The Effects of Item Grouping on the Reliability and Scale Scores of an Affective Measure

Abstract

Instrument developers find conflicting advice in the literature about either grouping items together that purport to measure each construct or to randomly intersperse items throughout the instrument. This paper examines the effects of grouping and random item placement on the alpha internal-consistency reliabilities of the scales and scale means on an affective measure. Out of a total of 992 randomly selected principals, 719 responded to the Administrative Attitudes Toward Unsatisfactory Teachers Survey; 350 received a grouped-item version of the instrument and 369, a random-item version. While there were no significant differences in scale means between the two methods of item placement, there was an effect on the scale alpha reliabilities.