Last Mile and Private Peer tests can be run in a Batch group which runs the tests through the same machine during a given test run to allow for added comparisons.
Between one running and the next various things can occur to stop the machine from retuning results. For instance the Peer could shut down or have a loss of connectivity. It is also possible that data could be marked as suspect and be 'hidden' from the results. This could be if a critical UTA error occurs, or there is an issue with the bandwidth test.
If any of these occur, you would see slightly different test counts in the results.