Frustrated at being unable to find a detailed chi-squared distribution for only 1 degree of freedom, as used for results of the Mantel Haenszel chi-squared test, to quickly reference your test statistic for the approximate two-sided P-value? I was, so delved into Excel to construct the following table – colour coded for glory! – using the formula “CHISQ.DIST.RT(x,deg_freedom)” for results from 0.000 (P-value = 1.00) to 15.975 (P-value = 0.000064) in increments of 0.025. Critical values are in bold.
To use, cross-reference your Mantel-Haenszel chi-squared result first with the whole numbers at the top of the columns and then with the approximate decimal point to find the corresponding approximate P-value. For example for MHX2 of 8.43 you would cross reference 8 at the top with 0.450 on the side to find the two-sided P-value of 0.0037.
Enjoy the rainbow of probability! It can be downloaded by clicking on the tools icon “>>” in the upper right-hand corner and selecting “download”.
You can also make your own table of areas in the upper tail of the standard normal distribution (one-sided P-value) in Excel using the formula “=1-NORMSDIST(‘Z-score array’!D5)” Where ‘z-score array’ refers to a seperate worksheet (called ‘z-score array’ with an the arbitrarily demonstrated cell ‘D5’ which is one of many in an array of z-score values from 0 to 3.99 (or higher, I went to 4.49, but that’s a pretty small area in the tail there….), arranged with x.x down the left column and -.-x across the top column. Ha, or you could just look it up in a book, hey? 🙂