Single-Item Reliability: A Replication and Extension

Abstract

The reliability of a single-item measure of student-rated college teaching effectiveness was estimated with two different methods and at two levels of analysis. The two methods are the correction for attenuation formula and factor analysis. The two levels of analysis are the group level (10, 682 classes) and the individual level (323, 262 students). Reliability estimates were higher using factor analysis (.88) than the correction for attenuation formula (.64), and they were higher using group-level data (.82) than individual-level data (.70). Based on the assumptions and limitations of each method used, the authors conclude that a minimum estimate of .80 for single-item reliability is reasonable for group-level data. The authors reaffirm a minimum reliability estimate of .70 for individual-level data, as previously concluded by Wanous, Reichers, and Hudy, who estimated single-item reliability for measures of overall job satisfaction using individual-level data.