Uncovering the Missing Pieces: Predictors of Nonresponse in a Mobile Experience Sampling Study on Media Effects Among Youth

Abstract

Mobile Experience Sampling (MES) is a promising tool for understanding youth digital media use and its effects. Unfortunately, the method suffers from high levels of missing data. Depending on whether the data is randomly or non-randomly missing, it can have severe effects on the validity of findings. For this reason, we investigated predictors of non-response in an MES study on displacement effects of digital media use on adolescents’ well-being and academic performance (N = 347). Multilevel binary logistic regression identified significant influencing factors of response odds, such as afternoon beeps and being outside. Importantly, adolescents with poorer school grades were more likely to miss beeps. Because this missingness was related to the outcome variable, modern missing data methods such as multiple imputation should be applied before analyzing the data. Understanding the reasons for non-response can be seen as the first step to preventing, minimizing, and handling missing data in MES studies, ultimately ensuring that the collected data is fully utilized to draw accurate conclusions.