Five questionnaire forms containing 61 items specifying potential inservice topics for public school teachers were sent to a stratified random sample of Indiana public school administrators and curriculum supervisors. The five forms differed in that, for two forms, the items were ungrouped and appeared in different orders; and, for three forms, the items appeared in different orders and were presented within different labeled item groupings. A 68 percent return rate provided 1,468 useable questionnaires for the data analysis. An item-order effect was demonstrated by comparing Pearson product-moment correlations, covariance matrices, residuals resulting from a fixed factor structure, and factor loading patterns for the various questionnaire forms. The item-order effect appeared to be more pronounced for adjacent item pairs than for item pairs separated by one or two items. Only a marginal effect was demonstrated for item grouping with captions. It is probable that low-inference type items factor structures obtained from survey data are dependent upon item order, but only minimally affected by item grouping. (Author)