What tool should be used to find if there is really a statistically significant difference in the measured parameter?