Measuring Implicit Attitudes Toward Digital Health Technologies in Older Adults Using an Influence-Aware Affect Misattribution Procedure: Development and Feasibility Study

Abstract Background Older adults often express positive attitudes toward digital health technologies in surveys, yet adoption remains low. Self-report measures may not capture automatic affective reactions such as anxiety or distrust. Implicit paradigms such as the affect misattribution procedure (AMP) can reveal these automatic attitudes, but parameters optimized for younger adults may not be suitable for older adults because of age-related slowing and changes in visual processing. Objective This study aimed to adapt and evaluate an influence-aware affect misattribution procedure (IA-AMP) for measuring implicit attitudes of older adults toward digital health technologies and to identify an age-appropriate prime duration that balances affect transfer strength with minimal conscious awareness. Methods A 2-phase methodological adaptation and feasibility study was conducted among older adults (aged ≥60 y). Phase 1 (n=40) involved the development and validation of age-relevant synthetic images depicting older adults using digital health tools. The images were evaluated based on valence, arousal, thematic relevance, and low-level perceptual features. Phase 2 (n=56) implemented an IA-AMP with 3 prime durations (75, 350, and 425 ms) across 2 sequential cohorts. The first cohort (batch 2A, n=29) used the initial awareness probe, whereas the second cohort (batch 2B, n=27) used a simplified awareness interface. The core IA-AMP target judgment task remained unchanged across batches. The primary inferential outcome was the binary trial-level target judgment, coded as pleasant or unpleasant. Trial-level responses were analyzed using binomial logistic mixed-effects models with prime valence, prime duration, their interaction, and batch as fixed effects, along with random intercepts for participant and prime image. Awareness analyses were restricted to batch 2B. Results The primary generalized linear mixed-effects model showed a significant prime valence × prime duration interaction ( χ 2 2 =38.1; P <.001). Positive primes increased the odds of pleasant target judgments relative to negative primes at all durations (odds ratio [OR] 11.2, 95% CI 5.8-21.6, at 75 ms; OR 31.9, 95% CI 16.0-63.4, at 350 ms; and OR 45.6, 95% CI 22.1-94.0, at 425 ms; all Holm-adjusted P <.001). The positive-negative contrast was smaller at 75 milliseconds than at 350 and 425 milliseconds, whereas the contrasts at 350 and 425 milliseconds did not differ significantly. Batch sensitivity analyses showed stronger overall priming in batch 2B, but the prime valence × prime duration × batch interaction was not significant ( χ 2 2 =0.76; P =.68). Conclusions The IA-AMP offers a promising approach for assessing the affective responses of older adults to digital health technologies beyond self-report. A prime duration of approximately 350 milliseconds appears to be a practical calibration point for AMP studies involving older adults, producing strong affect transfer effects while avoiding the longest exposure duration. Because reported influence awareness was common, AMP effects should be interpreted alongside awareness measures rather than as awareness-free implicit attitudes. AMP-based affective measures may complement usability and adoption research by identifying emotional responses that users may not readily articulate, thereby supporting more inclusive and evidence-informed development, evaluation, and implementation of digital health technologies for aging populations.