STATA to SAS: survey logistic regression

#1
I am getting very different coefficients and p-values with STATA and SAS. It would be great to figure out which program is set up wrong.

My sampling weights variable is samp_wt. My cluster variable is sta3n. My binary response is Screened.

STATA:
svyset sta3n [pweight=samp_wt]
svy: logit screened i.group
margins group

SAS:
proc surveylogistic data=mydata;
class screened (ref='no') ;
model screened=group;
cluster sta3n;
weight samp_wt;
run;
 
#3
Stata:
| Linearized

screened | Coef. Std. Err. t P>|t| [95% Conf. Interval]

-------------+---------------------------------------------------------------- group |

group1 | -.2372838 .0320043 -7.41 0.000 -.300562 -.1740057

group2 | -.3793562 .0524134 -7.24 0.000 -.4829868 -.2757257
|
_cons | 3.116272 .0655551 47.54 0.000 2.986658 3.245886

SAS: Analysis of Maximum Likelihood Estimates

Parameter DF Estimate Standard

Error Wald

Chi-Square Pr > ChiSq

Intercept 1 -2.9106 0.0597 2379.9834 <.0001

group group1 1 0.0316 0.0194 2.6723 0.1021

group group2 1 0.1740 0.0308 31.9014 <.0001