Voters’ Ideological Perceptions of 2014 Senate Candidates

In this post we use Bayesian Aldrich-McKelvey scaling to analyze voters’ perceptions of the ideological positions of Senators and Senate candidates who will be running in close races in 2014 (we also describe the Bayesian Aldrich-McKelvey scaling method in our book Analyzing Spatial Models of Choice and Judgment with R. To do so we use data from the 2012 Cooperative Congressional Election Study (CCES). The 2012 CCES asked respondents to place national figures like President Obama and Mitt Romney (as well as the Democratic/Republican Parties, the Tea Party, and the Supreme Court), their Senators and Representatives on a seven-point ideological scale ranging from extremely liberal to extremely conservative.

Bayesian Aldrich-McKelvey (BAM) scaling corrects for bias in respondents’ placements (for example, when a liberal respondent overstates the conservatism of political figures or vice versa) to create measures of citizens’ ideological perceptions of political figures that are comparable across states and districts. We have previously used BAM to show that raw liberal-conservative self-placement data understates the true level of polarization in the electorate.

Below we plot the BAM estimates of the liberal-conservative positions of incumbents and challengers (if they were included in the 2012 CCES) in selected 2014 Senate races. One of the advantages of using 2012 data is that we can also estimate the positions of Senatorial candidates who were unsuccessful that year; in particular, former Rep. Todd Akin (Missouri) and Richard Mourdock (Indiana). Both lost very winnable races after making controversial statements about abortion and rape. One of the big questions heading into 2014 is whether the Republican Party will again let races slip through their fingers by nominating candidates perceived to be too conservative and out of the mainstream.

According to the BAM estimates, the likely Republican nominees in close races like Arkansas and Louisiana, safer races like Montana and West Virginia, and greater long-shots like Colorado and New Hampshire are all perceived to be more ideologically moderate than the Tea Party in general or Akin and Mourdock specifically. In fact, perceptions of all of these candidates are virtually indistinguishable from those of Mitt Romney, with the exception of former Sen. Scott Brown (R-MA), who is located much closer to the center.

All of the Democrats are perceived to be closer to the center than President Obama, with Sen. Jeanne Shaheen (D-NH) as the most liberal of the group and Sen. Mark Pryor (D-AR) as the most moderate. Interestingly, Sen. Mary Landrieu (D-LA) is perceived to be slightly more liberal than Sen. Kay Hagan (D-NC) and considerably more liberal than Pryor by Louisiana voters. This does not bode well for her in her race for a fourth term. There is also the question of how much these races will be a referendum on the Senators personally rather than on national conditions like the economy, presidential approval, and the Affordable Care Act.

Click image to enlarge

Note: the correlations between the BAM scores and DW-NOMINATE Common Space scores are moderate-high, although the samples size is very small: r = 0.96 overall, r = 0.49 for Democrats, and r = 0.75 for Republicans.

Posted in Contemporary American Politics, Issue Scaling, Political Polarization | Leave a comment

An Update on the Presidential Square Wave through 2013

Below we plot the first dimension DW-NOMINATE Common Space scores of the presidents in the post-war period, which we refer to as the “presidential square wave” due to its shape. An ideological score is estimated for each president throughout the entirety of their tenure in office by scaling their “votes” on a subset of roll call on which they announce a position (measured using CQ Presidential Support Scores). Negative scores indicate greater liberalism and positive scores indicate greater conservatism. The presidential scores are directly comparable across time and with members of Congress.

Very little has changed from the last presidential square wave. President Obama fits the spatial model estimated by DW-NOMINATE extremely well, with over 95% of his “votes” correctly classified. Obama has moved very slightly leftward (-0.363) and is now just to the left of LBJ (-0.346) and right of Truman (-0.368), though this trio is virtually ideologically indistinguishable. President Eisenhower is the most moderate president (0.302) of the post-war era.

