For example, the sequence
contains two increasing runs, both of size 3. The value 51 that ends the first run is not included as part of the second run; it's skipped.
Terminating values are skipped to keep the runs independent. If the terminating run were included in the following run, the runs would be interdependent, which makes analyzing the results much more complicated. In particular, the chi-square test no longer works because it assumes independent trials.
This page last modified on 2 March 2005.