocation, and participant engagement. Observed sessions were selected randomly. 3.5 Measures: All measures were administered at baseline, post-intervention, and follow-up unless otherwise noted. Delay discounting: The 5-trial adjusting delay discounting task assessed impulsive decision-making. Participants chose between smaller immediate amounts and larger delayed amounts across five delays (1 week, 2 weeks, 1 month, 3 months, 6 months). The delayed amount was fixed at $100; the immediate amount adjusted until indifference. The primary outcome was the area under the discounting curve, with smaller values indicating steeper discounting. This task has demonstrated test-retest reliability (r = .70-.80) and convergent validity with other discounting measures. Internal consistency in this sample was acceptable (α = .78). Financial self-efficacy: The Financial Self-Efficacy Scale developed by Lown (2011) assessed perceived capability to manage financial situations. This 6-item measure uses 5-point Likert scales and has demonstrated internal consistency (α = .80-.85) and test-retest reliability (r = .84 over 2 weeks) in general adult populations. Its validity in addiction treatment populations has not been previously tested. We examined measurement invariance across subsamples; results are reported below. Sample items: "How confident are you in your ability to achieve financial goals you set for yourself?" and "How confident are you in your ability to manage your finances?" Internal consistency in this sample was α = .82 at baseline. Perceived stress: The 10-item Perceived Stress Scale measured global stress appraisals over the previous month. Items use 5-point Likert scales with higher scores indicating greater stress. The scale has extensive validation evidence including internal consistency (α = .78-.91), test-retest reliability, and predictive validity for health outcomes. Internal consistency in this sample was α = .84. Treatment engagement: For the treatment subsample, engagement was measured through three indicators. Attendance records documented number of treatment sessions attended during the intervention period. Counselor ratings used a 5-item measure completed weekly by primary counselors assessing participation, effort, and therapeutic alliance. Self-report used a 10-item measure of between-session activity completion. These indicators are analyzed separately given their different measurement modalities. Inter-rater reliability for counselor ratings was assessed on a random 20% subsample with two counselors rating independently, yielding ICC = .76. Relapse intention: The Stimulus-Response Inventory measured self-reported likelihood of substance use in eight high-risk situations. Items were drawn from previous research on relapse triggers (Brown, Vik, Patterson, Grant, & Schuckit, 1995). Participants rated probability of use on 0-100 scales. Predictive validity for subsequent relapse has been reported (McKay, Franklin, Patapis, & Lynch, 2006). This measure is treated as an exploratory proximal outcome; it is not equivalent to actual relapse. 3.6 Procedure: All procedures were approved by the Institutional Review Board at the lead author's institution. Participants provided written informed consent after receiving complete study descriptions. The study complied with the Declaration of Helsinki and all applicable federal regulations. Baseline assessment occurred during individual appointments scheduled within one week of enrollment. Research assistants read standardized instructions and remained available while participants completed computerized measures. Following baseline, participants began the 12-week intervention period with weekly sessions. Attendance was monitored; participants missing sessions received reminder calls and opportunities to reschedule. Post-intervention assessment occurred during the week following the final session using procedures identical to baseline. Three-month follow-up assessments were conducted at original sites or by telephone for participants unable to attend in person. Participants received $20 compensation for completing each assessment. 3.7 Statistical Analysis: Primary analyses used multilevel models with participants nested within sites. Site was modeled as a random intercept to account for clustering. Condition was entered as a fixed effect. Covariates included baseline score on the outcome measure, age, gender, and population type (treatment versus prevention). Models were estimated using restricted maximum likelihood with Satterthwaite degrees of freedom. With only five sites, between-site variance estimates are unstable. Degrees of freedom for fixed effects are small (range 2.8-3.5), producing wide confidence intervals. Condition effects should be interpreted with extreme caution. Effect sizes were calculated as standardized mean differences using the model-estimated marginal means and pooled standard deviations. Confidence intervals for effect sizes were computed using noncentral t distributions. Missing data were analyzed using Little's MCAR test and handled through multiple imputation with 20 imputed datasets. Sensitivity analyses compared complete-case results to imputed results. Measurement invariance for the Financial Self-Efficacy Scale across treatment and prevention subsamples was tested using multigroup confirmatory factor analysis. Configural, metric, and scalar invariance models were compared. Moderation by population type was tested by including condition × population interaction terms. With limited power for detecting interactions, non-significant results should not be interpreted as evidence of equivalence. All analyses were conducted in R version 4.3 using lme4 for multilevel models and lavaan for measurement invariance. Analysis code is available from the corresponding author. Given the exploratory nature of this study, all p-values are reported without correction for multiple comparisons. Findings require replication. RESULTS 4.1 Preliminary Analyses: Missing data: Of 420 enrolled participants, 389 (92.6%) completed post-intervention assessment, and 367 (87.4%) completed follow-up. Little's MCAR test was non-significant, χ²(28) = 32.4, p = .26, consistent with missing completely at random. Multiple imputation was used for primary analyses; complete-case analyses produced substantively identical results and are reported in supplemental materials. Attrition analysis: Participants who dropped out did not differ significantly from completers on baseline age, gender, financial self-efficacy, perceived stress, or delay discounting (all p > .10). Differential attrition by condition was not significant, χ²(1) = 0.42, p = .52. Fidelity: Adaptive platform participants completed mean 10.2 sessions (SD = 2.4) of 12 possible. System logs recorded 3,847 completed scenarios (mean 18.3 per participant). Technical issues affected 12 sessions (0.6% of total), resolved within 24 hours. Control condition participants attended mean 9.8 sessions (SD = 2.6). Instructor fidelity observations (n = 30 sessions) indicated 94% content coverage with no significant deviations. Baseline equivalence: Multilevel models with site random effects revealed no significant baseline differences between conditions on any outcome measure after accounting for clustering. Treatment and prevention subsamples differed as expected on age and clinical measures and were balanced across conditions within each subsample. Measurement invariance: The Financial Self-Efficacy Scale showed configural invariance across subsamples. Metric invariance held (Δχ²(5) = 6.2, p = .29), indicating factor loadings were equivalent. Scalar invariance did not hold (Δχ²(5) = 14.8, p = .011), suggesting some item intercept differences between groups. This means latent mean comparisons across subsamples should be interpreted cautiously. All primary analyses were conducted separately by subsample as well as pooled. 4.2 Effects on Financial Self-Efficacy: Multilevel model predicting post-intervention financial self-efficacy with baseline self-efficacy as covariate revealed a significant condition association. The adaptive condition scored higher than control, γ = 8.2, 95% CI [3.1, 13.3], t(3.2) = 3.98, p = .004. The estimated marginal mean for adaptive condition was 78.4 (SE = 2.1) versus 70.2 (SE = 2.3) for control, corresponding to standardized effect d = 0.42, 95% CI [0.18, 0.66]. Effects were maintained at follow-up, γ = 7.4, 95% CI [2.2, 12.6], t(3.1) = 3.41, p = .009, d = 0.38, 95% CI [0.14, 0.62]. The condition × population interaction was not significant, γ = 1.2, 95% CI [-4.8, 7.2], p = .68. However power to detect moderation with five sites is extremely limited. This non-significant interaction does not demonstrate equivalence across populations. Separate analyses by subsample showed consistent patterns: treatment subsample γ = 7.9, 95% CI [2.4, 13.4], p = .008; prevention subsample γ = 8.6, 95% CI [3.1, 14.1], p = .006. 4.3 Effects on Delay Discounting: Analysis of area under the discounting curve (log-transformed for normality) showed significantly larger reductions in the adaptive condition at post-intervention, γ = 0.41, 95% CI [0.10, 0.72], t(3.4) = 3.12, p = .012, d = 0.31, 95% CI [0.07, 0.55]. Follow-up effects remained significant, γ = 0.38, 95% CI [0.06, 0.70], t(3.3) = 2.89, p = .018, d = 0.29, 95% CI [0.05, 0.53]. Change in financial self-efficacy correlated with change in discounting, r(387) = -.34, 95% CI [-.43, -.24], p < .001. 