ABSTRACT
The best way to estimate non response bias and its potential effect in epidemiological research to know the characteristics of non respondents. The purpose of this study is to analyze what characteristics of those who refused to answer telephone and personal interviews in a follow-up study can predict non-response. As part of the Yale Health and Ageing Project, a follow-up study initiated in 1982 in the USA in a population sample of 2.806 over 65 years of age, the influence of several factors in non-response rate for those who refused to answer the interview was analyzed. A total of 21 variables related to demographic, health, and social characteristics were analyzed. The dependent variable was defined as response/non-response to the interviews. The results did not show significant statistical associations. The only factors that showed specific patterns were housing (in both types of interviews), and education level, cognitive status, and physical disability in personal interview in women, with p values for the Odds Ratios between 0.05 and 0.07. In conclusion, most results of this study do not support the findings already published. A disadvantaged sociodemographic situation was not a predictor factor of non-response, except for the education level of women in personal interviews. In relation to health status only a poor cognitive status, as it has been already reported, seems to induce refusal of women to personal interviews.