comparing outcome variable (dummy) over group

[ajay pasi]

Dear community,

May anyone suggest a way to compare the distribution of the outcome variable(which is a dummy) over a group? I know vioplot does it for continuous outcome variables over a group. Is there any way to do that?

regards,
ajay

[ajay pasi]

The Pearson's χ² test--> I believe is a test for that.

[Nick Cox]

What's a group here? The answer could be many things, depending, but if group means a categorical variable, a chi-square test could make sense, or a test comparing proportions.

[ajay pasi]

Thank you sir. Group is a categorical variable.

[Carlo Lazzaro]

Ajai:
do you mean something along the lines of the following toy-example?

. sysuse auto.dta
(1978 automobile data)

. g wanted=0 if rep78<=2

. replace wanted=1 if wanted==.

. tabulate wanted foreign, chi2

           |      Car origin
    wanted |  Domestic    Foreign |     Total
-----------+----------------------+----------
         0 |        10          0 |        10 
         1 |        42         22 |        64 
-----------+----------------------+----------
     Total |        52         22 |        74 

          Pearson chi2(1) =   4.8918   Pr = 0.027

.

[ajay pasi]

Hi Carlo, is my interpretation correct for chi square test for auto.dta ?

Result implies, distributions are not statistically different between who want a domestic and who want a foreign car. As the null hypothesis is rejected because chi2 p<0.05, where the null hypothesis is both distributions are independent.

regards,
ajay

[Carlo Lazzaro]

Correct.

[ajay pasi]

Thank you sir.