topic Re: Proc Freq Odds Ratios in Statistical Procedures

Proc Freq Odds Ratios

alexgouv — Thu, 19 Sep 2019 17:09:42 GMT

Hi all,

I am having a small problem with my odds ratios which I am generating from proc freq. They are giving me the inverse of what I want and I cannot figure out how to flip them. Right now I am comparing Treatment to Outcome while adjusting for Type. It is giving me the Odds Ratios for TRT to Outcome, when I want Outcome to TRT. Any ideas on how to flip them? I have tried flipping them in the tables statement and changing the order of the data but no luck. Here's the code I'm currently using:

proc freq data = p1 order = data;
	weight count;
	tables od*trt*out / 
			chisq expected nocol measures cmh alpha=.1 out = test;
run;

I just want to flip it so I don't have to calculate confidence intervals manually, so if this can't be done or is very complicated let me know.

Thanks!

Re: Proc Freq Odds Ratios

SAS_Rob — Thu, 19 Sep 2019 17:29:11 GMT

I am not sure I understand what you mean by your comment about flipping the variables. In general, your outcome variable should be the column, that is, the last variable on the TABLES statement, with the level of interest in the first column. Your row variable would be ordered such that the row1 level will be compared against the row2 level in the odds ratio.

You should be able to control the ordering of the rows and columns by using the ORDER= option. The easiest thing to do is to set up formats that properly reflect the ordering and use the ORDER=FORMATTED option.

Re: Proc Freq Odds Ratios

ballardw — Thu, 19 Sep 2019 17:55:16 GMT

Data? Values?