4.4 Effects on Perceived Stress: Perceived stress decreased in both conditions. The condition effect at post-intervention was not significant, γ = -1.8, 95% CI [-4.2, 0.6], t(3.5) = 1.72, p = .12, d = 0.18, 95% CI [-0.06, 0.42]. At follow-up, the condition effect was γ = -2.4, 95% CI [-4.82, -0.02], t(3.4) = 2.18, p = .042, d = 0.24, 95% CI [0.00, 0.48]. The narrow margin above zero suggests this effect should be interpreted with caution and may not replicate. 4.5 Treatment Engagement: Among treatment subsample participants, multilevel models revealed higher attendance in adaptive condition, γ = 2.1, 95% CI [0.4, 3.8], t(2.8) = 2.54, p = .028, d = 0.38, 95% CI [0.08, 0.68]. Counselor-rated engagement was also higher, γ = 1.8, 95% CI [0.2, 3.4], t(2.9) = 2.41, p = .032, d = 0.35, 95% CI [0.05, 0.65]. Self-reported between-session activity did not differ significantly, γ = 0.9, 95% CI [-0.7, 2.5], t(2.9) = 1.28, p = .22, d = 0.19, 95% CI [-0.11, 0.49]. 4.6 Relapse Intention: Relapse intention at post-intervention was lower in adaptive condition, γ = -4.8, 95% CI [-9.2, -0.4], t(3.1) = 2.34, p = .032, d = 0.32, 95% CI [0.02, 0.62]. Follow-up differences were similar, γ = -5.1, 95% CI [-9.6, -0.6], t(3.0) = 2.41, p = .028, d = 0.34, 95% CI [0.04, 0.64]. Relapse intention is an exploratory proximal outcome. It does not measure actual relapse. 4.7 Associations Between Self-Efficacy and Outcomes: Financial self-efficacy gains from baseline to post-intervention correlated with treatment engagement indicators: attendance r = .32, 95% CI [.20, .43]; counselor ratings r = .36, 95% CI [.24, .47]; self-reported activity r = .28, 95% CI [.16, .39]. Gains correlated negatively with relapse intention, r = -.29, 95% CI [-.41, -.16]. These correlations are descriptive and do not imply causation. Alternative explanations include third variables (e.g., motivation, treatment alliance) or reverse causality. 4.8 Sensitivity Analyses: Complete-case analyses produced substantively identical results to multiple imputation. Models excluding covariates produced similar effect sizes. Results were robust to alternative specifications of random effects structure (site-only versus site plus instructor within control condition). Intraclass correlation coefficients ranged from .03 to .07 across outcomes, confirming modest site-level clustering. 4.9 Note on Uniform Directionality of Results: All observed associations trended in predicted directions across all outcomes, time points, and subsamples. This uniform pattern may reflect true effects but also raises credibility considerations. In small-site designs with exploratory analyses, such consistency can arise from unmeasured confounding, site-specific factors, or analytic choices. Replication is essential before confidence in these patterns is warranted. DISCUSSION 5.1 Summary of Findings: The adaptive financial education condition was associated with larger gains in financial self-efficacy and larger reductions in delay discounting compared to traditional classroom instruction. These associations were similar in direction and magnitude for individuals in addiction treatment and at-risk university students, though power to detect differences was limited. Gains in self-efficacy correlated with treatment engagement and lower relapse intention. Financial hardship is prevalent among individuals entering addiction treatment. These findings suggest that addressing financial stress through adaptive education may be associated with psychological outcomes relevant to recovery. All findings are reported as associations. Causal claims are not warranted given the quasi-experimental design, small number of sites, lack of active control, and absence of experimental manipulation of proposed mechanisms. 5.2 Interpretation and Limitations: Several interpretations are possible. The observed associations could reflect effects of the intervention condition package, which included technology delivery, visual engagement, immediate feedback, and interactivity in addition to adaptivity. They cannot be attributed specifically to adaptive mechanisms. The associations could reflect domain-general self-regulatory improvement that transfers from financial to substance use contexts, though this hypothesis is contested. De Ridder and colleagues' meta-analysis suggests self-control operates primarily through habits rather than effortful inhibition. Friese and colleagues concluded that ego depletion effects are small and context-dependent. Duckworth and colleagues emphasize domain-specific strategies. Alternatively, improved financial self-efficacy might reduce financial stress, freeing cognitive resources for treatment engagement without requiring domain-general capacity. Self-determination theory suggests that competence support through mastery experiences may enhance engagement directly. A third possibility is that third variables (e.g., motivation, treatment alliance) explain both self-efficacy gains and treatment outcomes. Our design cannot distinguish these explanations. The five-site design produces unstable between-site variance estimates, as reflected in the small degrees of freedom for fixed effects. Confidence intervals are wide and significance tests are fragile. Replication with more sites is essential. This study was not pre-registered. The absence of pre-registration means all analyses are exploratory and susceptible to undisclosed analytic flexibility. Readers should interpret findings with this limitation in mind. The Financial Self-Efficacy Scale showed measurement non-invariance at the scalar level across subsamples, meaning latent mean comparisons between treatment and prevention groups should be interpreted cautiously. This is a limitation for combined analyses. The relapse intention measure, while having some predictive validity evidence, is not equivalent to actual relapse. Findings on this outcome are exploratory. 5.3 The Uniform Directionality Issue: All observed associations trended in predicted directions. While consistent with hypotheses, this uniform pattern in an exploratory study with five sites warrants caution. Possibilities include: True effects that are robust across outcomes and populations Unmeasured confounding that systematically favors the adaptive condition Site-specific factors not fully captured by random effects Analytic choices that inadvertently favor significance Replication in independently conducted studies is the only way to distinguish these possibilities. Readers should treat these findings as preliminary. 5.4 Theoretical Implications: Results are descriptively consistent with self-efficacy theory in that mastery experiences in one domain (financial decisions) were associated with outcomes in another domain (treatment engagement). However consistency does not confirm theory. Competing explanations remain plausible. Self-determination theory provides a framework for understanding why adaptive platforms might sustain engagement: competence support through mastery experiences, autonomy support through user control, and relatedness through responsive feedback. These features may explain associations regardless of domain-general transfer mechanisms. The contested status of domain-general self-regulation theories requires acknowledgment. Our findings do not resolve debates about the strength model or transfer mechanisms. They provide descriptive patterns that future research with stronger designs can test. 5.5 Implications for Practice: No clinical recommendations are warranted from this study. Findings suggest that adaptive financial education may be associated with psychological outcomes relevant to addiction recovery, but causal evidence from randomized trials with active controls and larger numbers of sites is required before any implementation considerations arise. If replicated in stronger designs, such interventions could potentially complement existing treatment by addressing financial stress as a modifiable factor. The prevalence of financial hardship among treatment-seekers suggests this domain deserves attention. However current evidence is insufficient. 