0

I am a brand new Stata user and I am not sure I understand the print for my logit. I have the following command in Stata:

logit won i.dum1##(c.srto)

dum1 is a dummy variable 1 or 0 srto is a continuous variable

I get the following result:

------------------------------------------------------------------------------

>          won |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
> > -------------+----------------------------------------------------------------
> >       1.dum1 |   .1482421   .0939667     1.58   0.115    -.0359293    .3324135
> >         srto |  -.0097423   .0006258   -15.57   0.000    -.0109689   -.0085157
> >             |
> >  dum1#c.srto |
> >           1  |      -.054   .0047478   -11.37   0.000    -.0633056   -.0446944
> >              |
> >        _cons |   -2.02613   .0493487   -41.06   0.000    -2.122852   -1.929408
> ------------------------------------------------------------------------------

I am interpreting this that dum1 effects srto to be -.054 when 1 and -.009 with a difference of .043 is greater when dummy1 =1. Therefore, srto has more effect when the dum1 =1. Since the p value is 0 for both of these they are significant. Secondly, the dum1 srto confidence intervals don't overlap with the srto which makes it a stronger effect. My question is this the correct way to read this result or is the -0.54 the margin or difference between dummy on and dummy off?

done_merson
  • 111
  • 4
  • You can find explanations of how to interpret interaction effects [here](https://stats.stackexchange.com/questions/284888/interpretation-of-interaction-effect-in-multiple-regression) and especially [here](https://stats.stackexchange.com/questions/137856/how-to-interpret-interaction-between-a-dummy-and-a-continuous-variables-in-ols). It might be easier to understand the model if you write it in the form of $logit(y) = \beta_0 + \beta_1x_1+\beta_2x_2+\beta_3x_1x_2 + \varepsilon$ and substitute coefficient estimates. Also, please note that logits might not be very useful for the interpretation. – T.E.G. Sep 03 '17 at 03:36
  • [This](https://stats.stackexchange.com/a/89980/7071) would be worth a close read. – dimitriy Sep 05 '17 at 20:39

0 Answers0