Nomothetic and idiographic determinants of dietary lapses and adherence: an exploratory mixed methods ecological momentary assessment study of open-text responses.

OBJECTIVE: Dietary interventions often fail to achieve sustained behavior change, partly due to limited understanding of momentary factors driving dietary lapses and instances of goal adherence. Given constrained item pools in ecological momentary assessment (EMA), it remains unclear which factors are most relevant to individuals in daily life. This study therefore examined dieters' self-reported momentary reasons for lapses and adherence and whether these are broadly applicable (nomothetic) or individual-specific (idiographic). METHODS: Data were drawn from two EMA studies (N = 199) in which participants reported lapses and adherence every 3.5 h (over 1 or 2 weeks) and provided open-text explanations. In total, 5,402 reasons were analyzed using AI-assisted qualitative content analysis. Categories were evaluated with three indices to assess their nomothetic versus idiographic relevance. RESULTS: Overarching categories were similar for lapses and adherence (self-regulation, food availability, social influences, lack of opportunity, unplanned circumstances, emotional, physiological, craving-related factors), but subcategories differed (e.g., deliberate exceptions vs. routines and strategies). No lapse category was reported by >47.73% of participants, indicating substantial idiographic variability. The most broadly applicable factors across participants were cravings, physiological factors (especially hunger), and social influences for lapses, and lack of opportunity (especially being occupied) for goal adherence. CONCLUSION: Participants highlighted (sub)categories typically underrepresented in lapse prediction models (e.g., health-related states, availability of goal-congruent foods, lack of opportunity). Substantial individual variability further suggests that lapse reasons do not generalize across participants and that personalized item selection would probably improve lapse prediction and just-in-time adaptive interventions.