These should be based on the population variance and the standard deviation. The wider the variability among the results and the wider the standard deviation among the same, the less accurate any one measured score is going to be in representing the overall set of data.