> And yet somehow whenever you take a closer look at mislabeled product prices, the average is always in favor of the store.

This could just as easily be selection bias: the errors in favour of the customer are less likely to get reported by customers.