Why is maximum likelihood ratio in surveylogistic using the t-test instead of the chi-square test?

juvepirlo — Wed, 10 May 2023 11:33:17 GMT

As shown, I'm processing nhanes data with the Survey Logistic process, but what I don't understand is why the t-test is used instead of the chi-square test in the maximum likelihood ratio estimation in Surveylogistic. In the global test, the same is not used chi-square test but the f-test, in normal logistics regression, both the global test and the maximum likelihood ratio estimation use the chi-square test, I don't know why surveylogistic and logistic are different..

proc surveylogistic data=sasuser.begin nomcar;
strata sdmvstra;
cluster sdmvpsu;
weight WTFOL2YR;
domain category('1');
class serumgroup/param=ref ref=first;
model diabete(desc) = serumgroup;
run;

proc surveylogistic data=sasuser.begin nomcar;
strata sdmvstra;
cluster sdmvpsu;
weight WTFOL2YR;
domain category('1');
class serumgroup/param=ref ref=first;
model diabete(desc) = serumgroup;
run;

Re: Why is maximum likelihood ratio in surveylogistic using the t-test instead of the chi-square tes

StatDave — Wed, 10 May 2023 13:39:44 GMT

This is discussed in the SURVEYLOGISTIC documentation. See the "Details: Testing Global Null Hypothesis: BETA=0" section. As noted there, you can use the CHISQ option in the MODEL statement, such as CHISQ(NOADJUST). See the discussion for details on the available suboptions that control the computation of the chi-square statistic.

topic Re: Why is maximum likelihood ratio in surveylogistic using the t-test instead of the chi-square tes in Statistical Procedures

Why is maximum likelihood ratio in surveylogistic using the t-test instead of the chi-square test?

Re: Why is maximum likelihood ratio in surveylogistic using the t-test instead of the chi-square tes