# Re: st: complex difference-in-difference approach with e.g. economic sectors

 From "Austin Nichols" <[email protected]> To [email protected] Subject Re: st: complex difference-in-difference approach with e.g. economic sectors Date Sat, 9 Feb 2008 09:37:39 -0500

```Pascal Sulser <[email protected]>:
If I understand your description, regression (2) exhibits perfect
collinearity and one regressor would be dropped, since
female = sector1_female + sector2_female + sector3_female
so (1) is what you would be estimating.  It is usually helpful in
these models to imagine a single real observation and mentally plug in
ones and zeros into your regression equation to see what coefficients
mean.

On Feb 9, 2008 7:29 AM, Pascal Sulser <[email protected]> wrote:
> Hi,
>
> Does anyone know the correct specification for a difference-in-
> difference (DID) approach with e.g. economic sectors included. In a
> standard regression (no DID) I use e.g.
>
> "reg unemployment_rate age age2 sector1 sector2"  --> assuming the
> base group is working in the tertiary sector
>
>
> Now, if I want to do a difference-in-difference approach to see
> weather there exists a significant difference between male workers and
> female worker, which of the two variants is the correct to chose,
> assuming there are differences between all independent variables?
>
>
> 1. reg unemployment_rate female age age_female age2 age2_female
> sector1 sector1_female sector2 sector2_female
>
> OR
>
> 2. reg unemployment_rate female age age_female age2 age2_female
> sector1 sector1_female sector2 sector2_female sector3_female
>
>
> I am not quite sure whether regression 1. is the right one assuming
> that if female==1 then this means she is a women working in the
> tertiary sector or whether I have to add that interaction term, too.
>
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```