Unless you are using a different meaning of 'reversion to the mean' than I am familiar with, it's not a statistical error and has nothing to do with the size of

Unless you are using a different meaning of 'reversion to the mean' than I am familiar with, it's not a statistical error and has nothing to do with the size of the sample. 'Regression to the mean' happens when you have an imperfect correlation between two data fields.