General Assembly party loyalty rankings.

1. I like the Senate one a lot. In the House, I would use Speaker Howell instead of Griffith. Griffith has a libertarian streak that a lot of Republicans do not have- and I could be wrong on this, but I bet you would find a lot more votes where Howell and Moran differed than votes where Griffith and Moran differed.

2. Or if not more votes, better votes to judge by. That the 2nd place Republican shows a 34 probably shows a little problem with comparing to Griffith.

3. It’s not clear to me whether the House Republican numbers indicate a) generally poor party loyalty in the House b) that Griffith doesn’t do a very good job c) that I should choose a different person to define the party standard or d) all of the above.

4. How interesting to see that Putney and Abbitt, as Independents, vote with the Republicans a lot. I would have guessed as much, but didn’t know for sure.

   Why does someone declare themselves to be independent if they vote with one side or another most of time?

5. Although I can hardly see someone sitting down with a score sheet, to determine if they ‘look’ independent.

6. It would be interesting to see a wheel graph, or something like it, to determine the ‘temperature’ of each of the legislators.

   For instance, Watkins Abbitt votes with Republicans more than 20 R’s in this chart.

   Do you see what I am getting at here?

   Timelines sound like what I am thinking of, but more about party ID, not so much time.

7. The benefit of the approach that you describe, Mark, is that it makes it possible to bury the vagaries of this somewhat, lest somebody try to do something absolute with the numbers. I’d be relatively confident taking these numbers here now, shaving off the top and bottom person from each chart (the people from whom we establish the extremes), recalibrating with that new range, and showing an little graph where I just mark a spot, from left to right, indicating approximately where this legislator sits.

8. Fascinating work here, Waldo. Thanks!

9. Kind of like the Political Compass thingies that we see around, right?

10. As I see it, recalibrating both House and Senate scales to 100 make them incomparable to each other, right?

    It might be more complicated, but shouldn’t we measure party loyalty by consensus of the party instead of the vote of their majority leader (although I find nothing wrong with everyone listening to Brian Moran more often)?

11. I agree this is an interesting analysis. But a word or two of caution. Neither Griffith nor Stosch are “in charge” in practical terms of their party votes — and the same goes for Moran and Saslaw. In fact, there are very few “party line” or “caucus line” votes and those tend to be measured by a handful or two each session.

    More useful might be an analysis of vote clusters, using maybe some form of regression analysis to group legislators into blocks. It’s only a hypothesis, but I suspect it would show a couple of blocks of a third to 40% in each house that vote together the vast majority of the time and a “middle” that might be bigger in the Senate than the House.

12. I think a better measure might be judging votes where half of the Republicans voted against half of the Democrats and then seeing how often each member was with her party on such votes.

13. As I see it, recalibrating both House and Senate scales to 100 make them incomparable to each other, right?

    That’s quite right.

    It might be more complicated, but shouldn’t we measure party loyalty by consensus of the party instead of the vote of their majority leader (although I find nothing wrong with everyone listening to Brian Moran more often)?

    That would absolutely be better, but I’ve had little luck with properly incorporating statistical data packages into the MySQL/PHP setup to calculate such things. I regard this current step as interim, since I’m just not getting anywhere with a more detailed analytical process.

    I think a better measure might be judging votes where half of the Republicans voted against half of the Democrats and then seeing how often each member was with her party on such votes.

    Party line floor votes, as Observer points out, are just really rare. There aren’t enough of them to do any sort of meaningful calculation based on them.

14. Waldo,

    d). All of the above!

15. Though this may fail at clearly showing party loyalty, it still does a decent job showing loyalty to party leaders. That’s pretty interesting on its own.

16. Maybe you should use the algorithm used here, if you can find a good way to do it.

17. Whoa. Though that’s clearly the right way to do it, KC, it’s way over my head. Richmond Sunlight needs a statistician.

Comments are closed.