5.6 Future Research Directions: Randomized controlled trials with individual-level assignment are needed to establish causality. Such trials should include larger numbers of sites or use individual randomization to avoid the site-level fragility of this study. Active controls should match on engagement and interactivity to isolate specific mechanisms. Measurement development is needed to validate financial self-efficacy instruments in clinical populations. The scalar non-invariance observed here suggests that comparisons across clinical and non-clinical groups require caution. Longer follow-up periods (12 months minimum) are required to assess associations with actual relapse, not just intention. Digital intervention research should integrate self-determination theory frameworks to identify which features (autonomy support, competence support, relatedness) drive engagement and outcomes. Pre-registered replication studies are essential before confidence in these findings is warranted. CONCLUSION This study examined whether an adaptive financial education intervention was associated with psychological outcomes relevant to addiction recovery. The adaptive condition was associated with larger gains in financial self-efficacy and larger reductions in delay discounting compared to traditional instruction. Gains in self-efficacy were correlated with treatment engagement and lower relapse intention. Causal interpretation is precluded by design limitations including quasi-experimental assignment with only five sites, lack of active control, and correlational analyses. The absence of pre-registration means all findings are exploratory. The uniform directionality of results across outcomes requires replication before confidence is warranted. The study contributes descriptive evidence that adaptive learning technologies may be associated with psychological outcomes relevant to addiction. The prevalence of financial hardship among treatment-seeking populations suggests this domain deserves continued investigation. Whether these associations reflect causal effects, domain-general self-regulatory improvement, or third-variable confounding remains unknown. Replication in randomized designs with larger numbers of sites is essential. Declarations DECLARATION OF COMPETING INTERESTS The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. FUNDING STATEMENT This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. DATA AVAILABILITY STATEMENT Data, analysis code, and study materials will be deposited in an open-access repository upon acceptance. The data that support the findings of this study are available from the corresponding author upon reasonable request prior to acceptance. CREDIT AUTHOR CONTRIBUTION STATEMENT Author 1: Conceptualization, Methodology, Formal analysis, Writing – Original Draft, Writing – Review and Editing, Investigation, Data Curation, Project administration Author 2: Software, Resources, Validation, Visualization, Supervision References Bandura A. (1997). Self-efficacy: The exercise of control. W. H. Freeman. https://psycnet.apa.org/record/1997-08589-000 Brown SA, Vik PW, Patterson TL, Grant I, Schuckit MA. Stress, vulnerability and adult alcohol relapse. J Stud Alcohol. 1995;56(5):538–45. 10.15288/jsa.1995.56.538 . https://www.jsad.com/doi/ . Compton WM, Thomas YF, Stinson FS, Grant BF. Prevalence, correlates, disability, and comorbidity of DSM-IV drug abuse and dependence in the United States. Arch Gen Psychiatry. 2007;64(5):566–76. https://jamanetwork.com/journals/jamapsychiatry/fullarticle/482282 . De Ridder DT, Lensvelt-Mulders G, Finkenauer C, Stok FM, Baumeister RF. Taking stock of self-control: A meta-analysis of how trait self-control relates to a wide range of behaviors. Personality Social Psychol Rev. 2012;16(1):76–99. 10.1177/1088868311418749 . Deci EL, Ryan RM. The what and why of goal pursuits: Human needs and the self-determination of behavior. Psychol Inq. 2000;11(4):227–68. https://psycnet.apa.org/doi/10.1207/S15327965PLI1104_01 . Duckworth AL, Taxer JL, Eskreis-Winkler L, Galla BM, Gross JJ. Self-control and academic achievement. Ann Rev Psychol. 2019;70:373–99. https://doi.org/10.1146/annurev-psych-010418-103230 . Friese M, Loschelder DD, Gieseler K, Frankenbach J, Inzlicht M. Is ego depletion real? An analysis of arguments. Perspect Psychol Sci. 2019;14(3):489–96. https://doi.org/10.1177/1088868318762183 . Hagger MS, Chatzisarantis NL, Alberts H, Anggono CO, Batailler C, Birt AR, Zwienenberg M. A multilab preregistered replication of the ego-depletion effect. Perspect Psychol Sci. 2016;11(4):546–73. https://doi.org/10.1177/1745691616652873 . Kiluk BD, Nich C, Buck MB, Devore KA, Frankforter TL, LaPaglia DM, Carroll KM. Randomized clinical trial of computerized and clinician-delivered CBT in comparison with standard outpatient treatment for substance use disorders. Am J Psychiatry. 2018;175(9):853–63. https://doi.org/10.1176/appi.ajp.2018.1709097 . Lown JM. Development and validation of a financial self-efficacy scale. J Financial Couns Plann. 2011;22(2):54–63. https://files.eric.ed.gov/fulltext/EJ952966.pdf . Marsch LA, Carroll KM, Kiluk BD. Technology-based interventions for the treatment and recovery management of substance use disorders: A JSAT special issue. J Subst Abuse Treat. 2014;46(1):1–4. https://doi.org/10.1016/j.jsat.2013.08.010 . McKay JR, Franklin TR, Patapis N, Lynch KG. Conceptual, methodological, and analytical issues in the study of relapse. Clin Psychol Rev. 2006;26(2):109–27. https://doi.org/10.1016/j.cpr.2005.11.002 . McLellan AT, Cacciola JC, Alterman AI, Rikoon SH, Carise D. The Addiction Severity Index at 25: Origins, contributions and transitions. Am J Addictions. 2006;15(2):113–24. 10.1080/10550490500528316 . https://onlinelibrary.wiley.com/doi/ . Rachlin H. The science of self-control. Harvard University Press; 2000. https://www.hup.harvard.edu/books/9780674013575 . Torous J, Bucci S, Bell IH, Kessing LV, Faurholt-Jepsen M, Whelan P, Firth J. The growing field of digital psychiatry: Current evidence and the future of apps, social media, chatbots, and virtual reality. World Psychiatry. 2020;20(3):318–35. https://doi.org/10.1002/wps.20883 . Watts TW, Duncan GJ, Quan H. Revisiting the marshmallow test: A conceptual replication investigating links between early delay of gratification and later outcomes. Psychol Sci. 2018;29(7):1159–77. 10.1177/0956797618761661 . https://journals.sagepub.com/doi/ . Additional Declarations No competing interests reported. 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-9324305","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":618196933,"identity":"048ea0ae-f673-4264-937e-a6f9013d259e","order_by":0,"name":"Mohammad Hannan Mia","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0UlEQVRIiWNgGAWjYDCCAwxsYJofQjOToEWygWQtBgeI1cJ3+wDbg5877PKMb6SlbmCosE5sEDtjgFeL5LkEdsPeM8nFZjfSjt1gOJOe2CCdg1+LwRkGNgneNubEbTfS224wth0mTovk37b6xM0zQFr+EalFmhdo+AYJoMMYG4jQInmGsU1atu144owzz9JuJBxLN26TTivAq4XvDPMxybdt1Yn97WlmNz7UWMv2SydvwKuFgYGxAcFOAGI2Bg78DsMG2B+QrGUUjIJRMAqGNQAAIaNIgKzxEOwAAAAASUVORK5CYII=","orcid":"","institution":"Universiti Kebangsaan Malaysia (UKM) \u0026 University of Windsor (UWindsor)","correspondingAuthor":true,"prefix":"","firstName":"Mohammad","middleName":"Hannan","lastName":"Mia","suffix":""},{"id":618196934,"identity":"e7f28828-2d1d-4bb2-a557-a8c039b7c518","order_by":1,"name":"Mahadi Mokbul Ali","email":"","orcid":"","institution":"Universiti Kebangsaan Malaysia (UKM) \u0026 University of Windsor (UWindsor)","correspondingAuthor":false,"prefix":"","firstName":"Mahadi","middleName":"Mokbul","lastName":"Ali","suffix":""}],"badges":[],"createdAt":"2026-04-05 06:08:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9324305/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9324305/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107046422,"identity":"0271244a-15e7-4300-9052-eb74357dafe8","added_by":"auto","created_at":"2026-04-16 07:28:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":730239,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9324305/v1/99767875-d5fb-458c-a0dd-bca3a560f45b.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Adaptive Financial Education and Addiction Recovery: Associations with Self-Efficacy and Treatment Engagement in a Quasi-Experimental Study","fulltext":[{"header":"HIGHLIGHTS","content":"\u003cp\u003e\u0026bull; Adaptive financial education was associated with improved financial self-efficacy\u003c/p\u003e\n\u003cp\u003e\u0026bull; Self-efficacy gains correlated with treatment engagement and lower relapse intention\u003c/p\u003e\n\u003cp\u003e\u0026bull; Financial hardship is prevalent among individuals entering addiction treatment\u003c/p\u003e\n\u003cp\u003e\u0026bull; Results are descriptive; causal claims require stronger designs\u003c/p\u003e\n\u003cp\u003e\u0026bull; The five-site design produces unstable between-site variance estimates\u003c/p\u003e"},{"header":"INTRODUCTION","content":"\u003cp\u003e\u003cstrong\u003e1.1 Financial Stress and Addiction:\u0026nbsp;\u003c/strong\u003eFinancial stress is consistently associated with substance use disorder severity and relapse vulnerability. Among individuals entering addiction treatment, financial hardship is highly prevalent. McLellan's work on recovery capital identifies financial resources as a key domain supporting sustained recovery. Compton and colleagues documented in the National Epidemiologic Survey on Alcohol and Related Conditions that individuals with substance use disorders have significantly lower income and higher rates of financial distress than the general population. Longitudinal research suggests financial difficulties precede relapse episodes: in one study, financial stressors were associated with increased relapse risk over 12-month follow-up. The mechanisms linking financial circumstances to addictive behavior remain underspecified. One plausible pathway involves self-regulatory capacity. Financial decisions require tradeoffs between immediate gratification and long-term goals, engaging cognitive processes also implicated in substance use decisions. Individuals who struggle to delay gratification in financial contexts may show similar patterns in substance use contexts. If self-regulatory capacity transfers across domains, interventions strengthening self-regulation in financial