Thursday, October 10, 2013

How Polarized Are We? A Look At Some Numbers

This is a reprint from WFAE's The Party Line blog:

As the game of gridlock "chicken" in Washington continues to march on, commentators are trying to explain 'how did we get to this point of polarization?"

Recently, long-time DC observer Dan Balz of the Washington Post offered his explanation of the roots of the government shutdown as being “deeply embedded divisions in America’s politics.” 

Among the factors, Balz contends, underlying the division is the fact that the congressional parties have separated themselves and increased the level of polarization, especially in the U.S. House of Representatives. 

As one of the leading scholars tracking this trend described it, the post-Civil War period of Reconstruction was “a highly polarized time,” but today’s levels “are far worse than we observed then.”  In fact, some of the measures are close to going off the charts.

Another factor that Balz cites is that the parties have become much more homogenous, especially due to the realignment of the South from the Democratic Party to the Republican Party and the demise of the Northeast and Midwest liberal wing of the Republican Party.

This realignment has sparked questions about how deep it went in terms of actual voters.  Some scholars contend that it is only the elites of society that are polarized; one of the leading proponents of this view, Morris Fiorina, argues that the sensationalism created by the media over red states versus blue states is exaggerated. 

Instead, he argues that Americans are closely divided, but not “deeply” divided.  Other scholars contend that Americans are deeply divided, and it’s not just the elites but the masses as well; Alan Abramowitz contends that a large segment of the public are deeply divided, but that this polarization and division “energizes the electorate and stimulates political participation.” 

One way to measure this possible polarization in the public is to look at how the public classifies themselves by both political party and ideology.

One great wealth of data is from the American National Election Studies, which has surveyed citizens since the mid-20th Century and allows scholars and researchers to explore a host of different questions on the nation’s political environment.

One basic way of looking at the political landscape is to classify the survey’s respondents by their self-identified partisan affiliation, going from “strong Democrats” to “pure independent” to “strong Republicans.”  In addition, the surveys asked respondents to classify themselves into different categories based on their ideology, from “extremely liberal” to “moderate, middle of the road” to “extremely conservative.”

Going back 40 years to 1972, we find a general mix among partisans when it comes to their ideology, or “their vision of how they see the world.” 

While one can see a concentration of liberals (extremely to slightly liberal) across the range of politicial party affiliation, conservatives were present in the Democratic affiliations.  Moderates appear to classify themselves more heavily within the Democratic ranks, but were distributed even along the independents to strong Republicans.

Flash forward to 2012, and a different picture emerges of the country.  Now, the ideological partisans have “sorted” themselves between the two parties, with the range of liberals camping out within the Democratic Party, while conservatives have fully migrated into the GOP.

Interestingly, modern-day moderates have adjusted themselves more into the middle as “pure independents,” with an alignment to the Democratic party over the GOP.

In fact, when you look at just “pure independents” who don’t see themselves aligned to either of the two major parties, you find a pretty stable distribution in 1972 across the ideological spectrum.

In today’s environment, there’s an even larger alignment to the “moderate, middle of the road” among pure independents. 

Unfortunately, in 2012, only 14% of the respondents said they were a pure independent.  With 86% of public aligning themselves with one of the two major parties, and voting at least 85% of the time for their party’s candidate, is it any wonder that it’s not just the political elites in government, but the rest of us who can’t seem to understand the other political side? 

Tuesday, October 8, 2013

Party voting by partisan identifiers

This graph shows how different partisan-identified voters cast their ballots in the 2012 presidential election, based on survey results from the American National Election Studies project.

The key point: besides "pure independent" voters, both sides of the political aisle voted at least 85% of the time for their party's presidential candidate. 

Tuesday, May 28, 2013

So Which Chamber Is The 'More Conservative' in the NC General Assembly?

This posting appears at WFAE's The Party Line--the graphics have been included here.

With the Republican-dominated North Carolina General Assembly nearing the end of its long session, many observers have taken to characterizing the legislature’s work under unified GOP control.

For left-leaning groups, the legislature’s treatment of minorities, the poor, education, and environmental protections have lead some leaders to describe the Republican-controlled state government as “Robin Hood in reverse.” 

Other observers outside of the Tar Heel state have described the “unimpeded GOP” as driving the “state hard to the right.” 

Republicans would contend, as Governor Pat McCrory noted in a number of media interviews recently, that they are implementing what they had campaigned in 2012 on, while seeking to put their stamp on state policies. 

But putting aside the partisan characterization on both sides, is there a way to independently analyze where both political parties, particularly in the legislature, are on an ideological spectrum?

In some new research on American state legislatures, two political scientists have used a dataset of roll call votes to scale the two majority parties in each state to place them on an ideological continuum.