Among members of the 113th Congress, President Obama is very ideologically close to Representatives John Garamendi (D-CA) [-0.362], Marcy Kaptur (D-OH) [-0.365], and Carol Shea-Porter (D-NH) [-0.354] in the House, and Senators Patty Murray (D-WA) [-0.367] and Mark Udall (D-CO) [-0.369] in the Senate.

Click image to enlarge

Posted in Contemporary American Politics, Political Polarization | Leave a comment

Senate: Vote to Invoke Cloture on the Nomination of Debo Adegbile

Below we use Optimal Classification (OC) in R to plot the Senate’s 47-52 vote to reject cloture on the nomination of Debo Adegbile to head the Justice Department’s Civil Rights Division. Adebbile is a controversial pick because of his ties to an appeal of Mumia Abu-Jamal’s sentence for the 1981 murder of a police officer in Philadelphia.

All 44 Senate Republicans who were present voted Nay and were joined by eight Democrats: Senators Chris Coons (D-DE), Bob Casey (D-PA), Mark Pryor (D-AR), Heidi Heitkamp (D-ND), Joe Manchin (D-WV), Joe Donnolly (D-IN), and John Walsh (D-MT). Senator Minority Leader Harry Reid (D-NV) also switched his vote to Nay so that he can bring the nomination back up in the near future. As seen below, the spatial model does a near-perfect job of classifying vote choices on the cloture motion, with only two errors: Senators Bob Casey (D-PA) and Chris Coons (D-DE). Of course, Senator Casey represents the state where the murder of Daniel Faulkner took place.

Click image to enlarge


Note: The plot shows only 51 Nay votes instead of the actual number of 52 because newly-appointed Sen. John Walsh (D-MT) voted Nay but has not yet cast enough votes to be included in the scaling.

Posted in 113th Congress | Leave a comment

House: Vote to Raise Debt Ceiling (updated with Senate Vote on Cloture)

Below we use Optimal Classification (OC) in R to plot the House’s 221-201 vote to pass a clean increase of the debt ceiling. The bill passed with the support of only 28 Republicans and all but two Democrats (Reps. John Barrow [D-GA] and Jim Matheson [D-UT]). As seen in the plot below, this vote clearly split Republicans along ideological lines. The mean first-dimension (representing liberal-conservative position) scores of Republicans who voted Yea is 0.51 compared to 0.61 for Nay Republicans (p < 0.01), meaning that Republicans who opposed the measure were more conservative the 28 who supported it.

The pattern seen in this vote is reminiscent of a proposal discussed during last October's government shutdown to pass a clean continuing resolution in the House with a coalition of Democrats and around 20 moderate Republicans.

Click image to enlarge


Note: The plot shows only 200 Nay votes instead of the actual number of 201 because Rep. Bradley Byrne (R-AL) voted Nay but has not yet cast enough votes to be included in the scaling.

Updated 14 February 2014

We also plot the Senate’s 67-31 vote to invoke cloture on the debt limit increase. The vote was a dramatic one as Republicans scrambled to find the 60th vote to block Sen. Ted Cruz’s (R-TX) attempted filibuster. 12 Republicans eventually joined all 55 Democrats to support cloture. As seen below, the push to assemble enough Republican votes to stop the filibuster created only a weakly ideological vote in the sense of spatial voting. But certainly, Sen. Cruz’s actions were ideologically motivated and clearly divided Senate Republicans.

Click image to enlarge


Note: The plot shows only 66 Yea votes instead of the actual number of 67 because newly-appointed Sen. John Walsh (D-MT) voted Yea but has not yet cast enough votes to be included in the scaling.

Posted in 113th Congress | Leave a comment

DW-NOMINATE Video, 1789-2013

Posted in American Political History, Political Polarization | Leave a comment

An Update on Polarization through the First Year of the 113th Congress (Part II)

Note: Raw data and additional graphs tracking congressional polarization can be found at: http://voteview.com/political_polarization.asp.

In our earlier post, we showed that polarization between the parties in Congress continued to rise through 2013: setting new records in both chambers. One natural question — one we tackle in this post — is whether this increase is mainly attributable to the exit of moderate legislators or the entrance of ideologically extreme legislators.

