Unequal-sized groups regression

[Maemorine2]

Hello,

I am performing a general probit regression model on the whole dataset but want to look at dissimilarities in coefficients between men and women.
I have got 2608 observations, with 1297 obs. for men and 1311 obs. for women.

My question is, can I split the sample between men and women and compare the regression coefficients even though both groups are not exactly the same size?
If not, how can I do it ?

Thanks in advance

[Nick Cox]

That difference in sample size is not a barrier to what you suggest.

I guess most researchers would prefer to put emphasis on working with a single model in which gender was an indicator variable, with scope for examining interactions explicitly.

Please note the request here to use full real names. You can change your identifier by contacting the administrators. See "Contact us" on the forum home page.

[Phil Schumm]

As Nick says, the unequal sample sizes is not a problem. Be careful, however, because comparisons of the coefficients between men and women in a model like this (e.g., probit, logit, etc.) depend on the assumption that the underlying variance is the same in both groups. The list archives are replete with posts describing this issue.