This research mirrors the research on the U.S. Congress that lines up members of Congress in terms of most liberal to most conservative, most notably based on roll call voting on economic issues.

Using roll call votes from 1996 to 2010, the state legislative ideological scores indicate where the two parties have aligned themselves in the fifty state assemblies, from most liberal to most conservative. 

If using zero as “moderate,” a score moving towards +1 would indicate a more conservative legislative party, while a movement towards -1 would indicate a more liberal bent to the legislative party. 

As depicted in the below graph for the House conferences and Senate caucuses in the North Carolina General Assembly, the Republican Senate caucus has been the most conservative group in the state legislature since 1996.

State Republican senators have ranged consistently in the more conservative end of the spectrum, with their Republican breathren in the NC House have been moving more and more conservative in their orientation. 

Among the Democrats in the legislature, the House conference has remained consistent in its moderate-to-liberal leanings, while the Senate caucus has moved, in recent times, from being more liberal towards a more moderate stance.

But what does this tell us about whether the North Carolina legislative parties are more “hard right” or “hard left”?  One way to approach this is to compare the parties to the other Southern states.

When comparing each N.C. party in each chamber to other parties in their respective chambers, two interesting patterns emerge.

Among the lower chambers from 1996-2010, North Carolina’s GOP conference started out in a fairly “moderate” scoring, being grouped with such states as Tennessee, Virginia, and Florida. 

But like our neighboring state to the west, N.C.’s GOP conference has moved more conservative, almost landing in between the moderate states and the more conservative states of Texas and Alabama, for example. 

The second interesting pattern was among Democrats in the senate (upper) chambers. 

Three distinct groupings appear in over the 1996-2010 time period: a grouping of fairly moderate Democratic caucuses (Louisiana to Alabama, though most all have moved more liberal in the past few years); a second grouping of South Carolina, Tennessee, and Georgia; and a third and distinctly more liberal grouping of Virginia, Florida, and North Carolina. 

For the year 2008 (the most recent with all the Southern states represented), North Carolina’s Democratic House delegation ranked as the fifth most liberal among the Southern states, behind Florida, Texas, Virginia and Georgia, while the Republican House conference was eighth most conservative in the region.

In the upper chambers, North Carolina’s Democratic Senate caucus tied for the second most liberal group, with Florida being the most liberal for its respective Democratic senators.  Conversely, the North Carolina Republican senate caucus was the fifth most conservative, being bested by the GOP senate conferences in Alabama, Arkansas, Mississippi, and Texas. 

Many political pundits have also questioned the relative “conservatism” of the GOP in both the House and Senate, usually by making the guess (based on ancedotal evidence and “gut-sense”) that the upper chamber is more conservative than their counterparts in the lower chamber. 

It would appear, at least from the historic trend lines before the GOP took over both chambers in 2011, that this analysis would be accurate, but only time will truly tell when it comes analyzing this year’s legislative votes. 

While the political scientists are still working to incorporate the 2011-2012 legislative voting records into their dataset, it will be important to watch what happens when the new scores reflect the GOP take-over from the 2010 elections. 

Thursday, May 9, 2013

Is the United States Becoming More Parliamentarian In Congressional Elections?

A re-posting from WFAE's The Party Line with the graphics included:

As many of the post-election commentaries pronounce a host of reasons (the better candidate in political workmanship, the novice challenger, a localized race that went national) why the once disgraced, now redemptive, Sanford won, we might want to view a more important component of his victory: the voters of the first congressional district.

If it wasn’t for Sanford and his baggage, most all analysts would have written off the Lowcountry contest as a “safe Republican” seat, due to the fact that Mitt Romney carried the district by 18 percent in the 2012 presidential election.

In fact, this kind of “landslide” district has become the national norm in U.S. House contests.

In Nate Silver’s analysis of the 2012 U.S. House races, he found that in 2012, only 35 districts—less than ten percent of the 435 contests—were “swing” districts, meaning that the district results were within five percentage points of the national popular vote margin.

More importantly, it appears that House elections are showing a closer alignment with the overall electoral patterns of voters, especially using the presidential returns.

In the 1st Congressional District election, I took the precinct returns from the 2012 presidential election and asked, would those presidential results have any possible predictive power to an election six months later? 

Meaning, would Romney performance in each precinct give an indicator of Sanford’s performance as well?  Conversely, would Obama’s performance indicate how Colbert Busch would perform as well? 

Using Romney’s performance on the horizontal axis and the preliminary numbers for Sanford on the vertical axis for the largest county (Charleston) in the 1st District, here’s the result:

Romney’s Vote Share in 2012 Presidential Election and Sanford’s Vote Share in 2013 Special Election in Charleston County

With a few exceptions (most notably a precinct where Romney got only 8% of the vote, but Sanford got 34% of the vote), the vote share alignment between Romney and Sanford is pretty striking. 

