The computer-based Outcome Probability Task uses facilitated, anticipatory mental imagery to measure the tendency to overestimate the likelihood of negative evaluation in imagined social interactions. This is done by prompting respondents to imagine they will soon experience what is shown in a social image, then collecting ratings of expected embarrassment, humiliation, or negative judgment (0-100%). Participants (n = 283) were predominantly students, who reported diverse sociodemographic identities and endorsed varying psychological characteristics. A confirmatory factor analysis was conducted to assess the stability of a three-factor model examining performance situations, probable judgment, and social gatherings. As hypothesized, the three-factor model was superior to an alternative bifactor model based on evaluation of fit metrics and factor loadings. Findings provide additional support that the Outcome Probability Task has potential to advance the multi-method assessment of social outcome probability bias subtypes. This could have positive implications for the innovation of transdiagnostic assessments and treatments.