Nonresponse in a Follow-Up to a Representative Telephone Survey of Adult Drinkers

Objective: Examined predictors of nonresponse among respondents who agreed to receive a follow-up questionnaire on alcohol use after participating in a representative telephone survey, and among respondents who did and did not return the follow-up questionnaire. Method: A total of 2,072 (52.2% female) respondents to a representative monthly telephone survey were assessed on sociodemographic variables and alcohol use. Respondents were asked whether they would be willing to fill out an additional mailed questionnaire on alcohol use and attitudes toward drinking. Almost half (n = 956; 46%) of respondents agreed to participate in the follow-up survey; 430 (45%) of those individuals completed and returned the questionnaire. Results: Agreement to receive the follow-up questionnaire was unrelated to alcohol use. Regarding gender, men were 1.42 times more likely than women to exhibit nonresponse in returning the follow-up questionnaire (95% CI: 1.08-1.42). After adjusting for the impact of demographic factors, respondents who consumed alcohol at least once per week were 1.43 times more likely than respondents who drank less frequently to exhibit nonresponse in returning the questionnaire (95% CI: 1.05-1.93). Respondents who consumed five or more standard drinks at least once per week were 1.83 times more likely to exhibit nonresponse in returning the questionnaire, compared with respondents who engaged in heavy drinking less frequently (95% CI: 1.15-2.92). Conclusions: Mailout questionnaires following a representative telephone survey may bias samples toward obtaining fewer men, fewer weekly drinkers and fewer heavier drinkers. Although the magnitude of these biases were relatively small, epidemiological studies on alcohol use may wish to oversample men and heavier drinkers in follow-up studies recruiting from population surveys. (J. Stud. Alcohol 62: 257-261, 2001)
ALCOHOL RESEARCHERS routinely use survey methods to assess drinking patterns and problems (e.g., Dawson, 1998; Harford et al., 1991; Hilton, 1991; Poulin et al., 1997; Rehm and Sempos, 1995). With [...]