And even though Obama won Charleston County in 2012, the relationship between his vote share and Colbert Busch’s vote share in that county is also striking.

Obama’s Vote Share in 2012 Presidential Election and Colbert Busch’s Vote Share in 2013 Special Election in Charleston County

So what might this mean?  One explanation might be that the United States is becoming more “parliamentary” in its national legislative elections: it doesn’t matter who the candidate is (hiking boots and all the relevant baggage), but what does matter is the voters’ party allegiance. 

This would tend to make us more along the lines of British elections, where the voters cast their ballots for the party; the “candidate” standing as that party’s choice to be the member of Parliament really doesn’t matter, because that candidate was picked by the party without any voter input. 

Granted, U.S. primary elections have become “the” election, rather than the general election, because, as Silver pointed out, more and more districts are “landslide” in their behavior (117 Democratic and 125 GOP districts in 2012’s House elections were 20 points or more above the national popular vote). 

So, it appears that even in a contest, headed by candidate who suffered from both self-inflicted wounds and a deep drive to win, the district behaved as it should—and gave the landslide win that most of us should have expected, but didn’t.  

Sunday, April 28, 2013

The Need for Independent Redistricting in North Carolina

This post appeared on WFAE's The Party Line

A bi-partisan bill has been introduced into the N.C. House of Representatives to hand over the redistricting process in the state to an independent commission, and thus give up one of the most important powers that any majority party has: pre-determining which party will win each district through political gerrymandering.

The legislation would charge the Legislative Services Office with drawing congressional and legislative districts without the use of “political affiliations of registered voters, previous election returns, [or] demographic information, other than population head counts.” 

In addition, the bill states that no district “shall be drawn for the purpose of favoring a political party, incumbent legislator, or member of Congress, or other person or group, or for the purpose of augmenting or diluting the voting strength of a language or racial minority group,” along with barring the use “of any of the addresses or geographic locations of incumbents.”

In the game of politics, this would be a radical departure for the next round of redistricting in the state in 2021—but one that would probably earn the legislators some form of public admiration, because the parties would be giving up their power over the voters.

In looking at the consequences of the most recent redistricting done by Republicans in 2011, we can see what the “victors” do with their spoils in winning control of the redrawing process. 

In the 2012 election under the new legislative maps, Republicans created safe districts for both themselves and for Democrats—but created disportionately more districts for the GOP than the opposition.

In the state senate, six contests out of fifty could be considered “competitive” (meaning that the winner received less than 55% of the vote).  In fact, three times as many seats (18) had no opposition contesting the election.

If all of the state senate votes for Democratic and Republican candidates were added up respectively across the state, Democratics garnered 47% to Republican’s 53%.  But Republicans ended up with 66% of the seats to Democrats’ 34%, creating a disproportionate advantage for the Republican party.

In the state house, a similar pattern emerged: Republican candidates garnered 51% of the total state-wide vote, but won 64% of the seats in the chamber.

In fact, nearly half of the 120 seats in the house went uncontested (27 automatically went to Republicans, while 28 went to Democrats).  A little over ten percent (14 seats) could be considered “competitive” in terms of the winner securing less than 55% of the vote.

So it appears that political gerrymandering seems to exist in the North Carolina General Assembly districts.

Another way to look at this gerrymandering is to explore the relationship between presidential voting in the district and the voting for the district’s legislative candidates. If the district was “competitive,” then there could be districts that, for example, voted for Obama while, at the same time, voted for a Republican for the state house.

In only ten percent of the 120 state house seats did the presidential candidate of one party win in the same district as the opposition party’s legislative candidate.  For example, in District 92, Obama carried the district with 54% of the vote while the Republican state house candidate won with 51% of the vote. 

In District 119, Mitt Romney won the district with 51% of the presidential vote while the Democratic legislative candidate won the same district with 52% of the vote. 

With so few districts having ‘split-ticket voting’ patterns, it is notable that the Romney vote in a district has a strong correlation to the vote cast for the Republican House candidate. 

Correlation between a District Voting for Mitt Romney
and the Republican State House and Senate Candidates

In the upper chamber, only two districts out of 50 saw the two parties split the contests at the presidential and state senate. 

In District 19, Obama won by a little over 500 votes, but the Republican won the state senate seat with 54% of the vote.  In District 25, Romney won with 58% while the Democrat claimed the senate seat with 53% of the vote.

One could attribute the strong correlation in both chambers between the presidential and state legislative races to the drawing of district lines that favor one party over another from the top of the ballot down. 

But if the districts were drawn without regards to party affiliation, past election results, and ignorance of an incumbent’s address, the legislative races in North Carolina might reflect a long-held belief of American democracy: that the voters should pick the candidates, rather than the candidates picking the voters.