Identifying Culturally Sensitive Physical Activities Using DIF Analysis

Document Type

Conference Proceeding

Publication Date




Promoting physical activity (PA) among traditionally disadvantaged PA subgroups as a means to reduce PA disparities is very important to help with the elimination of health disparities. However, the success of PA promotion and intervention programs for these subgroups has been very limited. Lack of culturally sensitive PA in the promotion and intervention is likely one of the reasons.

PURPOSE: To identify subgroup preferred PA using differential item functioning (DIF) analysis.

METHODS: A sub-unweighted sample of 3,083 (males = 1,511 and females = 1,572) from the National Health and Nutrition Examination Survey (NHANES 2003-2004) PA questionnaire data was used for the analyses. Using Mantel-Haenszel and SIBTEST DIF methods, 33 specific items related to leisure time moderate and vigorous PA (MVPA) were analyzed for DIF across race/ethnicity, gender, education, income and age groups.

RESULTS: Approximately 12% to 39% of MVPA questions used in NHANES PA questionnaire were identified as large DIF items according to widely accepted DIF identification criterions in educational and psychological measurement practice. The presence of DIF in this study implies a subgroup is more comfortable with a specific activity due to likely its social and cultural context; therefore, persons in that group report a preference to participate in that activity when engaging in the same amount of total leisure time MVPA as another group. For example, basketball and dance favored non-Hispanic Blacks while golf and hiking favored non-Hispanic Whites; Dance, hiking and soccer favored Hispanics while bicycling, golf, swimming, and walking favored non-Hispanic Whites; Aerobics, dance, stretching and walking favored females while basketball, fishing, golf, running, soccer, weightlifting and hunting favored males; Jogging, treadmill and yoga favored persons with high education levels; Golf and treadmill favored persons with high incomes; and finally basketball, dance, jogging, running, and weightlifting favored adults while walking and golf favored older adults.

CONCLUSIONS: DIF methods were able to identify subgroup preferred PA, which provides information to allow a deeper understanding of a subgroup's PA pattern which in turn can help design culturally sensitive, targeted interventions for disadvantaged PA subgroups.