Search

This study examines how low reliability of a measure affects its estimation and statistical inference and what happens when the measure is collinear with another variable with higher reliability as predictors in a statistical model. First, we provide a series of Monte-Carlo simulations to show that (1) the effect of an unreliable measure tends to be underestimated, and (2) its underestimation becomes more serious while a more reliable collinear predictor is overestimated in the same model. Then, we examine Male Attitudes Regarding Sexual Health (MARSH), a survey of 2,173 men in the United States, 40 years of age and older broken by racial/ethnic subgroups: Whites, African-Americans and Hispanics. We find in MARSH that the sum of IPSS-7 items for lower urinary tract symptoms (LUTS), as a risk factor for erectile dysfunction (ED), is considerably underestimated for Hispanics when another risk factor, the sum of CES-D8 items for depressive symptoms, is introduced. We infer that IPSS-7 could have been more unreliably measured among Hispanics than among Whites or African-Americans, noting comparably high correlations between CES-D8 and LUTS. LUTS questions may be particularly sensitive for Hispanic males, who are known for machismo, and may as well belong to T-ACASI module without an interviewer, rather than to CATI, in future survey. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]

An emerging body of research shows that perceived discrimination adversely influences the mental health of minority populations, but is it also deleterious to physical health? If yes, can marriage buffer the effect of perceived discrimination on physical health? We address these questions with data from Puerto Rican and Mexican American residents of Chicago. Multivariate regression analyses reveal that perceived discrimination is associated with a greater number of physical health problems for both Puerto Rican and Mexican Americans. Marriage has health-protective effects, however, for Mexican Americans only. In addition, an interaction effect between marital status and perceived discrimination was observed for Mexican Americans: married Mexican Americans who reported higher perceived discrimination had fewer physical health problems than their unmarried counterparts. The findings reveal that perceived discrimination has deleterious effects on physical health of both Puerto Rican and Mexican Americans, but that the buffering effect of marriage exists for Mexican Americans only. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]