Current measures of test profile similarity are inadequate since they are not applicable when different, independent dimensions of comparison exist. A method appropriate to the hypothesis underlying pattern analysis, i.e., a convergence of tests in a given profile with nonlinear joint functions, is generated from multiple-eta. The method proposed, "basically a curve-fitting technique,...requires the gathering of normative data and may be thought of as an actuarial method of pattern analysis. With this method, criteria and test data may be continuous or categorical. The method makes possible prediction of the criterion from the profile, measures the degree and significance of this predictability, allows for the assessment of similarity and dissimilarity of profiles with respect to the criterion dimension.".