Let sleeping dogs lie? How to deal with the night gap problem in ESM data

Abstract

Night gaps are inherent to data obtained with the experience sampling method (ESM). When such data are used to study lagged relations between variables--such as autoregression within the same variable, and cross-lagged regressions between different variables--the actual role of night gaps is typically not investigated. However, there are various methods to handle them in analyses. Common solutions involve: (1) ignoring the night gap by considering the night interval as a regular interval; (2) removing the night gap by not regressing the first measurement of the day on the last measurement of the previous day; or (3) treating the night gap as a missing data problem. The goal of this paper is to make explicit the theoretical implications of these three methods within the context of the first-order autoregressive (AR[1]) model. Additionally, we propose an alternative modeling approach that allows us to study the implications of the night gap in more detail. Moreover, given that the current methods are special cases of the proposed alternative, we can test which method best describes the process of interest. Through an empirical N=1 example with various ESM variables, we demonstrate that the best-fitting method differs per variable. This implies that some processes may exhibit different dynamics during the night than during the daytime, providing a stepping stone to understanding and modeling night gaps in ESM.