Exploring Effects Of Retrospective Reporting Within Ecological Momentary Assessment Research In Physical Activity

Ecological momentary assessment (EMA), which allows capture of time-series data, is gaining popularity to understand the complexities of physical activity (PA) behavior. Specifically, event-contingent EMA is necessary to assess integral aspects of PA experience (e.g., pre-activity mental / physical states; post-activity recall of pleasure). Collecting momentary data is prone to human error under real-world circumstances (e.g., delayed reporting) and may challenge data usability or skew results. PURPOSE: To explore the impact of non-momentary EMA data on associations between relevant pre- and post-PA variables. METHODS: The current project is a secondary analysis of a study wherein participants (n = 27) were asked to initiate EMAs before and after purposeful PA sessions for 10-wk (≥10 min; 602 sessions reported). Validated instruments (e.g. Feeling Scale, Acute Readiness Monitoring Scale) captured pre-exercise physical readiness (PR) and fatigue (PF), and recalled affective valence (rAV). To reduce momentary reporting barriers, an option to complete a retrospective report was included; 50% of PA sessions were recorded retrospectively. Repeated measures correlation procedures were used to estimate within-person associations for paired measures, both with and without the inclusion of retrospective data. Effects were interpreted based on standard values (small, r = .10 to .29; medium, r = .30 to .49; large, r ≥ .50) and included 95% confidence intervals. RESULTS: PA-related rAV was inversely related to PF based on all data (r = -.31; CI -.38, -.23), only momentary data (r = -.28; CI -.38, -.16), and only retrospective data (r = -.33; CI -.43, -.22). rAV was linearly related to PR using all data (r = .33; CI .26, .40), momentary data (r = .29; CI .18, .39), and retrospective data (r = .38; CI .27, .48). All correlations were statistically significant (p < .01). CONCLUSION: Regardless of the report type, association between PA experience variables were consistent in directionality, and magnitude as seen by the inclusion of r values in the overlapping CIs. Retrospective EMA reporting can aid in obtaining ample time-series observations for statistical power, but the potential impact of delayed survey completion should be transparently reviewed prior to conducting primary analyses to yield quality estimates of effects.