What Return Rate Is Necessary When Using Mailed Questionnaires?

Techniques have been developed to reduce the number of subjects who do not complete mailed questionnaires, but their use continues to be controversial. Nonresponse bias can influence both the distributions of variables studied and the relationships among the variables. However, nonresponse bias may be slight when we sample homogeneous populations. By using various techniques, nearly 100 percent of a homogeneous population of subjects, who had received a mailed questionnaire designed to evaluate a training impact model, were convinced to complete and return a mail questionnaire. Nonresponse biases on distributions of demographic and model variables and on relationships among model variables were analyzed by cumulative questionnaires returned. The means and standard deviations for demographic and model variables did not vary meaningfully when the first wave of returned questionnaires was compared to other cumulative waves. However, the regression coefficients used to evaluate the training impact model did vary. Some variables in the causal model being evaluated did not become significant until the next to last wave. These data suggest that researchers should continue to make every effort feasible to obtain a high return rate when using mailed questionnaires even when samples are drawn from homogeneous populations.