decades experience using and teaching statistical analysis procedures (Mardia, Kent and Bibby, 1979; and L. Poynor and H. Poynor, 2004) lead to the conclusions that using mean item-score preserves the original scale of 1...5, and distorts the data in no way. When making statistical comparisons across sub-samples using analysis of variance, the consideration of differing sample sizes, and between- and within-groups variation is built into the long-studied and well-used decision statistic.
The original intent of this study was to compare leader behaviour preferences across gender and race for Blacks and Whites. In the initial data analysis, results indicated significant differences between the responses of these two racial groups and the Asian and Coloured samples. As the Asian and Coloured sample sizes are too small for generalizations, they are included only as indications of requirements for future research, which is currently underway.
Multiple comparison analyses of variance for the groups for each factor were run to investigate specific differences. Complete and detailed analyses of differences are in Table 1. The leader behaviour preferences of South African managers are diverse and representative of many racial groups. There were significant differences in seven of the twelve factors of the LBDQ. In Table 1, with the exception of the Coloured sub-sample, there are relatively few differences between the groups defined by race and gender. South African Coloured samples in Table 1, while small in sample size, indicated very large differences between males and females. Means for Coloured female were