SAS code to calculate 95%CI for standardized estimates for categorical predictors in linear regress

[Kunko]

I am using to predict stronger lifestyle predictors (e.g., Physical activity: active. moderate, inactive) of a continuous outcome and wanted to have standardised estimates (beta-weights) using prog reg and also proc surveyreg in SAS.

proc surveyreg data=data;
class smoking Alcohol PhysicalActivity gender;
model BMI = smoking Alcohol PhysicalActivity gender age in years energy intake in kcal / stb solution CLPARM;
run;

However, the 95%CIs generated from both procedures are for unstandardised estimates. Is there anyone who can help me to get 95%CI for the standardised estimates? Thank you in advance for your unreserved support.    

[Rick_SAS]

The standardized estimates are the parameter estimates that you obtain by standardizing all the variables in the model (Xs and Y). So one way to obtain the standardized estimates is to output the design matrix and use PROC STDIZE to standardize the variables in the model. Then rerun the regression on the standardized variables. 

• How to generate dummy variables and output the design matrix
• How to standardize and rerun the regression to obtain standardized estimates

[Kunko · Accepted solution]

Dear Rick,

Thank you so much. It worked well.