To evaluate our mediation model, we used the PROCESS macro, developed by
Hayes (2013), to test a mediation model with multiple mediators operating in serial.
This macro allows us to test the direct effect of TMM-similarity on effectiveness and
the indirect effects of TMM-similarity on effectiveness through the mediators, even
when there is no association between TMM-similarity and effectiveness. For the path
analyses, team size and team composition were entered as control variables. We
resampled 5000 times and examined for 95 per cent confidence intervals (CI).
Bootstrapping produces a CI, based on bootstrapped sampling distribution, and we can
assume that the indirect effects are significant and that mediation occurred if zero falls
outside of the 95 per cent CI (Preacher and Hayes, 2008).