James Fowler (who earlier found that nicer people are more likely to vote) has a new paper on “who is the best connected legislator.” Here’s the abstract to Fowler’s paper:

Using large-scale network analysis I map the cosponsorship networks of all 280,000 pieces of legislation
proposed in the U.S. House and Senate from 1973 to 2004. In these networks a directional link can be
drawn from each cosponsor of a piece of legislation to its sponsor. I use a number of statistics to describe
these networks such as the quantity of legislation sponsored and cosponsored by each legislator, the
number of legislators cosponsoring each piece of legislation, the total number of legislators who have
cosponsored bills written by a given legislator, and network measures of closeness, betweenness, and
eigenvector centrality. I then introduce a new measure I call ‘connectedness’ which uses information
about the frequency of cosponsorship and the number of cosponsors on each bill to make inferences about
the social distance between legislators. Connectedness predicts which members will pass more
amendments on the floor, a measure which is commonly used as a proxy for legislative influence. It also
predicts roll call vote choice even after controlling for ideology and partisanship.

Once again, the “I” above is Fowler, not me. Now for some quick comments. Really quick, since I have time only to skim the paper, not to evaluate its method.

– First off, all the tables could (and should, in my opinion) be graphs. I know I’ve said this before, but it’s an important point. Making graphs takes work, but nothing compared to the effort of doing the research and writing the paper. To really see what’s going on with patterns, time trends, etc., you need a graph.

– Table 1 should be a set of time series graphs. y-axes could be on a logarithmic scale (since all outcomes are inherently positive), which would allow House and Senate time series to be shown as two lines (e.g., solid for House, dotted for Senate) on each graph. Thus, 7 little graphs vs. time. Could easily be placed within less space than the existing table. For example, on a 3×3 grid.

– Table 2 could be folded into the Table 1 display–2 more little time series graphs–or placed separately if you want. Again, would take less space! than the original table, if done right.

– Table 3 . . . this is OK, I guess. But what I’d really like to see is how stable these are over time. What I suggest is a time-series graph, showing the rankings over time for several prominent legislators (e.g., Tip O’Neill, Orrin Hatch, …) over this time period, so that we’d get a sense of how stable these measures are. You could probably legibly display about 5 of these time-series lines on a single plot. So how about 4 little plots showing 5 Dem House members, 5 Rep House members, 5 Dem Senators, 5 Rep Senators?

– Tables 4 and 5 . . . see Table 3

– Figures 3 and 4: These are pretty but you can do a little better by clearly distinguishing the parties. Perhaps open circles for Reps and solid circles for Dems?

– Tables 6, 7, 8: See comments on Table 1. Display as time series.

– Some other analyses I’d like to see: typical career paths: do congressmembers increase in connectedness during their careers? I’m surprised there isn’t a simple regression of connectedness on some legislator predictors: party, #years in congress, indicator for committee chairmanship, etc. You could run a separate regression for each Congress and then plot these estimated coefficients over time. (That’s the statistical technique called the “secret weapon.”)