Characteristics of non-respondents in a US national longitudinal survey on drinking and intimate partner violence.

Aims: To identify non-respondents' characteristics in a longitudinal household survey of intimate partner violence and drinking among United States couples.

Design: Multistage probability sampling procedure with over-samples from black and Hispanic populations. A total of 1635 couples 18 years of age or older were interviewed at baseline (1995) for a response rate of 85%. Five years later, 1392 respondents were re-interviewed, with a response rate of 72%.

Measurements: Sociodemographic variables, drinking volume and pattern, drug use, intimate partner violence, exposure to parental violence and childhood physical abuse.

Findings: Among men, those who are younger, those who did not graduate from high school, those who are unemployed and those who are drinkers are more likely to be among the non-respondents. Among women, those who are 40-49 years of age are two times more likely than those aged 50 years or older to be among non-respondents. Also, women who reported experiencing abuse during their childhood are less likely to be among non-respondents compared with those who were not victimized. Gender-specific regression models accounted for only 5% of the variance in the response status variable, even after the inclusion of a comprehensive range of predictors in the model.

Conclusions: A low proportion of the variance in survey participation was explained by the variables included in the analysis, suggesting that sociodemographic, drinking-related and violence-related variables might not be the main factors determining participation in the survey.