Analyzing the ebb and flow of the warp and woof with RIVAR mediation

Google map

Intra-individual variability is central to many theories of development. To study how intra-individual variability mediates slower timescale processes, we present random-intercepts vector autoregressive (RIVAR) mediation. This approach extends the level-2 model of the multilevel measurement error VAR model to enable analysis of fast-timescale mediators of slow-timescale effects. We demonstrate this method with an application to a study of parent-teen affect, examining how daily parent and teen affect dynamics mediate the relationship between parents' psychological distress at baseline and teens' perceived end-of-day stress. We find that adolescent's baseline affect credibly mediates the relationship (posterior median = 0.07, SD = 0.04, 95% CrI: [0.01, 0.16]). We conclude by discussing limitations and extensions of this approach.