 | Democrats Eye a Congressional Majority | |
| A True Blowout is Now Possible | |
| Evolving History of the House and Senate | |
| 2006: What About the Senate? | |
| The GOP's Many 'Micro' Advantages |
| Rising Wage Gap, But No Squeeze | |
| Health Care, Not Social Security, the Third Rail of 2008 | |
| Will Democrats Keep the Faith? | |
| Turning Toward Iran | |
| Can Republicans Count on a House Snapback? |
April 25, 2006
2006: What About the Senate? Update
By Jay CostSeveral astute readers have written to offer an objection to the thesis I presented in yesterday's column. Namely, they argue that we might expect a Democratic capture of the House, but not the Senate, because the Republicans have a numerical advantage in this year's Senate contests.
Specifically, with a Democratic caucus numbering 45, the party "should" have to defend 15. However, they must defend 18. The GOP "should" have to defend 18.33 seats this year, but only have to defend 15. The reason for this is that each Senate seat falls into a class that determines when it comes up for election. This year's class is the "First Class," which currently over-represents Democrats and under-represents Republicans.
This is an interesting point, one that I could have and should have addressed in my original column. What this objection implies is that I have committed a "fallacy of composition." In other words, I have correctly asserted that individual senators are less secure than individual representatives, but I incorrectly inferred that, because of this, Senate partisan caucuses are less secure than House caucuses. The division of partisan seats on the table per year, which exists entirely independent of a party's current electoral position and is therefore effectively random, can (so the objection goes) nullify this dynamic.
I have two responses to this objection.
First, historically speaking, a mathematical disadvantage like this has occurred at least three of the seven times the House and Senate have both switched. The GOP took the Senate in 1952 despite a numerical disadvantage greater than that from which the Democrats suffer this year. The Democrats took it back in 1954 despite the same obstacle. The GOP also had a slight disadvantage in 1946 of the same type. This, however, does not amount to decisive evidence against this objection. In fact, decisive evidence does not exist - as we would have to compare a year that the parties were separated by at least the same number of seats (6) and the minority party had to run at least the same number of "extra" seats (3). These years do indicate, nevertheless, that numerical disparities, even ones more severe than 3, are by no means decisive.
This also raises another point. A numerical disparity is no oddity. We should expect, on average, each party to have to put up a number of seats that is proportional to how much of the Senate it controls. But we should also expect regular deviation from this amount - in other words, the luck of the draw will usually give one party or the other fewer seats to defend than they "deserve." A deviation of 3 seats is, historically speaking, normal and, again, not prohibitive for a capture.
Let us think a little bit more about this objection - for a decisive response might be found without recourse to a directly falsifying instance. What is being asserted is that the essentially random distribution of seats has soured the Democrats' chances. The Republicans, thanks to dumb luck, have fewer to defend, so they shall be secure. Here is the key question: is the number of seats the only factor affected by the random distribution of class assignment? There might be a whole host of other, qualitative factors that might work out differently. What is more, as we increase the number of factors that can work to the advantage of either, we decrease the probability that one party will enjoy a net advantage simply by random distribution.
What I am implying, in other words, is that the Democrats might enjoy qualitative advantages that their numerical disadvantage belies. This is intuitive. The quality of seats should make a difference, possibly a decisive one. If for instance, the Democrats were to run their 31 strongest incumbents, and the Republicans were only to run Santorum, we would say that the Democrats have a severe disadvantage in terms of numbers, but we would still expect them to net one seat!
There are at least three qualities that could vary according to the random distribution of seats per election. First, one party or another might have to defend seats that were previously won by a relatively small margin. This year, this favors the Republicans, but only technically. The median percentage of the vote of the entire Democratic caucus of the Senate is 61%. In this cycle, excluding open seats, it is 59%. For the GOP, the median percentage of the caucus is 58%. For this cycle, it is 63%. There is no real difference between candidates pulling 58% and candidates pulling 63%. All of them would be classified as strong-but-not-impenetrable. In this sense, then, it seems most prudent to conclude that the parties are equal on this measure.
Median vote might not be a sufficient gauge of whether the incumbents are of average quality. Incumbents can severely damage their electoral prospects after victory - usually by angering some portion of the electorate. There are five members of the Republican caucus who have undoubtedly done this - (in order of weakness) Burns, Santorum, Chafee, Bunning and Specter. The GOP should have to run, on average, only 1.67 of these incumbents per election. This year, they must run 3 - and they must run their three weakest. This, I think, is an incredibly severe qualitative disadvantage - given that (in my judgment) both Burns and Santorum are virtually unelectable.
Another qualitative difference also slightly favors the Democrats. The Democratic seats to be defended this year skew toward blue states. 16 Democrats come from states that Bush won in 2004. In a "neutral" year, we would expect the Democrats to defend 5.33 such seats. This year, they must defend 5. The GOP, meanwhile, has 9 senators that come from states Bush lost in 2004. We should therefore expect them to defend 3 such seats. This year, they are indeed defending 3. This amounts to a slight Democratic advantage (much less important than the issue of unelectable incumbents, but more important than the difference between median vote levels). The game is being fought on turf that is "bluer" than it would be in an entirely proportional year.
Thus, the luck of the draw this year favors the Democrats twice, favors the GOP once, and favors nobody once. What is the net result? Since we are dealing with more than just quantities, no simple or clear aggregation is possible. Nevertheless, it seems to me that it is virtually zero - which is exactly what we should expect in any given year. Given that these dis/advantages are the consequence of chance, we expect them to cancel one another out. In any given year, we should expect no party to have a real advantage due to the particular class up for election.
| Sponsored Links |