Participation bias in a sexuality survey: psychological and behavioural characteristics of responders and non-responders.
BACKGROUND: Few studies of sexual attitudes and behaviour have quantified the direction and magnitude of participation bias, primarily because information on non-responders is difficult to obtain in cross-sectional surveys. METHOD: Australian adult twins (n = 9112) aged 17-52 years enrolled in a national, longitudinal research register were asked to participate in a postal survey concerning their sexual behaviour and attitudes. Individual consent was determined by separate return of a consent form; 27% explicitly refused, 19% initially agreed to receive a questionnaire, but subsequently did not return consent forms and 52% explicitly consented. Participation data were matched to social, psychological and behavioural information in a longitudinal data set. RESULTS: People who explicitly consented had higher levels of education, attended church less often, had less conservative sexual attitudes and voting preferences, were more likely to smoke cigarettes and drank alcohol more often than people who explicitly refused. On standard personality scales, responders were more novelty-seeking and reward-dependent and less harm-avoidant than refusers. Structured psychiatric telephone interview data from 3674 individuals showed that, compared to refusers, responders had higher lifetime prevalence of major depression, alcohol dependence and childhood conduct disorder and also reported an earlier age at first sexual intercourse and higher rates of sexual abuse. In general, those who had initially agreed to receive the sex questionnaire but were subsequently lost were more similar to consenters than to refusers. CONCLUSIONS: Effect sizes on most measures were small. The broad profile suggests that postal surveys of sexual attitudes and behaviour may overestimate sexual liberalism, activity and adversity, although this bias should not seriously compromise population estimates.