lalonde is an oft-used data frequently used for in statistical software packages. This data was drawn from the PSID and National Supported Work Demonstration.

Subsamples of the dataset are included with the R language's MatchIt and Matching packages.

I am using the data from these packages, yet I note that while one of the variables -- treat is defined as

"the treatment assignment (1=treated, 0=control)."

it doesn't actually seem to say what the treatment was.

Can anyone help me determine this?


This dataset was first used by Robert LaLonde in a paper published in the American Economic Review 1986: Evaluating the Econometric Evaluations of Training Programs with Experimental Data.

The "treatment" refers to participation in the National Supported Work Demonstration (NSW), an employment program to help disadvantage to move into the labour market. Eligible applicants were assigned randomly into this program. So the "treated" are those assigned to the program and the "controls" those not assigned.

But be careful, Lalonde has used several control groups. One groups stems from the experiment mentioned above other are "non-experimental" and stem from the Panel Study of Income Dynamics (PSID) an the Current Population Survey (CPS).

  • Do you know of an R package which also contains data of the non-experimental control groups? Nov 15 '16 at 7:40
  • 1
    No, but all the datasets are available via the following website, in txt and dta (Stata) format: users.nber.org/~rdehejia/data/nswdata2.html It is straightforward to read these formats in R. For the dta format you will have to load the foreign package. Nov 15 '16 at 17:11

I determined that treat represents participation in a 1970's job training program by reading the original paper by LaLonde (1986).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.