topic Re: Proc logistic: Different model fit statistic results using different reference/parameter coding? in New SAS User

Proc logistic: Different model fit statistic results using different reference/parameter coding?

fastandcurious — Tue, 08 Aug 2023 12:24:41 GMT

Hello,

I recently did a SAS Statistics 1 exercise (st107d05.sas) and got different results for the model fit statistics (Everything else was the same and displayed in a different order, I assume it's because I didn't apply a format)

I was wondering if there is a difference with the following reference coding techniques because I always thought they were same:

My code & some output

proc logistic data=STAT1.safety plots(only)=(effect oddsratio);
class Size (ref="1") Region (ref="Asia")/param=ref;
model unsafe (event="1")=Weight Region Size/ clodds=pl selection=backward;
UNITS weight= -1;
store issafe;

title 'LOGISTIC MODEL (3): Backwards Elimination';
run;

Solution code & some output

proc logistic data=STAT1.safety plots(only)=(effect oddsratio);
class Region (param=ref ref='Asia')
Size (param=ref ref='1');
model Unsafe(event='1') = Weight Region Size / clodds=pl selection=backward;
units Weight = -1;
store isSafe;
title 'Logistic Model: Backwards Elimination';
run;

Thank you!

Re: Proc logistic: Different model fit statistic results using different reference/parameter coding?

PaigeMiller — Tue, 08 Aug 2023 12:33:07 GMT

I recently did a SAS Statistics 1 exercise (st107d05.sas) and got different results for the model fit statistics

It looks to me like you got the exact same model fit statistics for both runs. I don't see a difference, do you? Please explain futher.

Re: Proc logistic: Different model fit statistic results using different reference/parameter coding?

fastandcurious — Tue, 08 Aug 2023 12:38:21 GMT

Sorry, thanks for the catch! I used the wrong snippet. The solution results should be this:

Re: Proc logistic: Different model fit statistic results using different reference/parameter coding?

fastandcurious — Tue, 08 Aug 2023 12:41:21 GMT

I'm sorry I restarted the program and I think I was comparing the wrong table models with each other.