: http://orcid.org/0000-0002-3721-020X

Date of Award


Document Type


Degree Name

MA in Psychology



First Advisor

Raymond DiGiuseppe

Second Advisor

William Chaplin


The present study examined two abbreviated versions of the Attitudes and Beliefs Scale-2 (ABS-2) to compare their factor structure and ability to achieve model fit to the data. The original scale, a measure of irrational and rational beliefs as defined by REBT theory, was designed with 72 items reflecting irrational and rational beliefs and each involving one of four cognitive processes and one of three content areas. The ABS-2 had been criticized for its length and the inconsistency of findings regarding its factor structure. Two groups of researchers independently created short forms of the ABS-2 using 24 of the original items. One scale used the items with the highest factor loadings, while the other also prioritized maintaining balance across all dimensions. To also explore the effects of using different estimators, the authors ran Confirmatory Factor Analyses (CFAs) for each short form twice, once using the Maximum Likelihood Robust (MLR) estimator and once using Diagonally Weighted Least Squares (DWLS). The sample consisted of over 1500 participants that included university students, psychotherapy outpatients, and individuals in a drug rehabilitation program. Results showed that both scales yielded virtually equal and excellent fit indices when using the DWLS estimator but not when using MLR. The model with the best fit was an eight-factor bifactor model with factors for the irrational and rational cognitive processes and a general factor. Two other models also yielded especially excellent fit, including a two-factor bifactor model for irrationality and rationality as well as a second-order model with items loading on either one of the four irrational cognitive processes and then a second-order irrationality factor or on one of the four rational cognitive processes and a second-order rationality factor. Ultimately, the results suggest that the assessment can provide meaningful subscales for scores of the total, irrationality, rationality, cognitive processes, and content domains. Additionally, the findings highlight the importance of critically considering one’s data and selecting an appropriate estimator as opposed to relying on default settings. Implications for the assessment of irrational and rational beliefs, furthering REBT research, and targeting treatment to client presentation across the three dimensions are discussed.

Available for download on Friday, September 09, 2022

Included in

Psychology Commons