Robert Baker

Publication Date


Advisor(s) - Committee Chair

Roy Mendel, John O’Connor, Elizabeth Erffemeyer

Degree Program

Department of Psychology

Degree Type

Master of Arts


The purpose of this study was to compare halo levels between a set of behaviorally anchored rating scales (BARS) developed by conventional methodology with a set of BARS developed by factor analytic means. The conventionally-developed BARS were developed at the University of California-Berkeley by Department of Psychology faculty members in conjunction with undergraduate psychology students. The factor analytically-derived BARS were developed during the course of the research reported here. In Phase 1 of the present study, undergraduate psychology students rated their instructors on a 1-7 Likert scale using the individual anchors which comprised the conventional BARS. A factor analysis of these ratings was used to form a modified set of BARS. In Phase 2, a new group of undergraduate psychology students was asked to rate their instructors using either the conventional (Harari-Zedeck) BARS or the newly developed modified BARS in order to test for differences in the halo levels for each form of BARS. Halo was measured by the strength of the dimension intercorrelations in each set of scales. It was hypothesized that the factor analytic approach would yield a set of scales with reduced levels of illusory halo. Results indicated, however, that the conventionally-developed and factor analytically-developed BARS were not significantly different in their halo levels. A post hoc reliability check revealed that the conventionally-developed BARS had significantly less interrater reliability than the factor analytically-derived BARS, which may partially account for the lack of difference in the scales’ halo levels.


Psychology | Social and Behavioral Sciences

Included in

Psychology Commons