Date of Award

1-8-2021

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Educational Policy Studies

First Advisor

Hongli Li

Second Advisor

T. Chris Oshima

Third Advisor

Sarah Carlson

Fourth Advisor

Randy W. Kamphaus

Abstract

Achievement motivation is a well-documented predictor of a variety of positive student outcomes. However, researchers have also found threats to fairness and measurement scale comparability in motivation items, including group differences in response scale use and response styles. As such, the measurement comparability of achievement motivation items was evaluated before and after using anchoring vignettes to account for the effect of group-specific response scale use as a source of differential item functioning (DIF) across gender and ethnicity. Within a combined item response theory/ordinal logistic regression DIF framework, gender DIF was assessed using pairwise comparisons and ethnicity DIF was tested using both multiple-group DIF with a common base group as the reference group and all possible pairwise comparisons.

Overall, using the vignettes changed both the form of DIF within items and the pattern of DIF between groups across items. Results indicated the presence of DIF between genders, but the DIF was unrelated to group differences in response scale use. Across ethnic groups, Black/African American students and Asian students demonstrated group-specific response scale use. When groups showed response tendencies, accounting for such scale use with the vignettes had a greater effect on reducing DIF in base group comparisons than in pairwise comparisons. Despite that DIF was identified in multiple items, the magnitude of all DIF was negligible and had little practical implication. Therefore, achievement motivation items appeared to demonstrate measurement comparability. As sources of DIF often go unidentified, a contribution of this study was the novel use of anchoring vignettes to account for group differences in response scale use as the source of DIF and to clarify the effect of those differences on measurement scale comparability and DIF.

DOI

https://doi.org/10.57709/20506213

File Upload Confirmation

1

Share

COinS