Exploring in-store and e-shopping against disruptive events: A cross-lagged panel SEM

This paper addresses a key gap in the literature by examining the dynamic and bidirectional relationship between in-store and e-shopping frequency during different stages of the COVID-19 pandemic. Previous studies primarily rely on cross-sectional data which fail to capture the temporal evolution and bidirectional nature of these behaviours. To overcome these limitations, this study implements a Random Intercept Cross-lagged Structural Equation Modelling (RI-CLPM) approach using three waves of panel data. Taking Luxembourg as the case study, the paper investigates the modifications in in-store shopping-related travel behaviour by evaluating shifts in trip frequency for three periods: pre-pandemic, post-peak, and relaxed measures phase. The results showed a significant shift in shopping frequency between the pre-pandemic and post-peak phase, evidencing substitution and complementarity effects both on individual as well as group level. Moreover, ANOVA and chi-square tests suggested that age and gender significantly influence in-store shopping frequency for these periods. However, no significant differences in e-shopping and in-store shopping frequencies were observed between the post-peak and the relaxed measures period. These findings provide critical insights for understanding shopping behaviour transitions and offer valuable guidance for transport policymaking. The paper closes by discussing how RI-CLPM models may improve transport policymaking, in the context of future disruptions, considering their potential for: (i) isolating policy impacts amid individual differences, (ii) addressing stable and dynamic shopping behaviours, and (iii) dealing with longitudinal data that allows for adaptive policy design.