Statement is not valid...Error when using the Slice statement in Proc Logistic

[FreddyAdams]

Hi,

Can someone give me the position of the Slice Statement in Proc Logistic syntax? I've put it all over the place and I keep getting the above error about invalidity or proper order. I'm using the param=glm option. Thanks, wardodd

[ballardw]

It would help to show the rest of your syntax as well. Some options may be incompatible with others.

A partial example from online help:

determines how to construct the partition of the least squares means for the model-effect. A slice-specification consists of an effect name followed by an optional list of formatted values. For example, the following statements creates partitions of the A*B interaction effect for all levels of variable A:

class a b;  model y = a b a*b;  slice a*b / sliceby=a;  

Similar to use of LSMEANS

[FreddyAdams]

Here's my code:

proc logistic data = test ;

class gender_code (ref='M') const (ref='1') back (ref='1')/param=glm;

model employed1 (event='1') = const back gender_code gender_code*back gender_code*const const*back/ rsq lackfit;

slice const*back/sliceby back;

oddsratio back;

oddsratio const;

oddsratio gender_code;

run;

[Doc_Duke]

Curious.  The code you posted here is different from that which came by e-mail.  The PARAM=GLM is required for the SLICE statement to be valid.  Perhaps you could post your log, as something strange is going on.

Doc

[FreddyAdams]

I pasted the wrong code at first (had param=ref) and edited it after I posted it. The log is vague - it just says it's invalid or is not in the proper order. Weird. I wonder if it's a patch thing or version issue.

[SteveDenham]

Unless you are operating on a version earlier than 9.2, I doubt that it is a version issue.  Try:

slice const*back/sliceby = back;

It may be that the syntax requires the equals sign in the sliceby option.

Steve Denham