topic Re: odds ratio interpretation in Statistical Procedures

odds ratio interpretation

odlum — Mon, 05 Jun 2017 15:42:06 GMT

A case-control study was conducted to assess the association between exposure to tobacco smoke (TS) and lung cancer. 300 lifelong nonsmoking lung cancer cases and 1,288 nonsmoking controls were interviewed about exposure to ETS. Let cancer=I(lung cancer), ts= I(exposed to TS), gender=I(female). A logistic regression model was fit using the following SAS code

proc logistic data=cancer descending;

model cancer/count = ets gender ets*gender; run;

Here is part of the SAS output.

Analysis of Maximum Likelihood Estimates

Parameter	DF	Estimate	Standard Error	Wald Chi-Square	Pr	> ChiSq
Intercept	1	-1.8946	0.2398	62.4070		<.0001
ts	1	-0.8134	0.2926	7.7308		0.0054
gender	1	1.0512	0.2653	15.6951		<.0001
ts*gender	1	0.8450	0.3320	6.4790		0.0109

How do you estimate odds lung cancer for males with ETS exposure?

Re: odds ratio interpretation

Reeza — Mon, 05 Jun 2017 16:27:10 GMT

Look at using the ODDSRATIO statement to determine the odds ratio for a variable in the interaction term.

Note that if your variables are categorical there should be a CLASS statement in your PROC and you should also define the coding type (GLM, REF, etc)

PS I'm going to move this to the Statistical Procedures forum.

Re: odds ratio interpretation

odlum — Mon, 05 Jun 2017 17:32:37 GMT

Thank you.This is one of questions in a problem set_neither complete output or data is given to rerun the codes.How can I determine the odds ratio of lung cancer in males with TS exposure using the information on the estimates in the table?

Re: odds ratio interpretation

Reeza — Mon, 05 Jun 2017 17:40:49 GMT

Ok...then that's a stats question, not a programming question.

I have no idea how males/females or treatment are coded but you can calculate the probability by hand if you write the model for males with treatment out and then calculate the estimate. The exponent of that value would be the odds ratio.