Discussions
The results of the foregoing analyses have been quite revealing. It is not surprising that the results of the empirical analysis agree largely with the findings from the literature review process. While the use of geometric mean to unify all performance parameter proxies showed no significant difference between itself and the a prioriexpectations or the theses researchers’ pre-study perceptive conclusions indicated in the analysed theses, using 10% level of significance (β (30) = .278, t (30) = .695, R2 = .0772, p > .10), all the individual surrogate variables excepting ROE returned significant figures for t at the 10% significant level. Buttressing the t statistics findings with F statistics, the geometric mean unified variable result was also not significant both at the 5% and 10% levels of significance (Adj. R2 = .0442, F (1, 28) = 2.426, p > .05), as against the test for the individual performance parameter surrogates which returned overall significant difference in relationship with the analysed theses pre-study expectations (Adj. R2 = .0291,F (4, 25) = 22.598, p < .05).
In other words, what the results of the various analysis is telling us is that the theses researchers’ pre-conceived conclusions were no different from the results obtained using the geometric mean unified dependent variables. Looking at the figures on table 3, it will be interesting to note that while the geometric mean unified performance variable maintained a positive relationship with the theses a priori expectations, one out of the four surrogate variables (ROA) showed strong negative relationship indicating that its individual influence inhibits the positive pull of the other variables, thereby distorting logic of joint conclusion. This trait was also exhibited on table 2 (summary statistics) which showed all the figures of the unified variable as positive but three of the individual surrogate variables returned negative minimum figures.