• Journal Article

Test of measurement invariance of the FTND across demographic groups: Assessment, effect size, and prediction of cessation

Citation

Johnson, E., Morgan-Lopez, A., Breslau, N., Hatsukami, D. K., & Bierut, L. J. (2008). Test of measurement invariance of the FTND across demographic groups: Assessment, effect size, and prediction of cessation. Drug and Alcohol Dependence, 93(3), 260-270. DOI: 10.1016/j.drugalcdep.2007.10.001

Abstract

BACKGROUND: Measurement non-invariance of the Fagerstrom test for nicotine dependence (FTND) across demographic groups could significantly bias group comparisons and screening for recruitment into treatment and genetic studies. Here clinically meaningful bias in the FTND across European-American and African-American men and women was assessed by: (1) testing measurement invariance; (2) estimating effect sizes of non-invariance; and (3) assessing impact of adjusting for bias on the association between FTND and cessation. METHODS: European-American and African-American current and former smokers (n = 8301) were identified from a community-based telephone screening of 25,265 individuals from metropolitan Detroit, MI and St. Louis, MO. The FTND was administered to measure current dependence and lifetime dependence when smoking the most. Cessation was measured as having smoked 100 or more cigarettes but not smoked in the past 30 days. RESULTS: Statistically significant measurement non-invariance for the FTND was found and more pronounced for lifetime than current dependence. However, the magnitude of effects appeared negligible. The largest variance in item response explained by measurement non-invariance was 3.1%. Adjusting for measurement non-invariance made no difference in the associations between nicotine dependence and quitting smoking across groups. CONCLUSIONS: Although European-American and African-American men and women often report different scores on the FTND and have different response patterns to items on the FTND, it does not appear that such differences result from meaningful item-level measurement bias.