[AlanBostick]

[ QUOTE ]
Finally, the 26 samples doesn't mean much to me. There's so much going on (rainbow boards vs twotone, how many times the flush card hits, how many times the straight draw hits, etc.) that intuition tells me it would take a hundred trials to get within 10%. Where does the +/- 8% come from?

[/ QUOTE ]

The +/- 8% comes from sampling theory using binomial statistics. 26 samples total, 21 hits, five misses. Standard error is sqrt(26 *21/26 * 5/26) = 2.01 hits. The best estimate of hit rate is then 21/26 +/- 2.01/26 = 0.807 +/- 0.077.