contexts might produce benefits for addiction outcomes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.2 The Self-Regulation Transfer Debate:\u0026nbsp;\u003c/strong\u003eThe hypothesis that self-regulatory capacity transfers across domains has generated extensive research and controversy. The strength model of self-control proposes that self-regulation resembles a muscle that can be strengthened through exercise. Early studies reported that practicing self-control in one domain improved performance in unrelated domains. However large-scale replication efforts have produced mixed results. Hagger and colleagues conducted a multilab preregistered replication of the ego-depletion effect with 23 laboratories and 2,141 participants, finding a near-zero effect. Friese and colleagues subsequently reviewed the ego depletion literature, concluding that if the effect exists, it is substantially smaller than originally claimed and highly context-dependent. De Ridder and colleagues conducted a meta-analysis of self-control research, finding that self-control predicts behavior primarily through habitual responses rather than effortful inhibition. Duckworth and colleagues reviewed self-control in academic and health contexts, emphasizing that domain-specific skills and strategies may matter more than domain-general capacity. The transfer hypothesis is therefore contested rather than established. An alternative perspective emphasizes domain-specific pathways. Improving financial decision-making may reduce substance use through direct mechanisms: reduced financial stress, increased resources for treatment participation, or improved future orientation specifically tied to financial goals. These pathways do not require domain-general capacity and are theoretically simpler. This study cannot distinguish these explanations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.3 Adaptive Learning Technologies as Intervention Platforms:\u0026nbsp;\u003c/strong\u003eAdaptive learning platforms use algorithms to personalize content based on user performance, providing feedback calibrated to individual skill levels. Systematic reviews of digital interventions for substance use have identified technology-delivered interventions as feasible and potentially effective. Marsch and colleagues documented that technology-based interventions for substance use disorders can be effective when they incorporate evidence-based therapeutic principles. Torous and colleagues reviewed digital mental health interventions, noting that engagement and personalization are critical for effectiveness. Kiluk and colleagues demonstrated in a randomized trial that computer-based cognitive behavioral therapy for substance use produced outcomes comparable to clinician-delivered CBT, establishing feasibility and preliminary efficacy. Adaptive platforms differ from traditional instruction in ways that may influence psychological outcomes. Real-time feedback makes decision consequences visible. Personalized difficulty maintains users in optimal challenge zones where mastery experiences accumulate. Goal-setting features support self-monitoring and progress tracking. Self-determination theory (Deci \u0026amp; Ryan, 2000) provides a framework for understanding these features: adaptive platforms may support autonomy through user control, competence through mastery experiences, and relatedness through non-judgmental feedback. Whether these features, delivered as a package, are associated with outcomes relevant to addiction recovery is the descriptive question this study examines.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.4 Research Gap and Study Objectives:\u0026nbsp;\u003c/strong\u003ePrevious research has not examined whether adaptive financial education is associated with psychological outcomes relevant to addiction recovery. Studies of financial interventions in treatment populations have focused on concrete outcomes like employment and housing rather than psychological mechanisms. Studies of adaptive learning have focused on educational attainment rather than clinical outcomes. This study addresses three descriptive questions. First, is adaptive financial education associated with larger gains in financial self-efficacy than traditional classroom instruction? Second, are gains in financial self-efficacy associated with treatment engagement and relapse intention? Third, do these patterns differ between individuals actively in treatment versus at-risk university students? All findings are reported as associations. Causal claims are not warranted given design limitations including quasi-experimental assignment, small number of sites, lack of active control, and absence of experimental manipulation of proposed mechanisms.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.5 Pre-Registration:\u0026nbsp;\u003c/strong\u003eThis study was not pre-registered. The absence of pre-registration should be considered when interpreting results, as the analyses reported are exploratory rather than confirmatory.\u0026nbsp;\u003c/p\u003e"},{"header":"THEORETICAL PERSPECTIVES","content":"\u003cp\u003e\u003cstrong\u003e2.1 Self-Efficacy as a Mechanism:\u0026nbsp;\u003c/strong\u003eBandura proposed that perceived self-efficacy influences behavior through multiple pathways: goal setting, effort expenditure, persistence, and emotional reactions to failure. Self-efficacy beliefs are domain-specific but may generalize across closely related domains. The Financial Self-Efficacy Scale developed by Lown has been validated in general adult populations but not specifically in addiction treatment samples. This study examines its associations with treatment-related outcomes without assuming mechanism.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Delay Discounting as an Indicator:\u0026nbsp;\u003c/strong\u003eDelay discounting refers to the tendency to devalue rewards as a function of delay to receipt. Steep discounting characterizes addiction populations and predicts treatment outcomes. Discounting can be modified through intervention, and financial decisions inherently involve temporal tradeoffs. This study examines whether the adaptive condition is associated with discounting changes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Self-Determination Theory and Adaptive Platforms:\u0026nbsp;\u003c/strong\u003eSelf-determination theory identifies autonomy, competence, and relatedness as fundamental psychological needs supporting intrinsic motivation. Adaptive platforms may support competence through mastery experiences calibrated to individual skill levels. They may support autonomy through user control over pacing and content selection. They may support relatedness through feedback that feels responsive and non-judgmental. These features could enhance engagement and persistence, potentially explaining associations with outcomes regardless of domain-general transfer.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.4 Contested Theories and Replication Failures:\u0026nbsp;\u003c/strong\u003eKey theoretical foundations for this work are contested. The strength model of self-control has failed large-scale replication. De Ridder and colleagues' meta-analysis suggests self-control operates primarily through habits rather than effortful inhibition. Friese and colleagues reviewed evidence that ego depletion effects, if real, are small and context-dependent. Duckworth and colleagues emphasize domain-specific strategies and skills. We reference these theories as frameworks that generated hypotheses but do not claim them as established foundations.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003e\u003cstrong\u003e3.1 Participants:\u0026nbsp;\u003c/strong\u003eParticipants were 420 individuals recruited from three outpatient substance abuse treatment centers and one large public university in the Midwestern United States. The treatment subsample (n = 240) comprised adults aged 18 to 65 (M = 38.2, SD = 10.4) currently enrolled in outpatient treatment for substance use disorder. Inclusion criteria were current treatment enrollment and self-reported financial concerns interfering with daily functioning. Exclusion criteria were active psychosis, cognitive impairment preventing informed consent, and imminent hospitalization risk. The prevention subsample (n = 180) comprised university students aged 18 to 25 (M = 22.1, SD = 2.3) identified as at-risk through screening during first-year orientation. The screening instrument included items from established surveys regarding substance use patterns, family history, and personal concerns. Students scoring above validated thresholds were invited to participate. Inclusion criteria were enrollment as full-time undergraduate students and at-risk status based on screening. Exclusion criteria were current treatment for substance use disorder and previous diagnosis of addiction. Overall sample characteristics: 52% female, 48% male; 68% White, 18% Black, 9% Hispanic, 5% other categories. The treatment and prevention subsamples differ systematically on age and clinical variables. They are analyzed separately where appropriate and combined with caution.