For each run, 10 sample data points were extracted, each one
representing the value of the measure in question at one of 10
equally spaced times throughout the run.
The basic idea of the paired sample *t* test in this situation is to
consider the 10 sample points for pairs of runs in turn.
By doing pairwise tests at 10 sample points we are comparing the
measures at a number of points through the run, with no point having
more significance than any other.
For each pair
of runs, the difference between corresponding samples is calculated,
together with the mean value for the 10 differences. We then ask
what the likelihood is of achieving this mean difference under the
null hypothesis that the two runs are statistically equivalent.
The method by which this is done will be explained shortly.

1999-05-29