Social Science Statistics Blog

Is the data behind some of this work available in an easy to use (ideally packaged up in R) format? I poked around a bit and couldn't find anything simple.

The reason I ask is that I hope to teach a course next spring which would involve the students learning about statistical graphics, a la Tufte and using R. Any project involving states and counties lends itself to pretty graphics. I am looking for examples to use in the assignments for the class.

see this paper and the associated replication data set.
Gary

Thanks for the citation. Comments:

1) Even though I entered my e-mail address to get updates on this thread, I am not getting them.

2) ICPSR does not make it easy to find things even when you know what you are looking for.

3) Does that replication really still work? I tried running the JudgeIt program that is there on my XP box, but it didn't seem to do anything. No need to walk me through it since I will not be able to teach the students a package other than R for this class, but I suspect that the passage of time has made a decade-old result not easily replicable, unless you have access to old hardware/OS. Another reason to use R for everything!

4) Recall that my question was:

Is the data behind some of this work available in an easy to use (ideally packaged up in R) format?

I think that the answer is "No." Please correct me if I am wrong. Again, my goal is to assign the *recent* papers that Gary and others have written and get students to play with data from those papers (especially the amici brief and the student version), ideally making some pretty graphics in R along the way. For those who care, a draft of the syllabus is here.

David: Gary and I have almost finished an R implementation of JudgeIt that can easily handle a large number of elections simultaneously - hence its name "Judgem". The package will have a few easy-to-use data sets included. ICPSR House 6311 is already loaded up in R format; you can download it here if you want a sneak peek before the package is ready.

Thanks!

For those still not converted to R, note how easy it is to grab and load up this data:

>load(url("http://www.fas.harvard.edu/~acthomas/house6311.RData"))

Alas, I had trouble understanding much of the data, no doubt because I lack experience with it. For those following along at home, here is a sample.

> head(house6311[[1]])
STATE DIST INC VOTE TURNOUT DELSOUTH
1 1 1 NA 0.28 38482 0
2 1 2 NA 0.38 58261 0
3 1 3 NA 0.33 22934 0
4 1 4 NA 0.34 46381 0
5 2 1 NA 0.31 28119 0
6 2 2 NA 0.27 30842 0
> summary(house6311[[1]])
STATE DIST INC
Min. : 1.0 Min. : 1.0 Mode:logical
1st Qu.:14.0 1st Qu.: 3.0 NA's:357
Median :31.0 Median : 6.0
Mean :30.6 Mean :11.4
3rd Qu.:45.0 3rd Qu.:11.0
Max. :73.0 Max. :98.0

VOTE TURNOUT DELSOUTH
Min. : 0.000 Min. : 3438 Min. :0.000
1st Qu.: 0.370 1st Qu.:26036 1st Qu.:0.000
Median : 0.467 Median :37012 Median :0.000
Mean : 0.488 Mean :34636 Mean :0.224
3rd Qu.: 0.565 3rd Qu.:43959 3rd Qu.:0.000
Max. : 1.000 Max. :75683 Max. :1.000
NA's :31.000 NA's : 31
>

Questions:

1) Do you have a timeline for when the package will be ready to go?

2) Do you have a sense of how connected this will be to the recent papers that would be appropriate for non-statistical undergraduates? My idea would be to have the undergrads read the papers and then "play" with some of the data in terms of making pretty graphics. For example, how does partisan asymetry vary across states? I suspect that this is not the primary audience that you have in mind.

Thanks again for the link.

David: Hopefully it'll be fully ready within the next month. We're just debugging some things.

I'm also putting together a data set of state legislatures for a related paper, from which we can get partisan bias before and after redistricting. It'll all be in the package release.

Our group released a study on gerrymandering today that you might find of interest.

freebuffalo.org