Forecasting Polarization in the 113th Senate

Posted on September 10, 2012 by voteview

As can be seen in the plot below, polarization has markedly increased in both chambers of Congress, but the Senate has remained less divided than the House. This does not mean that the Senate has been a more productive chamber (notably, it has not passed a budget in over three years). However, there are institutional explanations for gridlock in the Senate (e.g., a supermajority is required to invoke cloture) and, at least relative to the House, moderates have proven more resilient. There are several moderate, Blue Dog Democratic Senators in the South and Great Plains (Sens. Mary Landrieu (D-LA), Ben Nelson (D-NE), Bill Nelson, and Mark Pryor (D-AR) and moderate Republicans in the Northeast and Midwest (Sens. Scott Brown (R-MA), Susan Collins (R-ME), Mark Kirk (R-IL), and Olympia Snowe (R-ME)). Consequently, the ideological distribution of the 112th (current) Senate is less skewed towards the liberal and conservative extremes as the 112th House (using Common Space scores to allow for comparability across chambers).

Click images to enlarge

spacer

spacer

However, our research suggests that polarization — that is, the distance between the means of the Republican and Democratic Party means — will continue to worsen in the 113th Senate. Because most of the challengers in competitive Senate races are also members of the House (e.g., Reps. Todd Akin (R-MO), Danny Rehberg (R-MT), and Shelly Berkley (D-NV)), we can use Common Space DW-NOMINATE scores to compare their ideological positions to other Senators. For competitive candidates who do not have a congressional roll call voting record (e.g., Ted Cruz (R-TX) and Elizabeth Warren (D-MA)), we impute Common Space DW-NOMINATE scores by regressing the scores of members of the 112th Senate onto their state and personal characteristics. We then use the results to predict the scores of these Senate candidates.

Below are the results from two separate regressions (one for Democrats, the second for Republicans). In early models we included state median income and the Senators’ gender, but F-tests showed that these variables did not significantly contribute to the overall fit of the model (though gender did matter for Republicans). Not surprisingly, the most important factor is the partisanship makeup of the state (the percentage of the states’ 2008 two-party vote for Obama). The bluer the state, the more liberal the Senator. Female Republican Senators are significantly more moderate (b = -0.11, p <= 0.05), and Republican Senators affiliated with the Tea Party are significantly more conservative (b = 0.30, p <= 0.05).

DEMOCRATS:

. regress dwnom obamavote

      Source |       SS       df       MS              Number of obs =      53
-------------+------------------------------           F(  1,    51) =   25.02
       Model |  .183509575     1  .183509575           Prob > F      =  0.0000
    Residual |  .374091912    51  .007335136           R-squared     =  0.3291
-------------+------------------------------           Adj R-squared =  0.3160
       Total |  .557601488    52  .010723106           Root MSE      =  .08565

------------------------------------------------------------------------------
       dwnom |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   obamavote |  -.0073872   .0014769    -5.00   0.000    -.0103523   -.0044222
       _cons |   .0890863   .0841796     1.06   0.295    -.0799113    .2580839
------------------------------------------------------------------------------

REPUBLICANS:

. regress dwnom obamavote female teaparty

      Source |       SS       df       MS              Number of obs =      47
-------------+------------------------------           F(  3,    43) =   24.06
       Model |  .878020419     3  .292673473           Prob > F      =  0.0000
    Residual |   .52298669    43  .012162481           R-squared     =  0.6267
-------------+------------------------------           Adj R-squared =  0.6007
       Total |  1.40100711    46  .030456676           Root MSE      =  .11028

------------------------------------------------------------------------------
       dwnom |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   obamavote |   -.009166    .002154    -4.26   0.000    -.0135099    -.004822
      female |   -.109835    .054585    -2.01   0.050    -.2199161    .0002461
    teaparty |   .2996521   .0459595     6.52   0.000     .2069661    .3923382
       _cons |   .8304714   .0977509     8.50   0.000     .6333379    1.027605
------------------------------------------------------------------------------

Based on these results, we plug in the needed values to calculate predicted Common Space scores for the candidates. This is an imprecise method, but given this only affects a handful of candidates with no prior congressional service, is probably acceptable, especially since we are taking an aggregate measurement. Likely the biggest jump we make is in assigning a Common Space score to Angus King, the former Independent Governor of Maine and frontrunner for the seat of retiring Senator Olympia Snowe (R-ME). Gov. King is widely assumed to caucus with the Democrats, and we assign him the same score as former Republican-turned-Independent and fellow New Englander Jim Jeffords (I-VT), whose Common Space score is -0.309. This may seem too liberal a score for Gov. King, but a review of the issues page of his campaign website show that his positions are generally left-leaning, especially on the environment and social/cultural issues.

Next, we include the scores of all returning Senators with no opposition and those running for re-election in non-tossup races (as rated by Roll Call). We then consider three scenarios to determine the winners of toss-up Senate seats – one in which the Democrats win all a class=”www.rollcall.com/politics/2012_race_rating_map.html”>seven toss-up seats, one in which the Republicans win all of the toss-up seats, and one in which the current polling leaders win (the set of most likely outcomes) (*note: we use this helpful page from RealClear Politics, but change the Connecticut race outcome because of the “Likely Dem” rating from Roll Call).

In the plots below, we compare the predicted ideological distributions of the 113th Senate with those of the current, 112th Senate. In all three scenarios, the polarization measure (the distance between the Democratic and Republican means) is higher in the 113th Senate. The jump in polarization is not huge, but these results do suggest that the next Senate is not likely to be any more productive than the current Senate and will instead be somewhat more gridlocked.

Click images to enlarge

spacer

spacer

spacer

Much of the predicted growth in polarization is attributable to the retirement or electoral vulnerability of Senate moderates. This includes Ben Nelson’s (D-NE) likely replacement by conservative Deb Fischer, Scott Brown’s (R-MA) potential replacement by Elizabeth Warren (D-MA), and the likely ascendance of the Tea Party-supported candidates Ted Cruz in Texas and Richard Mourdock in Indiana (replacing the more moderate Republican Sens. Kay Bailey Hutchinson (R-TX) and Richard Lugar (R-IN), respectively.

Tweet
This entry was posted in Contemporary American Politics, Political Polarization. Bookmark the permalink.

Comments are closed.