The plots below track the proportion of moderates (those with DW-NOMINATE scores between -0.25 and 0.25) and non-centrists (those with DW-NOMINATE scores less than -0.5 or greater than 0.5) in the House and Senate over time.

By this specification, moderates comprised over 40% of the membership of both chambers as recently as the early 1980s. This number dipped below 20% after the 1994 midterm elections and has declined to less than 10% in 2013. But most of the disappearance of moderates, especially in the House, had already occurred by the 1990s.

Click images to enlarge



In contrast, this is when the ranks of non-centrists in Congress (those with DW-NOMINATE scores less/greater than -0.5/0.5) really started to take off. This is clearly an example of asymmetric polarization: in 2013, less than 20% of House or Senate Democrats had scores less than -0.5; while nearly 90% of House Republicans and 60% of Senate Republicans had scores in excess of 0.5. This appears to be fueling most of the increase in polarization over the last few Congresses.

Click images to enlarge



Posted in Political Polarization | Leave a comment

The House and Senate Votes on the Farm Bill

Below we use Optimal Classification (OC) in R to plot the House’s 251-166 vote and the Senate’s 68-32 vote to pass a five-year farm bill that cuts $8 billion from food stamps. In the House, Republicans voted 162-63 in favor of the bill while Democrats split 89-103 against it. However, in the Senate, more Nay votes came from Republicans (23) than Democrats (9).

The House vote is a classic example of a “two-ends-against-the-middle” vote, where relative centrists from both parties vote Yea while the most ideologically extreme members of both parties oppose passage. In this case, liberal Democrats were angered by the cuts to the food stamp (SNAP) program, while conservative Republicans were displeased by remaining subsidies for wealthy farmers. The mean first dimension (liberal-conservative) score of Yea House Democrats is -0.62, while it is -0.74 for Nay House Democrats (p < 0.01). The mean of Yea House Republicans is 0.58 compared to 0.64 for Nay House Republicans (p < 0.01).

The ideological picture is a bit more muddled in the Senate. Part of the reason is that the Senate Democrats are very ideologically compressed, and the 9 Senate Democrats who voted Nay do not appear to greatly differentiate themselves from the remainder of the caucus on the first dimension (for example, both liberal stalwarts Sens. Ed Markey (D-MA) and Elizabeth Warren (D-MA) and more moderate Democrats like Sens. Bob Casey (D-PA) and Kirsten Gillibrand (D-NY) both voted Nay). The same is also true of Senate Republicans. The main divide here seems to be along the second dimension that may represent an “establishment vs. outsider” divide but to which we remain hesitant about ascribing any substantive meaning.

Click images to enlarge



Posted in 113th Congress | Leave a comment

Political Polarization and Income Inequality

Two topics that will likely get a lot of attention in President Obama’s State of the Union address are gridlock and income inequality. In this post, we detail how trends in income inequality and political polarization are strongly intertwined with each other and describe how each reinforces the other.

Below we plot time series for indices of both trends. The polarization index is the distance between the party means on the liberal-conservative dimension, as measured by DW-NOMINATE scores. Greater values indicate greater ideological distance (i.e., polarization) between the two parties in the House.

We compare the polarization time series against two measures of income inequality. The first is the Gini coefficient, which measures the concentration of income among members of a given population (in this case, the United States). The Gini coefficient ranges between 0 (maximum equality: all incomes are equal) and 1 (maximum inequality: one person controls all of the income), so that higher values indicate greater income inequality.

The second measure of income inequality that we use is the income share of the top one percent, from an innovative dataset compiled by Thomas Piketty and Emmanuel Saez. Of course, higher values of this series correspond to greater income inequality. One attractive feature of this measure is that it dates back further (1913) than the Gini coefficient.

Clearly, political polarization and both indices of income inequality follow similar trajectories and are highly correlated. The polarization and Gini coefficient series are both roughly flat between 1947 and the mid-1970s, when both begin to trend upward and rise nearly unabated to the present. The polarization and top one percent share series both decline throughout the early twentieth century, reach a post-war dip, and begin to rebound in the 1970s-1980s. The two indices are especially highly correlated since 1945 (r = 0.92).