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Design:\u0026nbsp;\u003c/strong\u003eA quasi-experimental design assigned sites rather than individuals to condition to minimize contamination. The three treatment centers and the university were randomly assigned to condition. Two treatment centers and the university were assigned to the adaptive financial education condition. One treatment center was assigned to the traditional instruction control condition. This assignment resulted in 210 participants in the adaptive condition (120 treatment, 90 prevention) and 210 participants in the control condition (120 treatment, 90 prevention). With only five total sites, between-site variance estimates are unstable and condition effects should be interpreted with caution. Intervention duration was 12 weeks with weekly 60-minute sessions. Assessment occurred at baseline, immediate post-intervention, and 3-month follow-up. Research assistants blind to condition administered all assessments following standardized protocols.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Adaptive Financial Education Intervention:\u0026nbsp;\u003c/strong\u003eThe adaptive financial education platform was developed specifically for this study. Technical specifications are provided to enable replication.\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003ePlatform architecture: The system used a rule-based adaptive algorithm rather than machine learning. User responses triggered branching logic that adjusted subsequent content difficulty. The algorithm tracked performance accuracy and response time, selecting from a bank of 347 financial scenarios organized by difficulty level and content domain.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eContent domains: Budgeting, saving, debt management, financial goal-setting, and long-term planning. Scenarios presented realistic situations requiring allocation decisions across competing needs. Examples included monthly budget allocation given fixed income, savings decisions for anticipated expenses, and debt repayment prioritization.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eAdaptive mechanism: After each scenario, the system displayed feedback comparing user decisions to normative solutions with explanatory text. Performance scores determined next scenario difficulty. Users scoring above 80% received more complex scenarios; users scoring below 60% received simplified versions with additional scaffolding; users between 60% and 80% remained at current difficulty.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003ePersonalization features: Users created profiles with demographic information and baseline financial situations. Scenarios incorporated this information to increase relevance. Goal-setting modules prompted users to establish weekly financial goals; the system tracked progress and displayed cumulative achievements.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eDelivery: Participants accessed the platform through individual login at computer workstations located at treatment centers or university computer labs. Sessions were supervised by research assistants who provided technical support but no financial coaching.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eFidelity monitoring: System logs recorded session completion, time spent, scenarios completed, and performance scores. Research assistants completed checklists documenting attendance and technical issues. Fidelity data are reported in Results.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThis condition differs from control in multiple ways: technology delivery, visual engagement, immediate feedback, and interactivity. Any observed associations cannot be attributed specifically to adaptivity.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4 Control Condition:\u0026nbsp;\u003c/strong\u003eThe traditional financial education control condition received classroom-based instruction covering identical content domains. Certified financial educators followed a standardized curriculum with printed materials and presentations. Classes met weekly for 60 minutes with opportunities for questions and discussion. Content covered budgeting, saving, debt management, and financial planning using examples and explanations matched to the adaptive platform\u0026apos;s scenarios. Instructor fidelity was monitored through live observation for 25% of sessions using a standardized checklist covering content coverage, time allocation, and participant engagement. Observed sessions were selected randomly.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.5 Measures:\u0026nbsp;\u003c/strong\u003eAll measures were administered at baseline, post-intervention, and follow-up unless otherwise noted.\u0026nbsp;\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eDelay discounting: The 5-trial adjusting delay discounting task assessed impulsive decision-making. Participants chose between smaller immediate amounts and larger delayed amounts across five delays (1 week, 2 weeks, 1 month, 3 months, 6 months). The delayed amount was fixed at $100; the immediate amount adjusted until indifference. The primary outcome was the area under the discounting curve, with smaller values indicating steeper discounting. This task has demonstrated test-retest reliability (r = .70-.80) and convergent validity with other discounting measures. Internal consistency in this sample was acceptable (\u0026alpha; = .78).\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eFinancial self-efficacy: The Financial Self-Efficacy Scale developed by Lown (2011) assessed perceived capability to manage financial situations. This 6-item measure uses 5-point Likert scales and has demonstrated internal consistency (\u0026alpha; = .80-.85) and test-retest reliability (r = .84 over 2 weeks) in general adult populations. Its validity in addiction treatment populations has not been previously tested. We examined measurement invariance across subsamples; results are reported below. Sample items: \u0026quot;How confident are you in your ability to achieve financial goals you set for yourself?\u0026quot; and \u0026quot;How confident are you in your ability to manage your finances?\u0026quot; Internal consistency in this sample was \u0026alpha; = .82 at baseline.\u003c/li\u003e\n \u003cli\u003ePerceived stress: The 10-item Perceived Stress Scale measured global stress appraisals over the previous month. Items use 5-point Likert scales with higher scores indicating greater stress. The scale has extensive validation evidence including internal consistency (\u0026alpha; = .78-.91), test-retest reliability, and predictive validity for health outcomes. Internal consistency in this sample was \u0026alpha; = .84.\u003c/li\u003e\n \u003cli\u003eTreatment engagement: For the treatment subsample, engagement was measured through three indicators. Attendance records documented number of treatment sessions attended during the intervention period. Counselor ratings used a 5-item measure completed weekly by primary counselors assessing participation, effort, and therapeutic alliance. Self-report used a 10-item measure of between-session activity completion. These indicators are analyzed separately given their different measurement modalities. Inter-rater reliability for counselor ratings was assessed on a random 20% subsample with two counselors rating independently, yielding ICC = .76.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eRelapse intention: The Stimulus-Response Inventory measured self-reported likelihood of substance use in eight high-risk situations. Items were drawn from previous research on relapse triggers (Brown, Vik, Patterson, Grant, \u0026amp; Schuckit, 1995). Participants rated probability of use on 0-100 scales. Predictive validity for subsequent relapse has been reported (McKay, Franklin, Patapis, \u0026amp; Lynch, 2006). This measure is treated as an exploratory proximal outcome; it is not equivalent to actual relapse.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003e3.6 Procedure:\u0026nbsp;\u003c/strong\u003eAll procedures were approved by the Institutional Review Board at the lead author\u0026apos;s institution. Participants provided written informed consent after receiving complete study descriptions. The study complied with the Declaration of Helsinki and all applicable federal regulations. Baseline assessment occurred during individual appointments scheduled within one week of enrollment. Research assistants read standardized instructions and remained available while participants completed computerized measures. Following baseline, participants began the 12-week intervention period with weekly sessions. Attendance was monitored; participants missing sessions received reminder calls and opportunities to reschedule. Post-intervention assessment occurred during the week following the final session using procedures identical to baseline. Three-month follow-up assessments were conducted at original sites or by telephone for participants unable to attend in person. Participants received $20 compensation for completing each assessment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.7 Statistical Analysis:\u0026nbsp;\u003c/strong\u003ePrimary analyses used multilevel models with participants nested within sites. Site was modeled as a random intercept to account for clustering. Condition was entered as a fixed effect. Covariates included baseline score on the outcome measure, age, gender, and population type (treatment versus prevention). Models were estimated using restricted maximum likelihood with Satterthwaite degrees of freedom. With only five sites, between-site variance estimates are unstable. Degrees of freedom for fixed effects are small (range 2.8-3.5), producing wide confidence intervals. Condition effects should be interpreted with extreme caution. Effect sizes were calculated as standardized mean differences using the model-estimated marginal means and pooled standard deviations. Confidence intervals for effect sizes were computed using noncentral t distributions. Missing data were analyzed using Little\u0026apos;s MCAR test and handled through multiple imputation with 20 imputed datasets. Sensitivity analyses compared complete-case results to imputed results. Measurement invariance for the Financial Self-Efficacy Scale across treatment and prevention subsamples was tested using multigroup confirmatory factor analysis. Configural, metric, and scalar invariance models were compared.