https://communities.sas.com/t5/SAS-Procedures/Variance-of-Difference-of-Ratio-Means/m-p/66920#M19125 I have gotten a request from our statistician to compute the ratio mean, variance of the ratio mean, difference between ratio means, and variance of the difference between ratio means for our clinical trial. The goal is to examine rates of different events (e.g. headache) occurring between our clinics and between our studies. The statistical documentation I received deals with stratified unequal clusters and provides some simple, yet cumbersome formulas for these statistics. I have been searching for a SAS procedure that will calculate these values, but I haven't found anything, or perhaps I simply am not recognizing the right thing.<BR /> <BR /> Here are the forumlas given in case this illustrates what I need better (there are other more computational formulas I have as well):<BR /> <BR /> ratio mean: r = y/x<BR /> var(r) = (1/x^2)(sum(var(y)) + r^2*sum(var(x)) - 2r*sum(cov(y,x))), sums over the strata<BR /> <BR /> difference: r - r' = y/x - y'/x'<BR /> var(r - r') = var(r) + var(r') - 2cov(r,r')<BR /> <BR /> Any help would be appreciated, even if that I have to write code to compute these. Thanks. Mon, 22 Dec 2008 21:56:39 GMT Kastchei 2008-12-22T21:56:39Z https://communities.sas.com/t5/SAS-Procedures/Variance-of-Difference-of-Ratio-Means/m-p/66920#M19125 I have gotten a request from our statistician to compute the ratio mean, variance of the ratio mean, difference between ratio means, and variance of the difference between ratio means for our clinical trial. The goal is to examine rates of different events (e.g. headache) occurring between our clinics and between our studies. The statistical documentation I received deals with stratified unequal clusters and provides some simple, yet cumbersome formulas for these statistics. I have been searching for a SAS procedure that will calculate these values, but I haven't found anything, or perhaps I simply am not recognizing the right thing.<BR /> <BR /> Here are the forumlas given in case this illustrates what I need better (there are other more computational formulas I have as well):<BR /> <BR /> ratio mean: r = y/x<BR /> var(r) = (1/x^2)(sum(var(y)) + r^2*sum(var(x)) - 2r*sum(cov(y,x))), sums over the strata<BR /> <BR /> difference: r - r' = y/x - y'/x'<BR /> var(r - r') = var(r) + var(r') - 2cov(r,r')<BR /> <BR /> Any help would be appreciated, even if that I have to write code to compute these. Thanks. Mon, 22 Dec 2008 21:56:39 GMT https://communities.sas.com/t5/SAS-Procedures/Variance-of-Difference-of-Ratio-Means/m-p/66920#M19125 Kastchei 2008-12-22T21:56:39Z https://communities.sas.com/t5/SAS-Procedures/Variance-of-Difference-of-Ratio-Means/m-p/66921#M19126 Have you looked at proc means ? Mon, 22 Dec 2008 22:36:44 GMT https://communities.sas.com/t5/SAS-Procedures/Variance-of-Difference-of-Ratio-Means/m-p/66921#M19126 GertNissen 2008-12-22T22:36:44Z