Hypotheses Testing
Before examining the parameter estimates, we first needed to run null models (i.e.,
without entering any Level-1 and Level-2 predictors) to examine if the betweengroup
variance in the outcome variables were significant. The results of the null
model for extra-role performance were: t00 = 0.19, c2
(81) = 220.52, p < 0.01, and
ICC(1) = 0.24. The results showed that extra-role performance had 24 percent
between-group variance. Similar results were found for in-role performance
[t00 = 0.30, c2
(81) = 250.68, p < 0.01, and ICC(1) = 0.28]. All the null-model
results justified the application of cross-level HLM testing (Seibert et al., 2004).