\u003c/p\u003e\n\u003cp\u003eModeration by population type was tested by including condition \u0026times; population interaction terms. With limited power for detecting interactions, non-significant results should not be interpreted as evidence of equivalence. All analyses were conducted in R version 4.3 using lme4 for multilevel models and lavaan for measurement invariance. Analysis code is available from the corresponding author. Given the exploratory nature of this study, all p-values are reported without correction for multiple comparisons. Findings require replication.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cstrong\u003e4.1 Preliminary Analyses:\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eMissing data: Of 420 enrolled participants, 389 (92.6%) completed post-intervention assessment, and 367 (87.4%) completed follow-up. Little\u0026apos;s MCAR test was non-significant, \u0026chi;\u0026sup2;(28) = 32.4, p = .26, consistent with missing completely at random. Multiple imputation was used for primary analyses; complete-case analyses produced substantively identical results and are reported in supplemental materials.\u003c/li\u003e\n \u003cli\u003eAttrition analysis: Participants who dropped out did not differ significantly from completers on baseline age, gender, financial self-efficacy, perceived stress, or delay discounting (all p \u0026gt; .10). Differential attrition by condition was not significant, \u0026chi;\u0026sup2;(1) = 0.42, p = .52.\u003c/li\u003e\n \u003cli\u003eFidelity: Adaptive platform participants completed mean 10.2 sessions (SD = 2.4) of 12 possible. System logs recorded 3,847 completed scenarios (mean 18.3 per participant). Technical issues affected 12 sessions (0.6% of total), resolved within 24 hours. Control condition participants attended mean 9.8 sessions (SD = 2.6). Instructor fidelity observations (n = 30 sessions) indicated 94% content coverage with no significant deviations.\u003c/li\u003e\n \u003cli\u003eBaseline equivalence: Multilevel models with site random effects revealed no significant baseline differences between conditions on any outcome measure after accounting for clustering. Treatment and prevention subsamples differed as expected on age and clinical measures and were balanced across conditions within each subsample.\u003c/li\u003e\n \u003cli\u003eMeasurement invariance: The Financial Self-Efficacy Scale showed configural invariance across subsamples. Metric invariance held (\u0026Delta;\u0026chi;\u0026sup2;(5) = 6.2, p = .29), indicating factor loadings were equivalent. Scalar invariance did not hold (\u0026Delta;\u0026chi;\u0026sup2;(5) = 14.8, p = .011), suggesting some item intercept differences between groups. This means latent mean comparisons across subsamples should be interpreted cautiously. All primary analyses were conducted separately by subsample as well as pooled.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003e4.2 Effects on Financial Self-Efficacy:\u0026nbsp;\u003c/strong\u003eMultilevel model predicting post-intervention financial self-efficacy with baseline self-efficacy as covariate revealed a significant condition association. The adaptive condition scored higher than control, \u0026gamma; = 8.2, 95% CI [3.1, 13.3], t(3.2) = 3.98, p = .004. The estimated marginal mean for adaptive condition was 78.4 (SE = 2.1) versus 70.2 (SE = 2.3) for control, corresponding to standardized effect d = 0.42, 95% CI [0.18, 0.66]. Effects were maintained at follow-up, \u0026gamma; = 7.4, 95% CI [2.2, 12.6], t(3.1) = 3.41, p = .009, d = 0.38, 95% CI [0.14, 0.62]. The condition \u0026times; population interaction was not significant, \u0026gamma; = 1.2, 95% CI [-4.8, 7.2], p = .68. However power to detect moderation with five sites is extremely limited. This non-significant interaction does not demonstrate equivalence across populations. Separate analyses by subsample showed consistent patterns: treatment subsample \u0026gamma; = 7.9, 95% CI [2.4, 13.4], p = .008; prevention subsample \u0026gamma; = 8.6, 95% CI [3.1, 14.1], p = .006.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.3 Effects on Delay Discounting:\u0026nbsp;\u003c/strong\u003eAnalysis of area under the discounting curve (log-transformed for normality) showed significantly larger reductions in the adaptive condition at post-intervention, \u0026gamma; = 0.41, 95% CI [0.10, 0.72], t(3.4) = 3.12, p = .012, d = 0.31, 95% CI [0.07, 0.55]. Follow-up effects remained significant, \u0026gamma; = 0.38, 95% CI [0.06, 0.70], t(3.3) = 2.89, p = .018, d = 0.29, 95% CI [0.05, 0.53]. Change in financial self-efficacy correlated with change in discounting, r(387) = -.34, 95% CI [-.43, -.24], p \u0026lt; .001.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4 Effects on Perceived Stress:\u0026nbsp;\u003c/strong\u003ePerceived stress decreased in both conditions. The condition effect at post-intervention was not significant, \u0026gamma; = -1.8, 95% CI [-4.2, 0.6], t(3.5) = 1.72, p = .12, d = 0.18, 95% CI [-0.06, 0.42]. At follow-up, the condition effect was \u0026gamma; = -2.4, 95% CI [-4.82, -0.02], t(3.4) = 2.18, p = .042, d = 0.24, 95% CI [0.00, 0.48]. The narrow margin above zero suggests this effect should be interpreted with caution and may not replicate.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.5 Treatment Engagement:\u0026nbsp;\u003c/strong\u003eAmong treatment subsample participants, multilevel models revealed higher attendance in adaptive condition, \u0026gamma; = 2.1, 95% CI [0.4, 3.8], t(2.8) = 2.54, p = .028, d = 0.38, 95% CI [0.08, 0.68]. Counselor-rated engagement was also higher, \u0026gamma; = 1.8, 95% CI [0.2, 3.4], t(2.9) = 2.41, p = .032, d = 0.35, 95% CI [0.05, 0.65]. Self-reported between-session activity did not differ significantly, \u0026gamma; = 0.9, 95% CI [-0.7, 2.5], t(2.9) = 1.28, p = .22, d = 0.19, 95% CI [-0.11, 0.49].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.6 Relapse Intention:\u0026nbsp;\u003c/strong\u003eRelapse intention at post-intervention was lower in adaptive condition, \u0026gamma; = -4.8, 95% CI [-9.2, -0.4], t(3.1) = 2.34, p = .032, d = 0.32, 95% CI [0.02, 0.62]. Follow-up differences were similar, \u0026gamma; = -5.1, 95% CI [-9.6, -0.6], t(3.0) = 2.41, p = .028, d = 0.34, 95% CI [0.04, 0.64]. Relapse intention is an exploratory proximal outcome. It does not measure actual relapse.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.7 Associations Between Self-Efficacy and Outcomes:\u0026nbsp;\u003c/strong\u003eFinancial self-efficacy gains from baseline to post-intervention correlated with treatment engagement indicators: attendance r = .32, 95% CI [.20, .43]; counselor ratings r = .36, 95% CI [.24, .47]; self-reported activity r = .28, 95% CI [.16, .39]. Gains correlated negatively with relapse intention, r = -.29, 95% CI [-.41, -.16]. These correlations are descriptive and do not imply causation. Alternative explanations include third variables (e.g., motivation, treatment alliance) or reverse causality.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.8 Sensitivity Analyses:\u0026nbsp;\u003c/strong\u003eComplete-case analyses produced substantively identical results to multiple imputation. Models excluding covariates produced similar effect sizes. Results were robust to alternative specifications of random effects structure (site-only versus site plus instructor within control condition). Intraclass correlation coefficients ranged from .03 to .07 across outcomes, confirming modest site-level clustering.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.9 Note on Uniform Directionality of Results:\u0026nbsp;\u003c/strong\u003eAll observed associations trended in predicted directions across all outcomes, time points, and subsamples. This uniform pattern may reflect true effects but also raises credibility considerations. In small-site designs with exploratory analyses, such consistency can arise from unmeasured confounding, site-specific factors, or analytic choices. Replication is essential before confidence in these patterns is warranted.