Results

Across subjects the observed third central moment as well as the predicted third central moments for the PE model and the Gamma model differed considerably from the normal distribution. The distributions were very skew. The values for the Kolmogorov Smirnov Goodness of Fit statistic K was equal to 4.047 (p=0.000) for the observed distribution and equal to 5.876 (p=0.000) for the predicted distributions. Application of Students's t-test for the comparison of two means requires normal distributions for the variables involved. Therefore, it was decided to take the natural logarithm of the third central moment instead of the third central moment itself. However, in a number of cases (N=69) the observed third central moment was negative. Because the logarithm is only defined for positive real numbers, these cases were excluded from further analysis. The values for the Kolmogorov Smirnov Goodness of Fit statistic K were now equal to 0.479 (p=0.976) for the observed distribution and equal to 0.943 (p=0.337) for the predicted distributions. The results of Student's t-tests were as follows: in the case of the Poisson-Erlang model H₀ was not rejected (observed mean = 0.614, predicted mean = 0.889, N = 116, t = -1.19, df = 115, p = 0.235) and in the case of the Gamma model H₀ was significant at the 5 (observed mean = 0.614, predicted mean = 1.177, N = 116, t = -2.44, df = 115, p = 0.016). These results are clearly in favor of the PE model.