How to modify 'proc glm' in sas to find estimates per 10 units or per SD unit change of the exposure

[ph6]

I am using proc glm in SAS to find the estimates per unit change of exposure for the response variables. How could I modify my code so that it would calculate the estimates per 10 units or per SD change of the exposure variables? Here is my code:

 PROC GLM DATA=MyData PLOTS=NONE;
        CLASS &ClassAdj;
        MODEL &TraitList = &PVar &ClassAdj &ContAdj / SOLUTION E;
        %IF &ByVars ^= %THEN BY &ByVars;;
    QUIT;

[PaigeMiller]

Per 10 Units:

You don't have to modify your code, you can just multiply the regression coefficients by 10.

Per 1 SD unit change

As above, you just need to multiply the regression coefficients by the proper number.

Or you could modify the input data, but that isn't what you asked.

[ph6]

Thank you so much. For the SD do you have any thoughts how could modify the data?

[PaigeMiller · Accepted solution]

You can change the input data by standardizing it so each continuous variable has a standard deviation of 1.

Use PROC STDIZE with METHOD=STD

[jiltao]

You can also consider using the ESTIMATE statement in PROC GLM to estimate what you wanted. For example, if the model is --

model y=x1 x2 x3;

then estimate 'change in y for 10 units change in x1' x1 10;

estimate ' change in y for sd=3.4 units change in x1' x1 3.4;

might get what you wanted.

[ph6]

Thank you, what if I have logistic regression?

[jiltao]

You would use PROC LOGISTIC to fit your model with similar ESTIMATE statements to get what you wanted. The results are in the logit scale.

[Ksharp]

Logistic regression check UNITS Statement.

[ph6]

Would it be OK to divide x by 10, and standardize the SDs before running the model?

[Ksharp]

No. That is not OK. Logistic model have different LINK function .

[ph6]

.