(WITHDRAWN) Exploring Gender Differences in Everyday Discrimination: Item Response Theory Approach

Background

The Everyday Discrimination Scale (EDS) was developed by Williams, Yu, Jackson and Anderson (1997) to measure perceptions of subtle, but frequent and harmful forms of discrimination. The EDS has been extensively used to answer questions about the relationship between discrimination and physical and mental health. Within some of these studies, gender differences in frequency of everyday discrimination have been found. However, we limited understanding about gender differences perceived discrimination. Previous studies have found that females regularly experience discrimination, particularly within the job market; yet, when controlling for gender differences in perceived discrimination, most evidence suggests that females report less frequent experiences. Due to the widespread use of the EDS in studies on health and well-being, an important question in this area is whether females respond differently to particular questions on the EDS. This study explored whether there are gender differences in responses to EDS questions and the results provide information on gender differences that can aid in future work.

Methods

This study used secondary data from the 2017 Panel Study of Income Dynamics Transition into Adulthood Supplement (TAS) conducted by the Survey Research Center at the University of Michigan. The sample included 2,158 young adults between the ages of 17-28. To answer the question, Item Response Theory with measurement equivalence invariance testing was conducted using the R Lavaan package. Invariance testing is relevant to assess group differences and to investigate whether such difference relate to the way the respondents answer each question. Configural, metric, and threshold invariance models were specified using the seven EDS items available within the TAS dataset. The grouping variable in this study was sex, where 1 indicated female and 0 male.

Results

T-test results suggested no gender differences in EDS scores for items 1 through 3 or item 6. However, there were statistically significant gender differences on scores for items 4 and 5. Similar results were found during invariance testing. CFI, TLI and RMSEA statistics all showed satisfactory fit for both configural and metric invariance tests. Differences were found at the scalar invariance level. The partial structural invariance model showed the biggest difference for items 5 and 4. Item 5, which asked how often one was treated as dishonest, was endorsed more often by males. Likewise, males also reported greater frequency of discrimination related to item 4, how often others acted afraid of them.

Conclusions

The current study centered on assessing gender differences in acts of everyday discrimination. Specifically, this study aimed to identify gender differences in the types of discrimination experienced. Social dominance theory asserts that men, particularly those from subordinate groups, become targets for more overt acts of discrimination by those within dominant group. These findings showed that males report greater frequency of acts being treated as dishonest or having others fear them. Future work could explore gender differences in everyday discrimination within the context of racial or ethnic group.