Research master thesis | Psychology (research) (MSc)
open access
Ecological momentary assessment (EMA) is a data collection method in which participants’ current behaviors and experiences are sampled repeatedly in their natural environment. EMA has advantages...Show moreEcological momentary assessment (EMA) is a data collection method in which participants’ current behaviors and experiences are sampled repeatedly in their natural environment. EMA has advantages over retrospective research methods, in that it reduces retrospective bias, increases ecological validity, and offers the possibility to observe dynamical changes of variables. However, EMA protocols are burdensome for participants and may interfere with their daily activities. This can lead to non-compliance over the course of a study. Missing data can subsequently decrease statistical power, and even induce bias. This paper explored whether missing data can be predicted by various variables related to students’ primary motivation to participate, mental health, stress levels, and demographics. We analyzed data of the first cohort (N = 418) of the ongoing WARN- D project on student mental health. Participants completed a comprehensive baseline survey and took part in an 85-day long EMA study. We predicted overall rates of non- compliance by participant characteristics at baseline (Analysis 1) and weekly rates of non- compliance by time-varying factors during the EMA stage (Analysis 2). Analysis 1 showed that overall non-compliance can be predicted by baseline measures such as age, depression, substance use, and primary motivation to participate. Analysis 2 showed that weekly assessed time-varying measures like time into study, enjoyment of the study, weekly stress, anxiety, and depression may predict weekly rates of non-compliance. Participant’s sex and smartphone operating system were not related to overall non-compliance. Summarizing, non-compliance rates of participants can be predicted by participant characteristics at baseline as well as by time-varying predictors. Our findings may inform future research on potential mechanisms behind noncompliance in EMA designs that should be considered to maximize participation rates while avoiding biased conclusions.Show less
