Capacity, attention control, and secondary memory, also predicted gF. This model tests whether capacity, attention control, and secondary memory mediate the relation between WM storage and gF. If these factors do mediate the relation we should see that WM storage predicts all three factors, all three factorss significantly predict gF, but WM storage no longer has a direct effect on gF. This would suggest that the factors fully mediate the relation. If, however, WM storage still predicts gF after controlling for these other factors, then some other factor is also needed to explain the relation. As shown in Table 3 the fit of this model was good. Shown in Fig. 3 is

the resulting model. As can be seen, WM storage significantly

predicted each of the factors suggesting that WM storage is uniquely related to each of the factors (capacity, attention control, and secondary memory retrieval). CH5424802 molecular weight Additionally, each of the factors significantly predicted gF suggesting that each of the factors contributes to variation in gF. Most importantly, the direct path from WM storage to gF was not significant. That is, the correlation between WM storage and gF went from r = .57 to roughly zero after statistically controlling for the other factors. Thus, capacity, attention control, and secondary memory jointly mediated the relation between WM storage and selleck chemical gF. Once these three factors were taken into account WM span no longer predicted residual variance in gF. Furthermore, as shown in Table 3, fixing any of the paths from WM storage to the three factors (AC, SM, capacity) to zero resulted

in significantly worse model fits (all Δχ2’s > 6.5, p’s < .01). Likewise, fixing any of the paths from the three factors to gF to selleck zero resulted in significantly worse model fits (all Δχ2’s > 8.4, p’s < .01). However, fixing the path from the residual WM storage factor to gF to zero, did not change the model fit (Δχ2 = .04, p > .84). Thus, omitting any of the paths from WM storage to the three factors or from the factors to gF would reduce the fit of the model and limit the ability to account for variance in gF. These results are directly in line with the multifaceted view of WM which suggests that primary memory (capacity and attention control) and secondary memory underlie individual differences in WM span and account for their predictive power ( Unsworth and Engle, 2007a and Unsworth and Spillers, 2010a). Next, we added WM processing into the models to determine its relation with the other constructs. Specifically we specified the same measurement model shown in Fig. 2 (Measurement Model 5), and added in a factor for WM processing based on the three processing time measures taken from the complex span tasks. As shown in Table 3 the fit of this model was good. Shown in Fig. 4 is the resulting model.