B0865
Title: A continuous-time distributed lag model for experience sampling data
Authors: Biplab Paul - University of Haifa (Israel)
Philip Reiss - University of Haifa (Israel) [presenting]
Abstract: The experience sampling method (ESM) has emerged in the last fifteen years as a critical tool for mental health research. In ESM studies, participants are contacted at random times and asked about their activities and mental states at that moment. One class of questions that such a design can help to answer concerns the effect of past mental states on present ones, such as how stress experienced earlier in the day impacts one's current mood. Such time-lagged effects are often estimated by distributed lag models, but these presuppose a discrete set of lags. For ESM data acquired at random times, one needs a new variant of discrete lag modeling for time lags that are not discrete but continuous. A novel semiparametric model is proposed that meets this need and estimates the effect of past predictor values as a smooth function of the time lag. The model is implemented via the generalized additive mixed model software of Wood and coworkers. This allows for a wide range of response types, including ordinal responses, which are very common in ESM research. The proposed model is illustrated with data from a recent study of mental time orientation.