Interpretation bias and safety behaviours (Safe-B) have been proposed as factors perpetuating social anxiety (SA). However, longitudinal research on how they contribute to SA in everyday life is scarce.

The aim was to examine whether interpretation bias predicts daily Safe-B and SA. A mediated moderation was hypothesized, where the relationship between daily social stressors and Safe-B would be moderated by interpretation bias, and Safe-B, in turn, would mediate the association between stressors and SA levels. In addition, it was hypothesized that prior levels of SA would predict higher Safe-B use, especially in co-occurrence with stressors.

An intensive longitudinal design was employed, with 138 vocational training students (51% men, mean age 20.15 years). They completed initial measures of SA and interpretation bias and 7-day diaries with measures of social stressors, Safe-B, and SA. They reported SA levels two months later.

Both stressors and interpretation bias in ambiguous situations predicted Safe-B, which in turn predicted daily SA levels. However, neither interpretation bias nor Safe-B predicted SA levels at the follow-up, and interpretation bias did not moderate the association between stressors and daily SA. In addition, the relationship between stressors and Safe-B was stronger in people with higher initial SA levels.

The results suggest that Safe-B are a mechanism through which earlier SA levels and interpretation bias contribute to higher SA levels in daily life.