Category: Cognitive Science / Cognitive Processes
The Attentional Control Scale (ACS) is used to investigate executive features of attention. Psychometric studies of the ACS have found two correlated factors, Focusing and Shifting. Focusing is comprised of reverse-scored items and Shifting of straight-forward items, raising the possibility that two-factor solutions are due to method variance. Furthermore, two factor models have not included a general factor, though this is theoretically assumed, and factor variance is often better explained by a bifactor model which includes a superordinate general factor. The goal of this study was to test the two-factor model of the ACS and whether a bifactor model of the ACS provides superior fit relative to the two-factor model.
The ACS, with reverse-scored items reworded to be straight-forward, was completed by 419 participants through Amazon Mechanical Turk. Confirmatory factor analyses were conducted in Mplus version 7.4, using robust maximum likelihood. Nested models of the ACS two-factor and bifactor models were examined.
The two-factor model showed adequate model fit, χ2=2,266.08, p < .001, CFI=.93, RMSEA=.078, SRMR=.05. The fit of the bifactor model was significantly better than the two-factor model (Y-B χ2=113.69, ∆df=11, p < .001) and provided adequate fit overall (Y-B χ2=93.06, df=42, p < .001, CFI=.97, SRMR=.03, RMSEA=.05, 90% CI [.04, .07]). All items loaded significantly onto the General AC factor (λs=.49-.86). All focusing items also loaded onto the Focusing factor, although one item loaded below the commonly accepted .32 criterion (Tabachnick & Fidell, 2007; λs=.26-.49). Four items loaded significantly onto the Shifting factor, also with one item below the .32 threshold (λs=.31-.47). The factor determinacy scores were .95, .77, and .74 for the General AC, Focusing, and Shifting factors, respectively. The General AC factor accounted for 87.56% of the overall variance. The Focusing factor accounted for 31.71% of the variance in items posited to measure focusing, and the Shifting factor accounted for 11.57% of the variance in the items posited to measure shifting. The results suggested that the ACS consists of a general factor, explaining the preponderance of variance in the items along with two sub-factors, which offer less explanation of item variance.
These findings support the two-factor structure of the ACS, suggesting that these factors are not better explained as an artifact of reverse-scored items. However, the vast majority of item variance in the ACS is attributable to a common general factor. Future studies using the ACS may benefit by using a bifactor model. Given that the general factor explains most of the variance in the ACS, future research using the ACS should focus on the general factor.
Elizabeth Bauer– Old Dominion University, Virginia Beach, Virginia
Ekaterina Shurkova– Old Dominion University
Kevin Saulnier– Graduate Student, Ohio University, Athens, Ohio
Matt Judah– Old Dominion University
Nicholas Allan– Assistant Professor, Ohio University