Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Estimate statement in proc logistic for binary out...

- RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-18-2018 02:55 AM

Hi,

I am calculating the odds ratio of different treatment groups using proc logistic. I tried to do it in two ways: 1. by using estimate statement 2. by not using estimate statement. By just giving the odds option (code below). But I get different results for each option. Can someone tell me the difference between using an estimate statement and not using it in proc logistic?

proc logistic data = final;

class cohort (ref=LAST) tymcdpip / param=ref ;

model anl01fl(event='Y') = cohort tymcdpip / expb;

by segorder;

ods output OddsRatios = odds

ParameterEstimates = pval;

run ;

Where, cohort = treatment group , tymcdpip = covariate, segorder = variable which defines different treatment comparisons as a segment, and anl01fl = responder variable

- Mark as New
- Bookmark
- Subscribe
- RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Veen

04-18-2018 10:11 AM

For effects like COHORT which only appear as main effects in the model, the odds ratio estimate is provided automatically and the label in the provided table indicates its interpretation (which levels are compared). For any variable in the model, you can also request odds ratio estimates using the ODDSRATIO statement. This is far easier than using the ESTIMATE statement since you do not have to determine the correct linear combination of parameter estimates to yield the odds ratio estimate. Note that, in general, the EXPB option does not necessarily provide odds ratio estimates since, depending on the model parameterization, the odds ratio might not be computed simply by exponentiating a parameter estimate. There is no "odds" option. If you are referring to the ODS OUTPUT statement you show, that simply saves a displayed table in a data set.