Distinguishing Trait And Method Variance In Multitrait-Multimethod Matrices: A Reply To Golding

Area of Science:

• Psychometrics
• Statistical Analysis

Background:

• The Campbell and Fiske (1959) framework established criteria for convergent and discriminant validity.
• Evaluating multitrait-multimethod matrices is crucial for construct validation in psychological research.

Purpose of the Study:

• To evaluate the adequacy of proposed techniques for analyzing multitrait-multimethod matrices.
• To identify limitations in existing methods for assessing convergent and discriminant validity.

Main Methods:

• Critical analysis of Golding and Seidman's (1974) proposal for multitrait-multimethod matrix evaluation.
• Examination of Golding's (1976) conjecture regarding trait and method variance in factor analysis.

Main Results:

• Golding and Seidman's method was found inadequate due to inconsistencies with Campbell-Fiske requirements, confounding variances, and lack of clear variance isolation techniques.
• Golding's conjecture was based on redefinition of terms and lacked empirical verification.
• The proposed methods yielded problematic results when applied to matrices meeting validity criteria.

Conclusions:

• Existing methods for evaluating multitrait-multimethod matrices require revision.
• Empirical evaluation of trait and method variance can proceed through inductive generalization and theory, not solely relying on nomological networks.