How do I interpret the coefficient of this interaction variable. Should I just be setting my dummy to 1 instead?

I have a dataset which looks at people's number of visits to the doctor. It includes a dummy variable for gender called female and a dummy variable called hhkids which states whether or not someone has children. I want to look at whether women's visits to the doctor are more affected by children than that of men.

This was my original regression which didn't include gender:

areg  docvis hhkids age agesq married working linc addon, absorb(id)


Then I created an interaction variable and added it to my regression:

gen fekid = female*hhkids
areg  docvis hhkids age agesq married working linc addon fekid, absorb(id)


Here is the output:

    Linear regression, absorbing indicators                Number of obs =    6209
F(  8,  5314) =    8.25
Prob > F      =  0.0000
R-squared     =  0.4187
Root MSE      =  4.5747

------------------------------------------------------------------------------
docvis |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
hhkids |   .7493504   .2887167     2.60   0.009     .1833471    1.315354
age |  -.2326124   .1003786    -2.32   0.021    -.4293957   -.0358292
agesq |   .0038731   .0010802     3.59   0.000     .0017553    .0059908
married |  -.0923826   .3659839    -0.25   0.801    -.8098612     .625096
working |  -.5702973   .2491114    -2.29   0.022    -1.058658   -.0819367
linc |   .0886328     .23889     0.37   0.711    -.3796897    .5569553
addon |   .3009833   .6369426     0.47   0.637    -.9476857    1.549652
fekid |  -.1577091   .4279726    -0.37   0.713    -.9967111    .6812929
_cons |   5.793355   2.426897     2.39   0.017     1.035641    10.55107
-------------+----------------------------------------------------------------
id |      F(886, 5314) =      3.929   0.000         (887 categories)


I was wondering firstly if i'm going about this the right way, and then secondly how to interpret the coefficients in the output. How do i determine whether women are more affected by having kids than men are?

Additionally, would it be better to just have female==1 in the regression instead of using the interaction variable?

Typically, when you include an interaction variable you want both components of it also to be part of the regression. Otherwise your coefficients will be hard to interpret. If you run the regression as it is, the coefficient on the interaction term on fekid captures how much less women with kids go to the doctor, compared to the general population with similar amound of kids, age etc. (but not controlling for gender, so compared to a mix of men and women who are otherwise similar)

If you include female in your regression (but also keep fekid), the coefficient on fekid captures what you want: How much having kids increases the probability of going to the doctor for women compared to how it affects men.

If you only include female in your regression, the coefficient on that will just capture how much more/less women go to the doctor compared to men.

Checking areg, it appears you use panel data with fixed effects. In that case you don't need to include female as it's absorbed in the fixed effect. So if that is really what you do your approach is correct :)

In addition to these remarks, depending on how many doctor visits your typically have in your sample, running OLS might not be very appropriate, as your dependent variable is categorical. Check whether running ordered logit/probit makes a difference. (can maybe be messy because of the fixed effects)

• Thank you very much! – summer_ZUGG Mar 29 at 21:42
• I have an additional question. Is it correct to interpret from this that women are more affected by having children than men? That women with children are 15.77% less likely to go to hospital than men with children? – summer_ZUGG Mar 29 at 21:49
• In principle yes but your confidence interval is very wide and in particular includes zero, so while the point estimate points in that direction you cannot say with confidence whether women are affected more or less than men. – user1587692 Mar 30 at 0:19