Rasch analysis reveals multidimensionality in the public speaking anxiety scale.

Purpose: There is a lack of well-validated self-report measures to assess public speaking anxiety. This study explored the psychometric properties of the Public Speaking Anxiety Scale (PSAS). Methods: Seventy-two adults completed the PSAS as part of the baseline screening procedure of a randomized controlled trial. Rasch analysis was used to assess the scale's response category functioning, precision, targeting, unidimensionality, and differential item functioning. Construct validity was assessed using classical test theory methods. Results: While thresholds were ordered and no systematic bias in responses for age, gender, or screen failure was found, the PSAS demonstrated evidence of multidimensionality (variance by first factor = 39.7%, eigenvalue of first contrast = 2.76). Multidimensionality was resolved after splitting the scale into two discrete subscales: Emotional and Physiological. Three misfitting items (i.e. item 5 from Emotional, items 6 and 14 from Physiological) were removed. Scale precision and targeting remained suboptimal after subscale split and removal of misfitting items (PSI = 1.41, PR = 0.67 for Emotional; PSI = 1.49, PR = 0.69 for Physiological). Conclusion: The PSAS demonstrated adequate convergent validity. Psychometric properties of the PSAS after Rasch-guided modifications were overall promising. Further studies are needed to confirm our results. [ABSTRACT FROM AUTHOR]