Building a construct-valid battery of performance and self-report indicators of sustained attention consistency

Abstract

Previous work has argued that the ability to sustain attention consistency can be best modeled as the individual-difference covariation in objective performance-based measures (e.g., reaction-time [RT] variability; accuracy) and self-report measures of task-unrelated thought (TUT). Latent variable studies demonstrate that a general, higher-order attention consistency factor correlates more strongly with nomological network constructs than do either lower-order, measurement-specific factors. The present study aimed to replicate and extend this measurement approach by building a construct-valid battery of sustained attention consistency tasks and testing associations with the conative factors of task interest and success motivation. We analyzed data from 402 subjects who completed a battery of seven attention-consistency functions and found that the hierarchical model provided an adequate fit to the data. Further, attention-consistency associations with motivation and interest, while evident with the lower-order factors, were again stronger with the general higher-order factor (and each conative factor predicted unique variance in general attention consistency in structural regression models). We also refined our task battery by removing poor-performing indicators and demonstrated similar patterns of correlations among the attention and conative factors. We suggest that studies examining attention consistency should use a combination of performance and self-report indicators to capture its individual-differences variation in the most construct valid way. We finally provide recommendations on which tasks and measures might be most useful when measuring sustained attention consistency in future research.