Pandemic: Virulent Strain

[ralphmelton]

In February, I posted about my preliminary results with Virulent Strain. After 100 games with four players and five Epidemics with Virulent Strain, my win rate was nearly the same as without Virulent Strain. Was this a lucky streak, or a sign of something else

Comments

[ralphmelton]

I was playing purely VS games. My 4-player 5-Epidemic non-VS games mostly came in two big batches:
Games 124-544, spanning Dec-13-2013 to Feb-18-2014
Games 950-1344, spanning Jun-8-2014 to Sep-22-2014
(There are both games in those ranges that weren't 4P5E and 4P5E games that weren't in those ranges.)

My Virulent Strain games started Jan-3-2016 (game 2002) and continued to Jun-6-2016 (game 2576).

I agree that collecting new non-VS data is an obvious way to test the "improved player" hypothesis. But it seems hard to prove that:
Since I recorded that in 18 of 400 games I lost due to Virulent Strain and wasn't obviously doomed otherwise, let's assume for discussion that I'm now 4.5% more likely to win 4P5E (non-VS) than I was when I played those games. Could I prove that?
If I played another 500 games, it would take me six months and I'd get a confidence interval of about 3% - too wide a confidence interval to prove that I've gotten better. Confidence intervals are proportional to 1/sqrt(sample size), so to narrow the confidence interval enough to prove a 4.5% improvement, I'd have to play about 2000 games, taking me two years.

I'm not absolutely opposed to doing that, but I'm not feeling very eager. So first I will engage in speculation about possible causes and enlist the help of statistician friends to see whether there's more information I can tease out of the data I have so far.