Mantel-Haenszel statistics and direct standardization.

This paper demonstrates that the Mantel-Haenszel chi-square statistic may be employed to test the equality of two directly standardized rates. For certain epidemiological applications, this eliminates the need for special computer programs to contrast two directly standardized rates. Furthermore, since the weights are determined entirely by the data there is the advantage of avoiding the selection of an external standard population. Data analysts who compare two groups of rates with the Mantel-Haenszel chi-square statistic may view this statistic as a comparative measure of two directly adjusted rates.