topic Re: Getting standardized residuals with proc genmod in Statistical Procedures

Getting standardized residuals with proc genmod

sharcole — Thu, 02 Mar 2017 21:46:42 GMT

I'm having trouble getting output of standardized pearson residuals and standardized deviance residuals in proc genmod. I was able to get non standardize residuals with this code:

output out = overall3 reschi=reschi p=predicted;

but get all missing values when using this code: output out = overallstd stdresdev=stdresdev p=predicted;

Thank you for any assistance

Re: Getting standardized residuals with proc genmod

Reeza — Thu, 02 Mar 2017 21:59:50 GMT

Please post your full code and log.

I don't have any issues in SAS 9.4M3. Note what SAS version you have, and what SAS/STAT version you have as well.

If the code below doesn't run successfully, include the log.

data drug;
   input drug$ x r n @@;
   datalines;
A  .1   1  10   A  .23  2  12   A  .67  1   9
B  .2   3  13   B  .3   4  15   B  .45  5  16   B  .78  5  13
C  .04  0  10   C  .15  0  11   C  .56  1  12   C  .7   2  12
D  .34  5  10   D  .6   5   9   D  .7   8  10
E  .2  12  20   E  .34 15  20   E  .56 13  15   E  .8  17  20
;

proc genmod data=drug;
   class drug;
   model r/n = x drug / dist = bin
                        link = logit
                        lrci;
output out = overallstd stdresdev=stdresdev p=predicted;
run;

Re: Getting standardized residuals with proc genmod

sharcole — Thu, 02 Mar 2017 22:18:42 GMT

Thank you,

Your example works - I have 9.4 TS1M3

When I try with my own data, however, it is not working.

I note that your dependent variable is r/n . Would this have any implications?

The full code looks like this:

proc genmod data=stigma4 descending ;

class subid adependent (ref="0=No") time (ref="0")

marstat_bl (ref="1=Married/living with partner")

employ_bl (ref="1=Yes") educ9_bl (ref="0=No") msm2_bl (ref="2=N/A") suiclife_bl (ref="0=No")

moddep (ref="0=No") supportmedian (ref="1=Yes") drug_dependent (ref="0=No")/param=ref ;

model negimage_2 = adependent time marstat_bl employ_bl educ9_bl

msm2_bl suiclife_bl moddep supportmedian drug_dependent

mcs pcs hiv_yrs_bl ls_age_bl / dist=bin type3 link=logit lrci;

output out = overallstd stdresdev=stdresdev p=predicted;

repeated subject=subid/ type=ind corrw ;

Re: Getting standardized residuals with proc genmod

Ksharp — Fri, 03 Mar 2017 03:15:07 GMT

I noticed you are using REPEAT statement, that mean you are fiting GEE.

Can you see the standardized residuals in the output ?

Re: Getting standardized residuals with proc genmod

sharcole — Fri, 03 Mar 2017 15:26:53 GMT

Hi - Yes, this is a GEE.

I'm wondering if The standardized residuals cannot be computed mathmatically for GEE models?

I get all missing values in the output.

Thank you!

Re: Getting standardized residuals with proc genmod

Ksharp — Sat, 04 Mar 2017 02:47:03 GMT

If you can not see standardized residuals in RESULT printed by GEE, then you can not get it,

GEE take residuals as correlation.

Re: Getting standardized residuals with proc genmod

Ksharp — Sat, 04 Mar 2017 13:39:26 GMT

Add option OBSTATS into MODEL statement and see if you can get Pearson residual, if you can obtain it then you can output it.

Re: Getting standardized residuals with proc genmod

sharcole — Mon, 06 Mar 2017 19:14:39 GMT

Thank you all-

SAS technical support clarified: The standardized residuals cannot be computed for GEE models. They depend on the deviance or Hessian matrix which isn't involved in the GEE estimation method