next up previous contents
Next: Testing for Significance. Up: Procedure: Randomisation Version of Previous: Obtaining Raw Sample Points.

Obtaining Differenced Sample Points.

Because of the cumulative nature of evolution, it is possible that a small difference in the sampled value of a measure early on in a pair of runs will be magnified into a large difference later on, even if the two runs are actually proceeding in a fairly similar fashion. In order to gauge the magnitude of this effect, a duplicate set of tests was run, which used the difference in value between adjacent sample points as the figure to compare between runs, rather than the absolute value of the sample points. Using differenced data should reduce the influence of any cumulative disparity between runs.

Tim Taylor