(PS8-C55) Predictors of Reactivity and Adherence in Ecological Momentary Assessment Studies

Background: Ecological momentary assessment (EMA) entails repeated, real-time sampling of participants to capture moods, thoughts, behaviors, and situational context in vivo. Such methodology is beneficial as it can reduce retrospective recall biases; enhance ecological validity; establish temporal sequence; and track fluctuations in the variables being measured. However, previous research shows that repeated sampling of an individual can change reports about the variable(s) being assessed (i.e., reactivity). This can be beneficial for intervention purposes (e.g., decreasing unwanted moods or behaviors or increasing desirable states), but it can be a problem when the study goal is to simply capture a naturalistic picture of daily life as it is lived, or if such reactivity leads to an increase in negatively-valenced (or a decrease in positively-valenced) outcomes. Previous research has also suggested that some people might be more likely to adhere to EMA studies than others, and study design factors can also enhance or suppress adherence. For example, there is some indication that study compensation method, participant demographics, and baseline mental health symptoms might impact adherence, although specific findings have been inconsistent.

Methods: We combined data from five previous EMA studies (total n = 498 college students, total completed EMA assessments = 10,422; 76.31% female, 73.27% White, average age = 19.59) to investigate reactivity (change in negative mood/thoughts, positive mood/thoughts, stress, stressor appraisals over a week-long study) and predictors of adherence (total assessments completed, response or non-response, time delay to assessment prompts). We used hierarchical linear modeling, t-tests, and bivariate correlations, as appropriate, in our analyses.

Results: EMA reports of positive mood, positive thoughts, stress, and stressor control, stressor coping efficacy, and stressor undesirability decreased over the course of the study week (all p < .05). Reports of negative mood and thoughts did not change (all p > .05). Participants were less likely to respond to assessments and showed a greater response delay over the course of the study week (all p < .001). Response delay was greater in the mornings than in the afternoon/evenings (p = .003) but did not differ between weekdays and weekends (p = .22). Participants who identified as female, reported less neuroticism, and were compensated with cash (vs. a cash/extra credit combination) completed significantly more EMA surveys (all p < .05). Other baseline variables (age, GPA, class year, major, anxiety, depression) were unassociated with adherence rate. Upperclassmen and those compensated with cash took longer to respond to surveys (all p < .05). No other baseline or demographic variables were associated with response delay (all p > .05).

Conclusion: This study highlights important design considerations for EMA researchers interested in maximizing adherence and minimizing reactivity. Cash incentives should be offered, particularly to participants at risk of low adherence; assessment reactivity should be monitored; and surveys should be administered when participants are most likely to respond.