Speakers
Abstract
Although the Workplace Ostracism Scale (WOS) is a popular instrument for assessing workplace ostracism, research examining its psychometric properties is limited. Reliability is a fundamental quality of a psychometric tool. It is not a property of the test itself but of its test scores, as it depends on the sample’s characteristics, the number of items, and the response format. A good practice to foresee the reliability of a scale’s scores is to quantitatively integrate multiple reliability estimates derived from different administrations. Reliability Generalization (RG) meta-analyses synthesize these estimates, identifying typical reliability levels and factors influencing variability in reliability outcomes.
We performed an RG meta-analysis following the recommendations for conducting and reporting Reliability Generalization Meta-Analyses (REGEMA Checklist). Data were gathered from databases including Psychology and Behavioral Science Collection, MEDLINE, Scopus, Web of Science, Wiley Online Library, SAGE Journals, ScienceDirect, and Google Scholar.
Studies were included if they: (a) used the WOS with an adult sample, (b) reported reliability coefficients for the WOS scores, (c) were published in English in peer-reviewed journals, and (d) used the 10-item, seven-point Likert version of the scale. Moderator variables coded to assess reliability influences included: (a) participants’ gender distribution and mean age, (c) WOS language, (d) country of administration, (e) focus (psychometric vs. applied), and (f) mean and standard deviation of the WOS scores. Our RG meta-analysis focused on Cronbach’s alpha, the most commonly reported reliability coefficient.
Reliability estimates were standardized using Bonnett’s (2002) transformation to normalize their distribution. Between-study heterogeneity was assessed with Cochran’s Q test (p < 0.05) and Higgins’ I², with thresholds of 25%, 50%, and 75% indicating low, moderate, and high heterogeneity, respectively (Higgins et al., 2003). Meta-regressions were then conducted to evaluate the moderation effects of categorical and continuous variables.
From an initial 1,672 records, 44 articles (50 independent samples) were retained, comprising 15,868 participants. The sample was 40.43% male, with a mean age of 32.77 years (SD = 6.90). The WOS was primarily used in English (35.8%) and Chinese (38.3%) in applied research (84.9%). The mean WOS total score was 1.98 (SD = 0.08). The mean Cronbach’s alpha was .93 [CI: .92–.95], ranging from .71 to .99, with significant heterogeneity (Q = 2,634.29, p < .001; I² = 98.15). Moderation analyses revealed that the standard deviation of WOS scores was the only significant moderator (p < .001), explaining approximately 37% of the total variance.
A limitation of this RG meta-analysis is its reliance on a single reliability index, Cronbach’s alpha, which has known drawbacks. Including additional reliability coefficients could have offered a more comprehensive view of WOS score reliability. Furthermore, while variability in WOS scores explained much of the heterogeneity in alpha values, a significant portion remains unexplained. Future RG meta-analyses should consider additional moderators, such as work status, job category, or contract type. However, despite these issues, our findings suggest that the WOS is a reliable measure of workplace ostracism, with adequate internal consistency and no significant discrepancies across demographics, country, or language of administration.
| Poster | The Workplace Ostracism Scale: A Reliability Generalization Meta-Analysis |
|---|---|
| Author | Valenti and Faraci |
| Affiliation | University of Enna "Kore" |
| Keywords | RG meta-analysis; workplace ostracism scale |
