Analyses

Preliminary analyses were conducted using IBM SPSS Version 27.0. Path analysis with a cross-lagged panel configuration was used to test the relationship between caregiver R/S beliefs and practices, positive coping, and psychological distress using AMOS 27.0. The sample size of this study is consistent the range that is typically expected within complex mediation analyses (Fritz and MacKinnon, 2007). Model fit was evaluated using three fit indices: the chi-square measure of overall goodness of fit, the comparative fit index (CFI), and the root mean square error of approximation (RMSEA; Hooper et al., 2008). Attrition over time and any other missing data was handled using Full Information Maximum Likelihood (FIML) estimation that is based upon a missing at random (MAR) assumption. FIML estimation allows for the maximum use of the data available as it neither substitutes values for missing data points nor deletes cases that have missing data points. The FIML method instead estimates parameter values that have the highest likelihood of being observed in the current sample data and then accumulates and maximizes that data. The significance of any indirect effects was evaluated using confidence intervals constructed using Monte Carlo simulation methods appropriate for structural equation models (Falk and Biesanz, 2016). Mediation was considered as significant if the 95% bias corrected and accelerated confidence intervals for the indirect effects did not include 0. All pathways controlled for caregiver age, sex, race (white or non-white participants), and pandemic-related disruption (measured using the CoFaSS) at T1.