Convergent and divergent validity of three measures of conservation behavior. The multitrait-multimethod approach

Observations of the amount of reuse of glass, clothing, and metal were conducted at households of 130 individuals, and those direct observations were contrasted with the self-report of reuse of the same products. Two kinds of self-report were obtained: frequency of reuse self-reports and quantity of reuse self-reports. Thus, patterns of reuse of each particular type of material were assessed using the three methods of measurement. A multitrait-multimethod (MTMM) matrix of correlations between the reuse of these three materials, using these three methods of measurement, was analyzed. This matrix revealed the convergent and discriminant validities for the assessments of reuse. Higher correlations between direct observations and quantity self-reports were obtained than between observations and frequency reports. A confirmatory factor analysis of the MTMM matrix confirmed those results, adding significance testing to the validity assessment and to the partitioning of trait and method variance, modeled as latent factors.