CHAPTER 7
RESULTS
Including both fractal dimensions, ten variables were examined. Basic statistics
were computed for each variable independently (Tables 2 and 3). The parametric
medians for each group are graphically represented for both fractal dimensions in Figures
2 and 3. Since the samples sizes for these groups are small, the data was bootstrapped for
1000 iterations to try to obtain more reliable means and standard errors, since no
assumption is made regarding the distribution of the data. As shown in Tables 4 and 5,
there was little difference between the parametric mean and the bootstrapped mean.
The ruler and information fractal dimensions for each species were regressed
against each of the measured size/shape variables. Out of the 32 regressions performed,
only three resulted in significant P-values, i.e. P-values less than .05. However, the
coefficient of determination (r-squared) was very weak for these three regressions,
ranging from 12.1% to 37.2% (Table 6).
Tables 7 and 8 report the ruler and information fractal dimensions calculated for
both species. One interesting (and seemingly impossible) aspect of two of these fractal
dimensions is that they are below 1.0. Note in Table 3, Procolobus badius specimen
number 2107 (Figure 4) has a ruler fractal dimension of 0.99209 and P. badius specimen
number 9433 (Figure 5) has a ruler fractal dimension of 0.98466. However, their
information fractal dimensions are both above