| Cobretti2 said: uh your stats is BS because you cant have an outlier above 100 as the highest score is 100. also the data is heavy skewed to the right so there is no normal distribution. |
This. It doesn't make sense to use the IQR method with a distribution that (a) is heavily skewed (in this case, negatively skewed), and (b) bumps up against the maximum or minimum possible score (in this case, the maximum of 100), because it will be impossible to find outliers at one end of the distribution (in this case, at the high end). The review-score distributions for almost all decent video games fit these criteria, so using the IQR method with them is generally a bad idea.
If you are really concerned about outliers, you could argue for using the median score rather than the mean score. Or you could trim the distribution by eliminating the highest and lowest 10% of the review scores, and then take the mean of the remaining scores. However, as other have said, if you want to use an alternative method you should apply it to all games. Because game score distributions are almost always negatively skewed, the median and trimmed mean will almost always be a couple of points higher than the raw mean, as in the case of Halo 4 (median of 90 vs. raw mean of 87). So the final ranking of games from "best" to "worst" will look an awful lot like the metacritic list you started with, except that every game's "average" score will be a couple of points higher.
My advice: Don't bother, because it won't make a difference and it doesn't matter anyway.
