Hi Vipin and Rebecca, Sounds like an interesting paper. Following Ian's earlier comments on Justin & Clayton's abstract, I think perhaps you could omit the third to last sentence (b/c of specific criticism of authors) and maybe even the second to last (b/c it may be getting into smaller issues better covered in the meat of the paper). Great title! Best, Abby -----Original Message----- From: gov2001-l-bounces(a)lists.fas.harvard.edu [mailto:gov2001-l-bounces@lists.fas.harvard.edu]On Behalf Of ghumphr(a)fas.harvard.edu Sent: Sunday, April 23, 2006 8:17 PM To: gov2001-l(a)lists.fas.harvard.edu Subject: Re: [gov2001-l] Abstract For that sort of academic paper, you might want to look at the geneses of nations. Quoting Vipin Narang <vnarang(a)fas.harvard.edu>du>: _______________________________________________ gov2001-l mailing list gov2001-l(a)lists.fas.harvard.edu http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l _______________________________________________ gov2001-l mailing list gov2001-l(a)lists.fas.harvard.edu http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l

Hi All, With apologies for sending this out at this late hour, tonight in office hours, a few questions came up that I wanted to clarify for the list. *Not to worry: if you have already done these problems a different way based on your reading of the question, no worries, just be clear about what you did under what assumptions.* But for those still working, this might prove helpful. 1a) As we mentioned in an earlier email, you do not need to use covariates in question 1a--the mention of covariates was to provide a more general sense of how to estimate ATE for the problem set as a whole. 1c-e) In these questions, we had envisioned using only those assigned to control as the control group--and *not* those who were assigned to treatment but didn't receive the treatment. We can discuss why in section, but a simple way to think about it is this: those who were assigned to treatment but didn't receive it are indeed systematically different from those who did receive the treatment, and so wouldn't serve as good controls. Ignorability doesn't hold. 1d) By all "multiplicative interactions," we mean the one four-way interaction, the various three-way and two-way interactions, and the various squared terms. 1e) For the probit model on the matched dataset, simply using the four covariates (and not interactions) is fine. Good night, and good luck. Best, Dan ---- Ph.D. Student Department of Government Harvard University Tutor, Currier House dhopkins(a)fas.harvard.edu http://www.danhopkins.org _______________________________________________ gov2001-l mailing list gov2001-l(a)lists.fas.harvard.edu http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l

Justin, I encountered a similar problem a couple of weeks ago. You can find the snow package when you go directly to any of the CRAN mirrors (e.g. http://cran.xedio.de/) and then click on "contributed extention packages." After installing it snow reports a version conflict, though, and I didn't get it to work. It might be worth trying again now, as the new R version 2.3.0 has come out a couple of days ago. Please let me know if you get it to work; I would very much like R to use multiple threads. best, Holger Justin Ryan Grimmer wrote: