Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

 From loprinzp@onid.orst.edu To statalist@hsphsun2.harvard.edu Subject st: Adjusted Proportions Date Sun, 15 May 2011 17:01:17 -0700

```Dear All:

```
Any help you can provide on the following would be sincerely appreciated. I am using complex survey data and, fortunately, have had luck obtaining adjusted means for age for each level of a frailty variable (i.e., 1 = frail, 2 = pre-frail, and 3 = not frail) using the following commands:
```
svy: regress age sex race smoker2 school2 income2 bmicat frailtycat
adjust sex race smoker2 school2 income2 bmicat, by(frailtycat) se ci

Then to get p-values I ran the following command:
xi:svy: regress age i.frailycat sex race smoker2 school2 income2 bmicat
testparm _Ifr*

```
Now for my question. I am having difficulty obtaining adjusted proportions for categorical variables, specifically for different BMI categories (1 = underweight, 2 = normal weight, 3 = overweight, and 4 = obese) across the different frailty groups.
```
```
Using the following glm command I was able to get the mean BMI category value for each level of the frailty variable.
```
xi:svy: glm bmicat i.frailtycat sex race smoker2 school2 income2
adjust sex race smoker2 school2 income2, by(frailtycat) se ci

```
While adjusting for various covariates, however, I am trying to see how many people are in different BMI categories (1 = underweight, 2 = normal weight, 3 = overweight, and 4 = obese) across three different frailty groups (1 = frail individuals, 2 = approaching frailness, and 3 = not frail).
```
```
```