How to calculate Cramer's v by group, or using by() command

[Chian Jones Ritten]

Hello,
I am trying to calculate Cramer's V statistic for the same variable for different groups. For example, I am trying to calculate the effect size of height between women and men. Thus, the variable of interest is one variable (height) by two groups (women v men). I have tried to use the by() command, but I receive the "option by() not allowed" error. I have tried variations of the following command that have given me an error:

tabulate height, by(gender) V

Does anyone have experience on how to include the by() command when calculating the Cramer's V statistic? Thank you!

[Chian Jones Ritten]

And just for more information, the height variable is categorical.

[Nick Cox]

You won't get that out of a one-way tabulation, or even repeated one-way tabulations. It's only possible with a two-way tabulation.

Note that even as a measure of association, this one ignores the ordering, important here for rep78 and important for your height variable.

. sysuse auto , clear
(1978 Automobile Data)

. tab fore rep78, V

           |                   Repair Record 1978
  Car type |         1          2          3          4          5 |     Total
-----------+-------------------------------------------------------+----------
  Domestic |         2          8         27          9          2 |        48
   Foreign |         0          0          3          9          9 |        21
-----------+-------------------------------------------------------+----------
     Total |         2          8         30         18         11 |        69

               Cramér's V =   0.6286

Note that by() is an option, not a command.

[Chian Jones Ritten]

Thank you so much for your response. It works perfectly!