Measurement of attitudes and behaviors in public health surveys

We divided 900 general practitioners into two groups: one group received a questionnaire measuring both attitudes and behaviors toward preventive aspects of medical care, while the other received two separate questionnaires--one measuring behaviors, and one measuring attitudes. Response rates of the two groups were similar. Respondents in the first group retrieved their answer to the attitudinal items when responding to behavioral items, increasing the correlations between attitudes and behaviors by an average of .147 as compared to the second group; hence the second procedure was preferable.