RT Journal
A1 Hemmila M
T1 Propensity score methods: Setting the score straight—reply
JF Archives of Surgery
JO Archives of Surgery
YR 2011
FD July 1
VO 146
IS 7
SP 887
OP 889
DO 10.1001/archsurg.2011.160
UL http://dx.doi.org/10.1001/archsurg.2011.160
AB 
P values are numbers. Significance is set by the investigator, and a P value less than the set point for significance allows for rejection of the null hypothesis. Interpreting smaller P values as impressive is potentially misguided.1 A relevant criticism of P values is whether a very high significance value yields any practical implications. As Mayo and Pawlik point out, a finding with a high degree of statistical significance could be of very little use to patients, and clinically meaningful differences might be overlooked because they do not meet the arbitrary cutoff. A likely better representation of the precision of the statistical estimate is to use a 95% confidence interval.