Secondly, the sample was considered to be representative of the population for the inference predictions based on the normal distribution curve. Thirdly, the Durbin-Watson test statistic was used to test for variable independence through an inspection of the autocorrelation in the regression residuals (Durbin & Watson, 1950). As a rough rule of thumb, if Durbin-Watson is very small (d ≤ 1) there may be cause for alarm whereas small values (1 < d < 2) indicate that successive error terms are close in value to one another or positively correlated and large values (d > 2) suggest that successive error terms are very different in value from one another (i.e. negatively correlated). From the sample data, most of the residual autocorrelations fell within the 95% confidence bands around zero (0.287 < autocorr < 0.305). Similarly, a Durbin-Watson autocorrelation test statistic of 1.725 was established, indicating positive autocorrelation or a perfect estimation of the level of statistical significance in the model.