Data Preparation and Analysis

Mean RTs for each sequence (repetition, alternating, switching) of the task switching were collected and computed using SuperLab (Cedrus Corporation, Inc.). Trial errors were not included in the calculations of backward inhibition and switch cost indices. In addition, RTs of ±2 standard deviations from the mean were not included in the calculations. Statistical analysis was computed using IBM SPSS 24. Data were checked for normal distribution (substantial departure from normality was considered as skew value >2 and kurtosis value >7). Descriptive analysis of the sample was performed. One-way analysis of variance (ANOVA) was performed to assess the differences in RTs associated with AAA, ABA, and CBA sequences of the task switching. Furthermore, mixed-design factorial ANOVAs were implemented to examine the interactions between the within-factor task-switching RTs and the between-factor of the group (insomnia vs. good sleepers). Bivariate correlations between the main study variables (PTQ, BSRI, ISI, backward inhibition, switch cost, accuracy) were performed. Furthermore, to estimate the effects of PTQ on task-switching paradigm at different insomnia levels, the sample was split according to ISI scores (Morin et al., 2011) in high insomnia symptoms (ISI > 10) or low insomnia symptoms (ISI < 10), and partial correlations were computed in the two subsamples between PTQ and task-switching indices controlling for the effects of BSRI. Finally, conditional process modeling (Hayes, 2013) was performed to assess the hypothesis that insomnia severity may moderate the relationship between PC and executive functions. In these analyses, PTQ was inserted as independent variable, ISI was inserted as moderator, and task-switching indices were inserted as outcomes. Moreover, to control for the effects of state rumination on task switching, BSRI was inserted as covariate. Two separate analyses were computed for each task-switching outcome (backward inhibition, switch cost). According to our main hypothesis, we expected to find significant PTQ × ISI interactions on backward inhibition and switch cost. Moderation analysis was computed on PROCESS macro for SPSS (Hayes, 2013).