Robustness checks when outliers considered

[Marthe Dpnt]

Dear all

I have to replicate a paper and take into account outliers and do robustness checks. But I was novice with econometrics and stata. First, I don't know how I can integretate outliers since for me, the authors seem not to drop outliers.
This is the following paper :
https://www.aeaweb.org/articles?id=1...page=21&from=j
Second, robustness checks mean boxplot and so on ?
Thanks in advance !

[Felix Bittmann]

The first question seems to be a conceptual one. If the original study does not define and handle outliers, it is up to you to set up a definition (hint: there is no universally accepted definition of outlier). Searching the Forum, you will find many related discussions.

Regarding the second question, diagnostics depend on the model you run. For example, for a linear regression (regress), the following website might be very helpful:
https://stats.idre.ucla.edu/stata/we...n-diagnostics/
Again, there is no universally accepted set of tests to run. This depends on your field of study and probably also the journal you want to publish in. The link above is probably a good point to start.

[Carlo Lazzaro]

Marthe:
welcome to this forum.
As an aside to Felix's helpful advice, you should also define beforeheand robustness with respect to what. Expected results (which is often a dramatically poor approach), model misspecification or else?
Outliers (if they are not self-apparent results of mistaken data entry) are often a matter of fact and you have to live with them..
That said, if you are dealing with -regerss-, you may find -help lvr2plot- and related implications helpful for your research.