Since the last beep or in the last 30 minutes: The effect of items’ retrospective time frame on their functioning and usability in an experience sampling study

Experience sampling method (ESM) studies often use items capturing retrospective time frames (e.g., the last hour). However, the effects of retrospective time frames on responses and response processes remain unclear. We experimentally manipulated retrospective time frame (in the last 30/60/90 minutes; since the last beep) in N=192 students. Outcomes included distributional properties of responses, item user-friendliness (ease of responding, burden, response time), and predictive (relationship between stressful events and negative affect) and convergent (relationship between baseline- and ESM-assessed alcohol use) validity. We found that 30-minute time frames yielded strongest predictive validity, whereas 90-minute time frames produced the slowest responses, greatest overlap with adjacent assessment periods, and lowest convergent validity. In the since the last beep condition, time frame length predicted longer response times. We recommend matching time frames to study goals, and transparently reporting on the choice, wording, and analysis of items with retrospective time frames.