How are differences between the sample mean and the population mean explained when the result isn't statistically significant?