Understanding the Multitrait-Multimethod Design

The multitrait-multimethod (MMMT) design is a key strategy in psychology and research methodology used to assess the construct validity of measures. Proposed by Campbell and Fiske (1959), it involves measuring multiple traits using multiple methods simultaneously.

Evidence Provided by MMMT Design

The primary goal of the MMMT design is to provide specific types of validity evidence:

• Convergent Evidence Validity: This is demonstrated when different methods used to measure the same trait yield similar results (i.e., high correlations). The MMMT design helps show this convergence.
• Discriminant Evidence Validity: This is shown when measures of different traits using different methods yield low correlations. The MMMT design helps confirm that a measure is unique to its intended trait and not overly correlated with other traits.

Evidence NOT Provided by MMMT Design

The MMMT design is fundamentally focused on establishing the construct validity of measures by examining their relationships with other measures of the same or different constructs. It does not directly assess how well a measure predicts future outcomes or criterion behaviors.

• Predictive Evidence Validity: This type of validity focuses on the extent to which a measure predicts future performance or behavior (criterion). The MMMT matrix does not inherently contain information about future outcomes, making it unsuitable for providing predictive validity evidence.

Relationship with Reliability

While the MMMT design itself doesn't directly measure reliability (e.g., internal consistency, test-retest stability), reliable measures are a prerequisite for interpreting the validity coefficients obtained. Low reliability can attenuate correlations, making it harder to establish convergent validity or ensure discriminant validity. Therefore, reliability is often assessed and reported alongside the MMMT analysis, but the design's core contribution is to validity evidence, not reliability estimation.

Conclusion

Based on its focus on internal construct structure (convergent and discriminant patterns), the multitrait-multimethod design does not directly provide predictive evidence validity.