We then constructed a structural equation model to test the media- tion hypotheses. Four item parcels (each consisting of two randomly selected items) were used as indicators of the conscientiousness and ex- troversion latent constructs. Researchers traditionally had difficulty establishing suitable CFA models for personality constructs (Church & Burke, 1994; Nye, Roberts, Saucier, & Zhou, 2008) due to the covariance among personality constructs, which leads to cross-loaded items. We en- countered the same issue; however the use of item parcels improved the fit of our model and lowered our participant-to-parameter ratio.1 Fur- ther, following the guidance of Williams and O'Boyle (2008) we deter- mined that item parceling was appropriate due to our interest in structural relationships and not the measurement of personality itself. For all other measures we used items as indicators of the latent con- structs. Pathways were added to the model as indicated in Fig. 1.