Estimating the Reliability of a Single-Item Measure

Abstract

Single-item measures of employees' attitudes and beliefs are generally discouraged because their (internal consistency) reliability cannot be estimated. This results in the concern that reliability may be unacceptably low, particularly when compared to scales used to measure the same construct. A method for estimating the reliability of a single-item measure is demonstrated on original data that included both a single-item and a multiple-item measure of three constructs, namely, Over-all Job Satisfaction, Perceived Amount of Participation, and Desired Amount of Participation in decision-making. The average minimum estimated reliability for these single-item measures is .57; however, a realistic yet conservative estimate of their likely minimum reliability is at least .70.