The Study of Growth - An Invariant Approach to Statistical Analysis of Shapes - page 234

G obs statistic (max/min): 1.478
p-value : 0.204
1.047 |******

1.152 |****************************************************

1.257 |********************************************

1.362 |***************************************

1.468 |G************************

1.573 |*************

1.678 |*******

1.783 |***

1.888 |*

1.994 |*****

2.099 |

2.204 |*

2.309 |*

2.415 |*

2.520 |

2.625 |

2.730 |*

2.836 |

2.941 |*

3.046 |*

Confidence intervals for GDMA

Although we have some idea of the magnitude of differences in growth
pattern by inspecting the growth difference matrix, this cannot provide
information regarding which of these growth magnitudes is statisti-
cally different between the samples. Large within-sample variability
can underlie estimates of the mean and this may produce unexpected
statistical results. As explained previously, confidence intervals can be
calculated for the comparison of growth patterns for each linear dis-
tance. Confidence interval testing for differences in facial growth in
Cebus apella and Macaca fascicularis (shown below) is based on 200
non-parametric Bootstrap resamples. The chosen alpha level is 0.10. If
growth for a specific linear distances is significantly different in the
two samples, that linear distance is marked with a “**.”