Experiment 1: As numbers - near median

seconds frequency Cumulative %
0.019995 687 9.92%
0.019996 895 10.75%
0.019997 1334 11.99%
0.019998 209 12.18%
Mean 0.019999 1898 13.95%
0.020000 2671 16.44%
0.020001 3403 19.60%
0.020002 4747 24.02%
0.020003 7742 31.22%
Median 0.020004 13059 43.37%
Mode 0.020005 17121 59.30%
0.020006 13630 71.98%
0.020007 8211 79.62%
0.020008 5404 84.64%
0.020009 570 85.18%
0.020010 3158 88.11%
0.020011 2305 90.26%
0.020012 1787 91.92%
0.020013 1262 93.09%
0.020014 886 93.92%

Transcript

And if we look at our statistics data of those frequencies for those bins.  We see that the median is down here at 20.004 milliseconds that was the frequency.  43.37% of the data was in the same bin as the median The mode (the most commonly occurring value) was just a little bit longer - one microsecond longer and that was 59.30% of the data. But if we look at the mean, we see that the mean is way down here at 0.019999 seconds (sorry) 19.999 milliseconds, But that represented only 13.95% of the data. So we see that there's a pretty big issue here - the data isn't arriving at the twenty-millisecond separation, so I think some are arriving a little sooner and some are arriving a little later - shifting our median value out just slightly.  So half the data takes longer than this half the data takes less.