Abstract
Ecological momentary assessment (EMA) prompts participants to report on their current emotions, cognitions, or behaviors multiple times per day across several days, yielding rich, high-frequency data. However, the high frequency of measurement can burden participants, leading to disengagement in the form of non-compliance or careless/insufficient effort responding (C/IER), which compromises the validity of conclusions drawn from EMA data. To minimize these effects, it is crucial to understand how EMA design factors influence respondent disengagement. This study investigates how response format, a comparatively subtle design choice, affects C/IER as an important but hard-to-detect aspect of disengagement. We re-analyzed data from two experimental EMA studies, in which surveys were administered either in a Likert or a visual analogue scale (VAS) format in between-subjects designs. Each dataset was analyzed with one of two state-of-the-art confirmatory mixture modeling approaches to examine whether response format affects (a) the overall level of C/IER and (b) its associations with two fatigue- and attention-related contextual factors: the cumulative number of EMA study days and time of day. Across datasets and analytical approaches, we found indications that Likert scales yield up to 3 percentage points higher rates of C/IER than VASs, though the temporal patterns varied across studies. In Study I, differences were observed in overall C/IER levels, with parallel increases across EMA days. In Study II, initial C/IER was comparable between formats, but the Likert format showed a steeper increase over time. Implications for EMA design are discussed.