voteview / WebVoteView

Webbased rollcall vote visualization software
MIT License
35 stars 3 forks source link

Nomenclature for second dimension #364

Open aaronrudkin opened 5 years ago

aaronrudkin commented 5 years ago

I am going to write some code that relabels the second dimension depending on the congress you're in. For congresses generally regarded as unidimensional, it'll read something like "Other / Off Dimensional vote" and for historic congresses with a useful second dimension, it'll read whatever it should read.

Can someone check Ideology and Congress and confirm for me the exact congresses I should be labeling what? Obviously the three main 2d periods are Civil War / Slavery, Gold/Silver Metallism, and Civil Rights / Social Issues, but I'm not sure the exact cutoff.

If we could run a summary statistic for the average dimensionality of a congress (in terms of additional PRE from the second dimension, maybe? I seem to remember that's what Howard and Keith did in the second on testing higher dimensionality) then we could also sort of deduce this in a data-driven way rather than qualitatively.

I'll implement the code now and close the issue when I hear back about where to put the labels.

aaronrudkin commented 5 years ago

This was easier than I thought. Code is ready. It's live on dev. Check between the 25th and 32nd congresses versus any other congress and you'll see the difference.

In Ideology and Congress, Poole and Rosenthal write: "[In the House,] before the formation of the Democratic-Whig party system in the mid-1830s, the public works issue is the one most picked up [by 2nd dim]... After 1837, slavery and the disposition of public lands... until the fateful 32nd Congress and the compromise of 1850, after which slavery becomes [1st dim]... From the Civil War until the realignment of the 1890s, the [2nd dim is mostly] bimetalism... there is [no 2nd dim issue until] civil rights issues split the Democratic Party and created the three-party system..."

And then there's a lengthy section about how in the Senate there are lots of second dimension issues.

Okay, so having said that, someone other than me should pick labels for ranges of congress and I'll implement in code, with a default to "Other Issues" or "Off-Dimensional Votes" for the not specifically enumerated stuff.

JeffreyBLewis commented 5 years ago

I think one possibility for labeling is to look at the topics of the votes that strongly divide the members along the 2nd dimension in each congress and what fraction of the votes that is. Of course, we need to think what "strongly" means. One way to think about that is the how "horizontal" the cutline line and another would be the classification of the votes by the cutline (or loglik, GMP, or whatever). I can take a little look into how we might do this.

On Thu, Jan 24, 2019 at 3:11 PM Aaron Rudkin notifications@github.com wrote:

I am going to write some code that relabels the second dimension depending on the congress you're in. For congresses generally regarded as unidimensional, it'll read something like "Other / Off Dimensional vote" and for historic congresses with a useful second dimension, it'll read whatever it should read.

Can someone check Ideology and Congress and confirm for me the exact congresses I should be labeling what? Obviously the three main 2d periods are Civil War / Slavery, Gold/Silver Metallism, and Civil Rights / Social Issues, but I'm not sure the exact cutoff.

If we could run a summary statistic for the average dimensionality of a congress (in terms of additional PRE from the second dimension, maybe? I seem to remember that's what Howard and Keith did in the second on testing higher dimensionality) then we could also sort of deduce this in a data-driven way rather than qualitatively.

I'll implement the code now and close the issue when I hear back about where to put the labels.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/voteview/WebVoteView/issues/364, or mute the thread https://github.com/notifications/unsubscribe-auth/AEinZdjqcZaD_S5TSxmyVvm860yd15l3ks5vGj2GgaJpZM4aR5AJ .

--

Jeffrey B. Lewis Professor, Department of Political Science University of California, Los Angeles BOX 951472, 4289A Bunche Hall Los Angeles, CA 90095-1472