Author ORCID Identifier

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Page Anderson

Second Advisor

Erin Tone

Third Advisor

Michael Beran

Fourth Advisor

Lindsey Cohen


People with social anxiety disorder tend to expect that the likelihood of social embarrassment or negative judgment is much higher when anticipating a social encounter than people without the disorder (Lucock & Salkovskis, 1988). This outcome probability bias is theorized to contribute to the development and maintenance of symptoms (Clark, 2001; Heimberg et al., 2010). The causal role of outcome probability bias in social anxiety disorder is unknown, however, because all known studies have relied exclusively on paper-and-pencil questionnaires to operationalize this construct. The present study aimed to assess the reliability, validity, and factor structure of a novel computer task designed to measure outcome probability bias in social anxiety. Important components of this task included incorporation of social images and assessment of outcome probability bias resulting from both automatic and controlled levels of processing. Results from a pilot study indicated the images in the task were appropriate but modifications were warranted for assessment of outcome probability bias at an automatic level of processing. Results from the main study indicated good to excellent internal consistency among social images (α = 0.86 – 0.96). Correlations were consistent with the nomological network. Outcome probability bias task ratings were higher in response to social items compared to nonsocial items and among people with social anxiety symptom scores above a clinical cutoff compared to people with scores below the cutoff (all p values < 0.05). Social images in the task also predicted self-reported symptoms of social anxiety, self-reported safety behaviors during a behavioral avoidance test, and subjective distress during a behavioral avoidance test (all p values < 0.05). The task did not, however, significantly predict performance on a measure of behavioral avoidance. Exploratory factor analyses revealed a tripartite factor structure. These findings offer preliminary support for the reliability, convergent validity, discriminant validity, construct validity, and criterion validity of the outcome probability bias task. This task may potentially be used in future research for multimodal assessment and experimental manipulation of outcome probability bias in social anxiety.

File Upload Confirmation


Available for download on Friday, May 14, 2021