Text continues after figures.



Of course, correlation does not imply causation, but McCarty, Poole, and Rosenthal in Polarized America explain how political polarization and income inequality are mutually reinforcing trends. On the one hand, polarization can exacerbate income inequality by hampering support for redistributive policies, especially as congressional Republicans have moved to the right.

Income inequality, in turn, obstructs electoral incentives for parties to move back toward the center. For one, those at the bottom of the income scale are disproportionately less likely to vote or otherwise participate in the political arena. Noncitizen immigrants comprise a large segment of this group, which of course deflated political participation rates. At the other end, the wealthiest group gains an over-sized influence through their generous contribution patterns, which favor ideologically extreme over moderate politicians. Hence, there are good theoretical reasons to believe that the trends of political polarization and income inequality are not coincidentally intertwined.

Posted in Contemporary American Politics, Political Polarization | Leave a comment

The House and Senate Votes on the 2014 Omnibus Spending Bill

Below we use Optimal Classification (OC) in R to plot the House’s 359-67 vote and the Senate’s 72-26 vote to pass a $1.1 trillion omnibus spending bill that funds the federal government through October 1, 2014.

All of the Nay votes in the Senate and all but 3 of the Nay votes in the House were Republicans. It is also clear from the plots below that the votes split the Republican Party on ideological lines, and that more conservative Republicans were most likely to oppose the measure. This is somewhat more muddled in the Senate, as the cutting line that divides predicted Yeas from predicted Nays is more diagonal than vertical, but even here the pattern is still fairly unambiguous.

Click images to enlarge



Posted in 113th Congress | Leave a comment

An Update on Polarization through the First Year of the 113th Congress (Part I)

Note: Raw data and additional graphs tracking congressional polarization can be found at: http://voteview.com/political_polarization.asp.

Now that 2013 — the first year of the 113th Congress — has wrapped up and all 640 roll call votes in the House and 291 roll call votes in the Senate have been run through DW-NOMINATE, we can begin to provide some updates on political polarization in Congress.

The primary measure of congressional polarization — the distance between the party means on the liberal-conservative dimension in each chamber — has continued to set new record post-Reconstruction highs over the last few Congresses, and thus far the 113th Congress appears is no exception to this trend. By this measure (shown in the plot below), polarization increased slightly in the House but more drastically in the Senate. This is likely due to the exit of moderates like Senators Richard Lugar (R-IN) and Ben Nelson (D-NE) and the entrance of non-centrists like Senators Ted Cruz (R-TX) and Elizabeth Warren (D-MA) in the 113th Senate.

Polarization in the Senate continues to lag somewhat behind that in the House, but the two series are very highly intertwined (correlated at r = 0.89). This casts serious doubt on the popular explanation that rising polarization is attributable to the redistricting (or “gerrymandering”) process, since only House districts are affected by this process and of course not state boundaries (see also McCarty, Poole, and Rosenthal, “Does Gerrymandering Cause Polarization?”).

Click image to enlarge

We next plot the party means themselves in each chamber over time. In both the House and the Senate, the Republican Party mean continued to shift rightward on the liberal-conservative dimension. On the other side, the Democratic Party mean actually shifted just slightly towards the center in the House, while moving leftward in the Senate.

In addition, while Southern Democrats as a group remain considerably more moderate than non-Southern Democrats in the Senate, regional ideological differences among House Democrats have shrunk considerably, especially during the 112th House after the wave elections of 2010, which wiped out dozens of moderate-to-conservative Blue Dog Democrats in the South. This group has thus far fared somewhat better in the Senate, but the 2014 elections may not be so kind to Senators Kay Hagan (D-NC), Mary Landrieu (D-LA), Mark Pryor (D-AK), or even Mark Warner (D-VA).

Click images to enlarge



Posted in Political Polarization | Leave a comment