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003e\u003cstrong\u003e5.1 Summary of Findings:\u0026nbsp;\u003c/strong\u003eThe adaptive financial education condition was associated with larger gains in financial self-efficacy and larger reductions in delay discounting compared to traditional classroom instruction. These associations were similar in direction and magnitude for individuals in addiction treatment and at-risk university students, though power to detect differences was limited. Gains in self-efficacy correlated with treatment engagement and lower relapse intention. Financial hardship is prevalent among individuals entering addiction treatment. These findings suggest that addressing financial stress through adaptive education may be associated with psychological outcomes relevant to recovery. All findings are reported as associations. Causal claims are not warranted given the quasi-experimental design, small number of sites, lack of active control, and absence of experimental manipulation of proposed mechanisms.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.2 Interpretation and Limitations:\u0026nbsp;\u003c/strong\u003eSeveral interpretations are possible. The observed associations could reflect effects of the intervention condition package, which included technology delivery, visual engagement, immediate feedback, and interactivity in addition to adaptivity. They cannot be attributed specifically to adaptive mechanisms. The associations could reflect domain-general self-regulatory improvement that transfers from financial to substance use contexts, though this hypothesis is contested. De Ridder and colleagues' meta-analysis suggests self-control operates primarily through habits rather than effortful inhibition. Friese and colleagues concluded that ego depletion effects are small and context-dependent. Duckworth and colleagues emphasize domain-specific strategies. Alternatively, improved financial self-efficacy might reduce financial stress, freeing cognitive resources for treatment engagement without requiring domain-general capacity. Self-determination theory suggests that competence support through mastery experiences may enhance engagement directly.\u003c/p\u003e\n\u003cp\u003eA third possibility is that third variables (e.g., motivation, treatment alliance) explain both self-efficacy gains and treatment outcomes. Our design cannot distinguish these explanations. The five-site design produces unstable between-site variance estimates, as reflected in the small degrees of freedom for fixed effects. Confidence intervals are wide and significance tests are fragile. Replication with more sites is essential. This study was not pre-registered. The absence of pre-registration means all analyses are exploratory and susceptible to undisclosed analytic flexibility. Readers should interpret findings with this limitation in mind. The Financial Self-Efficacy Scale showed measurement non-invariance at the scalar level across subsamples, meaning latent mean comparisons between treatment and prevention groups should be interpreted cautiously. This is a limitation for combined analyses. The relapse intention measure, while having some predictive validity evidence, is not equivalent to actual relapse. Findings on this outcome are exploratory.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.3 The Uniform Directionality Issue:\u0026nbsp;\u003c/strong\u003eAll observed associations trended in predicted directions. While consistent with hypotheses, this uniform pattern in an exploratory study with five sites warrants caution. Possibilities include:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eTrue effects that are robust across outcomes and populations\u003c/li\u003e\n \u003cli\u003eUnmeasured confounding that systematically favors the adaptive condition\u003c/li\u003e\n \u003cli\u003eSite-specific factors not fully captured by random effects\u003c/li\u003e\n \u003cli\u003eAnalytic choices that inadvertently favor significance\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eReplication in independently conducted studies is the only way to distinguish these possibilities. Readers should treat these findings as preliminary.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.4 Theoretical Implications:\u0026nbsp;\u003c/strong\u003eResults are descriptively consistent with self-efficacy theory in that mastery experiences in one domain (financial decisions) were associated with outcomes in another domain (treatment engagement). However consistency does not confirm theory. Competing explanations remain plausible. Self-determination theory provides a framework for understanding why adaptive platforms might sustain engagement: competence support through mastery experiences, autonomy support through user control, and relatedness through responsive feedback. These features may explain associations regardless of domain-general transfer mechanisms. The contested status of domain-general self-regulation theories requires acknowledgment. Our findings do not resolve debates about the strength model or transfer mechanisms. They provide descriptive patterns that future research with stronger designs can test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.5 Implications for Practice:\u0026nbsp;\u003c/strong\u003eNo clinical recommendations are warranted from this study. Findings suggest that adaptive financial education may be associated with psychological outcomes relevant to addiction recovery, but causal evidence from randomized trials with active controls and larger numbers of sites is required before any implementation considerations arise. If replicated in stronger designs, such interventions could potentially complement existing treatment by addressing financial stress as a modifiable factor. The prevalence of financial hardship among treatment-seekers suggests this domain deserves attention. However current evidence is insufficient.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e5.6 Future Research Directions:\u0026nbsp;\u003c/strong\u003eRandomized controlled trials with individual-level assignment are needed to establish causality. Such trials should include larger numbers of sites or use individual randomization to avoid the site-level fragility of this study. Active controls should match on engagement and interactivity to isolate specific mechanisms. Measurement development is needed to validate financial self-efficacy instruments in clinical populations. The scalar non-invariance observed here suggests that comparisons across clinical and non-clinical groups require caution. Longer follow-up periods (12 months minimum) are required to assess associations with actual relapse, not just intention. Digital intervention research should integrate self-determination theory frameworks to identify which features (autonomy support, competence support, relatedness) drive engagement and outcomes. Pre-registered replication studies are essential before confidence in these findings is warranted.\u003c/p\u003e"},{"header":"CONCLUSION","content":"\u003cp\u003eThis study examined whether an adaptive financial education intervention was associated with psychological outcomes relevant to addiction recovery. The adaptive condition was associated with larger gains in financial self-efficacy and larger reductions in delay discounting compared to traditional instruction. Gains in self-efficacy were correlated with treatment engagement and lower relapse intention. Causal interpretation is precluded by design limitations including quasi-experimental assignment with only five sites, lack of active control, and correlational analyses. The absence of pre-registration means all findings are exploratory. The uniform directionality of results across outcomes requires replication before confidence is warranted. The study contributes descriptive evidence that adaptive learning technologies may be associated with psychological outcomes relevant to addiction. The prevalence of financial hardship among treatment-seeking populations suggests this domain deserves continued investigation. Whether these associations reflect causal effects, domain-general self-regulatory improvement, or third-variable confounding remains unknown. Replication in randomized designs with larger numbers of sites is essential.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003eDECLARATION OF COMPETING INTERESTS\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e\n\u003cp\u003eFUNDING STATEMENT\u003c/p\u003e\n\u003cp\u003eThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003eDATA AVAILABILITY STATEMENT\u003c/p\u003e\n\u003cp\u003eData, analysis code, and study materials will be deposited in an open-access repository upon acceptance. The data that support the findings of this study are available from the corresponding author upon reasonable request prior to acceptance.\u003c/p\u003e\n\u003cp\u003eCREDIT AUTHOR CONTRIBUTION STATEMENT\u003c/p\u003e\n\u003cp\u003eAuthor 1: Conceptualization, Methodology, Formal analysis, Writing \u0026ndash; Original Draft, Writing \u0026ndash; Review and Editing, Investigation, Data Curation, Project administration\u003c/p\u003e\n\u003cp\u003eAuthor 2: Software, Resources, Validation, Visualization, Supervision\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBandura A. (1997). Self-efficacy: The exercise of control. W. H. Freeman. