Department of Psychology
Master of Arts
Developers of multidimensional tests must decide whether to group items measuring the same construct together or list the items randomly. The effects of item grouping are not well established by the few existing studies. This study examined the effects of item grouping on the psychometric properties of a personality inventory measuring conscientiousness and extraversion. Two hundred and four undergraduate students were administered the test with the items listed in either a grouped or ungrouped format. Internal consistency reliability was estimated with coefficient alpha. Grouping test items failed to increase scale reliability (.80 for ungrouped versus .76 for grouped) or decrease scale intercorrelation (.31 for grouped items versus .27 for ungrouped items). The differences between these correlations were not significant. Reasons for the findings are offered and recommendations are given for future studies.
Hunt, Julian, "An Examination of the Effects of Item Grouping on Internal Consistency and Scale Intercorrelation" (2007). Masters Theses & Specialist Projects. Paper 500.