摘要：This research investigated whether the same inferences would have been made from a survey without callbacks as with them. Comparison of the characteristics of respondents contacted on the first call with those contacted on subsequent calls over a range of behavioural and demographic variables confirmed the respondents who are not at home at an interviewer's first attempt are more likely to be in regular, paid employment, to be younger, male, better educated and to earn higher incomes. They also tend to watch less television, listen to the radio less and attend the cinema less often. However, for the behavioural and demographic variables in the study, none of the differences between estimates based on the first call sample and the total sample are significant and, for many practical purposes, the differences which do exist could be regarded as irrelevant. But although these differences may be small, they may be important, depending on the decisions which are based on them (for example, small differences are important when survey proportions are applied to population figures to determine the cost-per-thousand estimates which influence the purchase of media time and space). Consequently, the answer to the question, "Are callbacks a waste of time.", depends ultimately on how accurate the results of a survey need to be. This in turn depends on the nature and purpose of the study and on how the results will be used
关键词：callbacks; interviewing; survey research; non response bias
