Date of Award
Doctor of Psychology (PsyD)
Graduate Department of Clinical Psychology
Dr. Rodger Bufford
The purpose of this study was to determine if any of the twenty items on the Spiritual Well-Being Scale (SWBS) show differential item functioning (DIF) between gender and item response. Archival data were obtained on two groups: (a) 331 members of the Christian Association for Psychological Studies (CAPS, USA), 197 males and 134 females, average age 44.5 years (Adams, 1993; Stratton, 1992); and (b) 392 incoming freshmen students at a liberal arts college in the Pacific Northwest, 148 males and 244 females, average age 18 years (Foster & LaForce, 1991). For each sample, partial correlations were computed for gender-related response differences (DIF) on each item with total SWBS score, Religious Well-Being Subscale (RWBS) score, and Existential Well-Being Subscale (EWBS) score held constant.
For these two populations, evidence was found for small effects (each item that showed significance accounted for less than two percent of the total variance) indicating DIF in six items [Item 6 (college population) and Item 18 (CAPS population) on the SWBS; Item 7 (college population) and Item 17 (CAPS population) on the RWBS; and Item 6 (college population) and Item 20 (college population) on the EWBS]. Due to the number of significance tests (80 at ~ < .OS), four items could be expected by chance to show a response difference.
After post-hoc partial correlations to remove variance due to other factors known to be correlated with gender in these samples such as licensure, years of professional service, and age in the/CAPS population; and parents' income, mothers' education, fathers' education, theological classification, high school GPA, and age in the college population, Item 18 in the CAPS sample and Item 6 in the college sample still showed DIF between gender groups. The small magnitude of the relationships (less than two percent of the total variance) suggests these significant DIF scores have little practical importance for this sample.
Further post-hoc analysis of those items no longer showing significance revealed that the factors that appeared to account for original significance were a combination of all six additional items on Item 6 (SWBS, college population}; years of professional practice for Item 7 (CAPS population); parents' income for Item 17 (college population); and theological classification (conservative or liberal) for Item 20 (college population).
In comparing mean scores for SWBS, RWBS, and EWBS in the two samples, the male mean scores were .2 to .6 points higher than the female mean scores in the CAPS sample. In the college sample, the female mean scores were 2.1 to 5.0 points higher than the male mean scores. Developmentally, these two samples were apparently at different stages (middle adulthood, average age 44.5 years, in the CAPS sample, and young adulthood, average age 18 years, in the GFC sample) which could account for the mean differences. Also, the college sample may have been identity foreclosed while the CAPS sample was more likely to be identity achieved.
Kellums, Karen J., "Gender Analysis of the Spiritual Well-Being Scale" (1995). Doctor of Psychology (PsyD). 171.