SilentWolf said:
I thought that in statistics you only needed 31(random..ish) samples to be able to safely say that you have enough to represent the whole. If true that would put OoT at (barely) enough reviews to say that its current percentage is a good representation. |
You can get by with that number but you'd rather not have too. Ideally you'd have at least 120 reviews (If I recall correctly, I don't have my books with me at the minute). The lower the number of reviews the wider the confidence intervals must be. My analysis was flawed because I applied an identical confidence interval to all games (95% confidence interval, represented by +- 1.96 standard deviations from the mean). Had I not done that the OoT result would've been a lot weaker (and so would the other games but not to the same extent).







