DATA Step, Macro, Functions and more

how to use only one level in chi square??

Reply
Contributor
Posts: 25

how to use only one level in chi square??

Hi i want to compare the proportion of males between two groups. 

 

first, am I correct in thinking a chi-square is appropriate?

second, how would I compare group by only one level of gender and not by both levels?

 

thanks for any help!

PROC FREQ DATA=_____;
TITLE3 'Chi-Square Test';
TABLES group*gender/CHISQ;
RUN; 
Esteemed Advisor
Posts: 5,484

Re: how to use only one level in chi square??

Posted in reply to iressa131

A Chi-square test is appropriate if there are enough cases in every table bin. Otherwise, you might consider using an exact Fisher test.

What is there to compare statistically between males and females in a single group?

 

/* Extract a few observations at random  from SASHELP.HEART dataset */
data heart;
call streaminit(78967896);
set sashelp.heart;
if rand("uniform") < 0.05;
run;

/* Compare the live/death status between males and females, 
   using two tests */
proc freq data=heart;
tables sex*status;
exact Fisher chisq / mc;
run;
PG
Super User
Posts: 10,698

Re: how to use only one level in chi square??

Posted in reply to iressa131
PROC FREQ DATA=_____;
TITLE3 'Chi-Square Test';
TABLES group*gender/CHISQ relrisk riskdiff;
RUN; 
Super User
Posts: 13,358

Re: how to use only one level in chi square??

Posted in reply to iressa131
PROC FREQ DATA=_____;
TITLE3 'Chi-Square Test males';
where gender='Male';
TABLES group/CHISQ;
RUN; 

could be one way. Of course without data I do not know your exact values of gender...

 

Ask a Question
Discussion stats
  • 3 replies
  • 97 views
  • 0 likes
  • 4 in conversation