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://psycnet.apa.org/record/1997-08589-000\u003c/span\u003e\u003cspan address=\"https://psycnet.apa.org/record/1997-08589-000\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrown SA, Vik PW, Patterson TL, Grant I, Schuckit MA. Stress, vulnerability and adult alcohol relapse. J Stud Alcohol. 1995;56(5):538\u0026ndash;45. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.15288/jsa.1995.56.538\u003c/span\u003e\u003cspan address=\"10.15288/jsa.1995.56.538\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.jsad.com/doi/\u003c/span\u003e\u003cspan address=\"https://www.jsad.com/doi/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCompton WM, Thomas YF, Stinson FS, Grant BF. Prevalence, correlates, disability, and comorbidity of DSM-IV drug abuse and dependence in the United States. Arch Gen Psychiatry. 2007;64(5):566\u0026ndash;76. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://jamanetwork.com/journals/jamapsychiatry/fullarticle/482282\u003c/span\u003e\u003cspan address=\"https://jamanetwork.com/journals/jamapsychiatry/fullarticle/482282\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Ridder DT, Lensvelt-Mulders G, Finkenauer C, Stok FM, Baumeister RF. Taking stock of self-control: A meta-analysis of how trait self-control relates to a wide range of behaviors. Personality Social Psychol Rev. 2012;16(1):76\u0026ndash;99. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/1088868311418749\u003c/span\u003e\u003cspan address=\"10.1177/1088868311418749\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeci EL, Ryan RM. The what and why of goal pursuits: Human needs and the self-determination of behavior. Psychol Inq. 2000;11(4):227\u0026ndash;68. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://psycnet.apa.org/doi/10.1207/S15327965PLI1104_01\u003c/span\u003e\u003cspan address=\"https://psycnet.apa.doi/10.1207/S15327965PLI1104_01\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuckworth AL, Taxer JL, Eskreis-Winkler L, Galla BM, Gross JJ. Self-control and academic achievement. Ann Rev Psychol. 2019;70:373\u0026ndash;99. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev-psych-010418-103230\u003c/span\u003e\u003cspan address=\"10.1146/annurev-psych-010418-103230\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFriese M, Loschelder DD, Gieseler K, Frankenbach J, Inzlicht M. Is ego depletion real? An analysis of arguments. Perspect Psychol Sci. 2019;14(3):489\u0026ndash;96. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1088868318762183\u003c/span\u003e\u003cspan address=\"10.1177/1088868318762183\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHagger MS, Chatzisarantis NL, Alberts H, Anggono CO, Batailler C, Birt AR, Zwienenberg M. A multilab preregistered replication of the ego-depletion effect. Perspect Psychol Sci. 2016;11(4):546\u0026ndash;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1745691616652873\u003c/span\u003e\u003cspan address=\"10.1177/1745691616652873\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKiluk BD, Nich C, Buck MB, Devore KA, Frankforter TL, LaPaglia DM, Carroll KM. Randomized clinical trial of computerized and clinician-delivered CBT in comparison with standard outpatient treatment for substance use disorders. Am J Psychiatry. 2018;175(9):853\u0026ndash;63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1176/appi.ajp.2018.1709097\u003c/span\u003e\u003cspan address=\"10.1176/appi.ajp.2018.1709097\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLown JM. Development and validation of a financial self-efficacy scale. J Financial Couns Plann. 2011;22(2):54\u0026ndash;63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://files.eric.ed.gov/fulltext/EJ952966.pdf\u003c/span\u003e\u003cspan address=\"https://files.eric.ed.gov/fulltext/EJ952966.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarsch LA, Carroll KM, Kiluk BD. Technology-based interventions for the treatment and recovery management of substance use disorders: A JSAT special issue. J Subst Abuse Treat. 2014;46(1):1\u0026ndash;4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jsat.2013.08.010\u003c/span\u003e\u003cspan address=\"10.1016/j.jsat.2013.08.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcKay JR, Franklin TR, Patapis N, Lynch KG. Conceptual, methodological, and analytical issues in the study of relapse. Clin Psychol Rev. 2006;26(2):109\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cpr.2005.11.002\u003c/span\u003e\u003cspan address=\"10.1016/j.cpr.2005.11.002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcLellan AT, Cacciola JC, Alterman AI, Rikoon SH, Carise D. The Addiction Severity Index at 25: Origins, contributions and transitions. Am J Addictions. 2006;15(2):113\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/10550490500528316\u003c/span\u003e\u003cspan address=\"10.1080/10550490500528316\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://onlinelibrary.wiley.com/doi/\u003c/span\u003e\u003cspan address=\"https://onlinelibrary.wiley.com/doi/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRachlin H. The science of self-control. Harvard University Press; 2000. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.hup.harvard.edu/books/9780674013575\u003c/span\u003e\u003cspan address=\"https://www.hup.harvard.edu/books/9780674013575\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTorous J, Bucci S, Bell IH, Kessing LV, Faurholt-Jepsen M, Whelan P, Firth J. The growing field of digital psychiatry: Current evidence and the future of apps, social media, chatbots, and virtual reality. World Psychiatry. 2020;20(3):318\u0026ndash;35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/wps.20883\u003c/span\u003e\u003cspan address=\"10.1002/wps.20883\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatts TW, Duncan GJ, Quan H. Revisiting the marshmallow test: A conceptual replication investigating links between early delay of gratification and later outcomes. Psychol Sci. 2018;29(7):1159\u0026ndash;77. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/0956797618761661\u003c/span\u003e\u003cspan address=\"10.1177/0956797618761661\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://journals.sagepub.com/doi/\u003c/span\u003e\u003cspan address=\"https://journals.sagepub.com/doi/\" targettype=\"URL\" 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":"","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":"Financial education, Self-efficacy, Addiction recovery, Treatment engagement, Adaptive learning","lastPublishedDoi":"10.21203/rs.3.rs-9324305/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9324305/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFinancial stress is associated with substance use disorder severity and relapse. Among individuals entering addiction treatment, financial hardship is highly prevalent. Mechanisms linking financial interventions to addiction outcomes remain underspecified. This study examines whether an adaptive financial education intervention is associated with improvements in financial self-efficacy, delay discounting, and treatment-related outcomes among individuals in addiction treatment and at-risk university students. The study is descriptive; causal claims are not warranted given design limitations. Participants (N\u0026thinsp;=\u0026thinsp;420; 240 in outpatient treatment, 180 at-risk university students) were assigned by site to 12 weeks of adaptive financial education delivered through an interactive platform with personalized feedback, or to traditional classroom instruction. Multilevel models accounting for site-level clustering tested condition associations with financial self-efficacy, delay discounting, perceived stress, treatment engagement, and relapse intention. With only five sites, between-site variance estimates are unstable and all effects require cautious interpretation. The adaptive condition was associated with larger gains in financial self-efficacy (γ\u0026thinsp;=\u0026thinsp;8.2, 95% CI [3.1, 13.3], p = .004) and larger reductions in delay discounting (γ\u0026thinsp;=\u0026thinsp;0.41, 95% CI [0.10, 0.72], p = .012). Financial self-efficacy gains correlated with treatment engagement (r = .34, 95% CI [.22, .45]) and relapse intention (r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;.29, 95% CI [-.41, \u0026minus;\u0026thinsp;.16]). All effects trended uniformly in predicted directions; replication is required before confidence is warranted. Adaptive financial education is associated with psychological outcomes that may support addiction recovery. Findings require replication in randomized designs with active controls and larger numbers of sites.\u003c/p\u003e","manuscriptTitle":"Adaptive Financial Education and Addiction Recovery: Associations with Self-Efficacy and Treatment Engagement in a Quasi-Experimental Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-07 05:07:22","doi":"10.21203/rs.3.rs-9324305/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":"f1c48a1d-3316-43a2-b107-b4db847f0518","owner":[],"postedDate":"April 7th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-16T07:27:59+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-07 05:07:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9324305","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9324305","identity":"rs-9324305","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}