Psychometric Evaluation of the Posttrauma Risky Behaviors Questionnaire: Item Response Theory Analyses

Abstract

The Posttrauma Risky Behaviors Questionnaire (PRBQ) assesses extent of engagement in posttrauma reckless and self-destructive behaviors (RSDBs). Given PRBQ’s recent development with limited psychometric investigations, we used item response theory to examine (a) item analysis, (b) person fit, and (c) differential item functioning (DIF) across gender-based groups and two different samples. One sample included 464 participants reporting potentially traumatic experiences (Mechanical Turk [MTurk], recruited online), and the other sample included 171 trauma-exposed women reporting current intimate partner violence and substance use (recruited in-person). All PRBQ items contributed to the RSDB scale, and all PRBQ items and the PRBQ scale provided maximum information for high levels of the RSDB latent trait. Seven and 11 items were conceptualized as low information items in the MTurk and intimate partner violence samples, respectively. Eight MTurk participants’ responses did not fit the overall pattern of responses as expected. Seven items were flagged for DIF between the two samples, and eight items were flagged for DIF between men and women in the MTurk sample. However, all effect sizes were <8%. Conclusively, results suggest good psychometric properties for the PRBQ and support its use to compare RSDBs across different samples and gender-based groups.

Relations between PTSD symptoms and reckless behaviors among military veteran students: Moderating roles of negative and positive life experiences.

“…For the specific analyses examining the number of delinquent behaviors, we excluded the PRBQ Item 7 examining illegal behaviors due to its overlap with the delinquency construct when computing the total PRBQ score. The PRBQ scores have good psychometrics (Contractor et al, 2020, 2021; Natesan Batley et al, 2022), and in the current study, the Cronbach’s α was .78 for the total score, which excluded Item 7.…”

Revised Chinese resident health literacy scale for the older adults in China: simplified version and initial validity testing

“…Recently, IRT was more and more applied to assess and revise the scales in public health or clinical studies (43). Different from the previous simplified studies just focused on the difficulty and discrimination when using IRT model (31,(62)(63)(64), our findings also integrated the equivalence and information of items into the criteria of remaining items in order to obtain a relative accuracy of measurement of health literacy for older adults. The previous study by Shen et al reported the overall reliability of original version was 0.95 among the population aged 15-69 years old in China (16).…”

“…Where was the response to item in the scale; was the discrimination parameter; was the difficulty parameter, which indicated the point where an individual would have a 50% chance of endorsing item ; and is the ability value being measured. The basic assumptions of IRT included unidimensionality, local independence, and monotonicity ( 25 – 29 ), which were assessed, respectively, by the eigenvalue examination in factor analysis ( 30 ), Yen’s Q3 statistic ( 29 ), splines modeling via flexible IRT models and Pearson χ 2 statistics ( 31 ) in this study.…”

“…Item response theory (IRT) was a complex approach that attempted to explain the relationship between latent traits (unobservable characteristic or attribute) and their manifestations (i.e., observed outcomes, responses or performance) (23)(24)(25)(26)(27)(28)(29)(30)(31). Recently, more and more researchers used IRT models to analyze the latent properties of the scales and furtherly revise or simplify the scales (32).…”

The cyborg method: A method to identify fraudulent responses from crowdsourced data

“…A knowledge-check item is one that asks for information that would only be known by a member of a target population (e.g., specific military information for a veteran-specific sample) (Weiss et al, 2021). This approach has been used to identify potential fraudulent responses as well as those who are inattentive (Natesan Batley et al, 2021;Shapiro et al, 2013;van Stolk-Cooke et al, 2018). This method also has limitations.…”