Some effects of position on opinion survey items

Tested the effect of putting opinion survey items in different positions in a questionnnaire. Equivalent samples of 284 and 281 employees in an industrial corporation completed similar questionnaires in which the placement of 46 Likert-type items were reversed. Respondents answered with less extreme responses and were slightly more likely to omit replies when items were placed later in the questionnaire. Findings suggest that comparisons of responses to identical items used in different surveys may be misleading if they appeared in different position or context.