We used the multiple indicators, multiple causes (MIMIC) model framework to estimate measurement bias in the CES-D responses of participants in the New Haven Established Populations for the Epidemiologic Studies of the Elderly study (N = 2,340).
Measurement bias attributable to race was significant for the following two CES-D items: people ȁc;are unfriendlyȁd; and ȁc;dislike meȁd;. The proportional odds of a higher-category response by blacks relative to whites on these items were 2.35 (95%confidence interval [CI]: 1.65, 3.36) and 3.11 (95%CI: 2.04, 4.76), respectively. The proportional odds were higher among women (2.03 [95%CI: 1.35, 3.06]) relative to men for the CES-D item ȁc;crying.ȁd;
Our findings confirm that three items on the CES-D show strong evidence of item response bias. The MIMIC model is preferable to the Mantel–Haenszel approach because it conditions on a latent variable, although the effect estimates can also be interpreted using a proportional odds framework.