Gender differences in the reliability of the EPQ? A bootstrapping approach

Abstract

Reliability indicates the degree of stability or homogeneity of a measurement, but also places an upper limit on the degree of association with other variables. Various methods are available to estimate the reliability of a measurement scale. However, an issue that has rarely been examined is that the reliability of a measurement, as estimated by coefficient alpha, may differ between groups. If a measurement has a different reliability for groups within a sample, spurious moderator effects may occur. The present study examines the reliability of the four subscales of a widely used psychological measurement instrument, the Eysenck Personality Questionnaire-Revised (EPQ-R), across gender. A bootstrapping methodology is employed which allows empirically derived standard errors to be calculated, and therefore tests of significance of difference to be computed. No significant differences were found in the reliability of the EPQ-R across sexes.