A Note on the McQuitty Hierarchical Linkage Analysis
The questionnaire forms were punched on to IB M punch cards from which all subsequent analyses were made.
Descriptive analyses were made on the basis of marginal counts and selected cross-tabulations. Additionally:
(a) A total matrix of coefficients of correlations between each variable and all other variables in the questionnaire was constructed. The list of variables is given in Appendix I. The Pearson product-moment correlations were used for the linearlinear scales, Guttman’s coefficient of predictability for nominalnominal correlations, and the Goodman-Kruskal for ordinalordinal correlations. (b) A McQuitty Hierarchical Linkage analysis was performed on the total matrix, and nearly 50 clusters were discerned, averaging seven variables per cluster. A cut-off point of *i was used in defining clusters. Thirteen clusters were selected because they contained variables which were highly correlated (i.e. had a coefficient of correlation of at least *25) with either variable 61 - development as a manager - or 62 - development as a person - or both.