Measurement properties of web-based attentional bias in depression with cursor-directed aperture in online studies

Negative attention bias is thought to maintain depression; however, past research has been hampered by psychometrically poor behavioral tasks. The present study examines the psychometrics of a free-viewing attention task that uses a 4 x 4 matrix of dysphoric and neutral facial expressions with mouse-tracking – movement of an aperture via a computer mouse that allows for clear viewing within the aperture of otherwise blurred stimuli. We assessed internal consistency, factor structure, and unidimensionality via percent uncontaminated correlations (PUC), explained common variance (ECV), and omega hierarchical (ωh) of trial-level scan time for dysphoric stimuli across four online samples: college students (Sample 1, N = 61; Sample 2, N = 114; Sample 4, N = 170) and an adult clinical sample (Sample 3, N = 181). Across all four samples, internal consistency was strong (alpha and omega ranged from .80 to .94) and 57%, 73%, 100%, and 87% of trials loaded on a single factor. Evidence for unidimensionality was modest in the college student samples (PUCs = .76, .75, and .78, ECVs = .25, .34, .and 46, ωh = .32, .40, and .52). In the clinical sample, unidimensionality (PUC = .72, ECV = .58, ωh = .69), within- and between-person test re-test reliability across 5 weekly assessments were strong (omegas > .90). Construct validity was also observed as higher depression severity was associated with more time spent viewing dysphoric stimuli. This behavioral task, when paired with mouse tracking, provides a psychometrically sound alternative for the assessment of attentional bias for dysphoric stimuli that could be used in large-scale, online studies.