topic Re: Goodness of fit for surveylogistic in SAS Programming

Goodness of fit for surveylogistic

sammmy — Tue, 17 Apr 2018 03:01:39 GMT

Hi,

I actually wanted testing the goodness of fit of my model developed from PROC SURVEYLOGISTIC but since it doesn't give such options, I opted for PROC LOGISTIC just to test the goodness of fit of my model it gave me the output below. can I say my model is adequate?

SAS Output

Re: Goodness of fit for surveylogistic

PeterClemmensen — Tue, 17 Apr 2018 06:22:00 GMT

Do you have 57.000 degrees of freedom in your model?

Re: Goodness of fit for surveylogistic

sammmy — Tue, 17 Apr 2018 07:45:55 GMT

Not sure where the 57000 is coming from but my observations are 66000

Re: Goodness of fit for surveylogistic

PeterClemmensen — Tue, 17 Apr 2018 08:00:10 GMT

what does your code look like?

Re: Goodness of fit for surveylogistic

sammmy — Tue, 17 Apr 2018 12:14:54 GMT

PROC LOGISTIC DATA= risk ;
CLASS X1 X2 X3 X5 X6 X7 X8 X9 X10 X13 X14 X15 X16 X17/param=glm;
 MODEL Y (Event = '1')=X1 X2 X3 X4  X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 X16 X17/ 
    LINK=LOGIT expb LACKFIT scale=none 
	aggregate=(X1 X2 X3 X4  X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 X16 X17) ;
RUN;QUIT;

Re: Goodness of fit for surveylogistic

Ksharp — Tue, 17 Apr 2018 12:31:49 GMT

aggregate=(X1 X2 X3 X4  X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 X16 X17) ;
====>
aggregate ;

Re: Goodness of fit for surveylogistic

Ksharp — Tue, 17 Apr 2018 12:35:04 GMT

If your model looks good, Value/DF should be near 1 .

Re: Goodness of fit for surveylogistic

sammmy — Tue, 17 Apr 2018 12:49:41 GMT

Okay, for both statistic or for any of them?

Re: Goodness of fit for surveylogistic

Ksharp — Tue, 17 Apr 2018 13:26:32 GMT

Good question. Should be both, but I would accept any of it .