Originally Posted by
Juret
Hi,
I have a data set of goal totals that are like a positively skewed Poisson distribution (tail to the right) due to a couple of outliers. The variance is for example larger than the mean and it doesn't pass a Chi-Square goodness-of-fit test for Poisson, failing heavily because of the tail to the right (otherwise fitting quite good). Any ideas how to deal with this? I don't think ignoring the outliers would be a good idea, it seems consistent that sometimes the score drifts away. When the score has drifted away, it is always due to a team winning huge, with only one exception when there was a high-scoring tie. It seems to me that sometimes a team just gives up, and this disturbs the Poisson process.
Thank you a lot