Factor structure and validation of the attentional control scale

Abstract

The Attentional Control Scale (ACS; Derryberry & Reed, 2002) has been used to assess executive control over attention in numerous studies, but no published data have examined the factor structure of the English version. The current studies addressed this need and tested the predictive and convergent validity of the ACS subscales. In Study 1, exploratory factor analysis yielded a two-factor model with Focusing and Shifting subscales. In Study 2, confirmatory factor analysis supported this model and suggested superior fit compared to the factor structure of the Icelandic version (Ólafsson et al., 2011). Study 3 examined correlations between the ACS subscales and measures of working memory, anxiety, and cognitive control. Study 4 examined correlations between the subscales and reaction times on a mixed-antisaccade task, revealing positive correlations for antisaccade performance and prosaccade latency with Focusing scores and between switch trial performance and Shifting scores. Additionally, the findings partially supported unique relationships between Focusing and trait anxiety and between Shifting and depression that have been noted in recent research. Although the results generally support the validity of the ACS, additional research using performance-based tasks is needed.

Keywords:

• Attentional control
• Factor analysis
• Saccade
• Self-report questionnaire
• Validity

The authors would like to thank Courtney Cavett, Lucinda Chee, Mike Dunn, William Ely, Mike Fitzgerald, Jenna Gilmore, Zach Johnson, Morgan McCulley, Jessie Riess, Sheridan Smith, and Evan White for assisting in data collection and processing.

The authors would like to thank Courtney Cavett, Lucinda Chee, Mike Dunn, William Ely, Mike Fitzgerald, Jenna Gilmore, Zach Johnson, Morgan McCulley, Jessie Riess, Sheridan Smith, and Evan White for assisting in data collection and processing.

Notes

¹ Because the ACS items could be considered ordinal, the CFA analyses were replicated with robust weighted least squares estimation using Mplus version 7.1 (Muthén, du Toit, & Spisic, 1997; Muthén & Muthén, 2013). This resulted in comparably good fit for our two-factor model, χ² = 107.56, p < .001;CFI = .96, TLI = .95, RMSEA = .07. Similarly, fit indices were comparably poor for the Ólafsson and colleagues (2011) model, χ² = 390.44, p < .001;CFI = .86, TLI = .84, RMSEA = .08.