Thursday, June 21, 2012

On Symmetry and Political Party Polarization in Texas

Several recent books on United States politics and polarization have noted that the most extreme polarization has occurred in the Republican Party, where members have become more conservative in the policy views and voting behavior than Democrats have become liberal. Notable books include Jacob Hacker and Paul Pierson’s Off Center: The Republican Revolution & the Erosion of American Democracy (2005) and, most recently, Thomas Mann and Norman Ornstein’s It’s Even Worse Than It Looks: How the American Constitutional System Collided with the New Politics of Extremism (2012). Mann and Ornstein make the case that that the extremism attributable to the Republican Party and the “Young Guns”—Eric Cantor, Kevin McCarthy, and Paul Ryan, but it started with former Speaker of the House and presidential candidate Newt Gingrich in 1995. Has the same type of asymmetry occurred in Texas? Has the Republican Party become even more conservative ideologically while the Democratic Party has remained moderate to slightly liberal in its ideology? The following information from the 81st and 82nd Texas legislatures offers a partial answer to the questions?

Here are the results from the Texas House in the 81st legislature for Republicans and Democrats (the scores are shown as absolute values; the Conservative scores should be positive and the Liberal scores negative):


In 2009, there were 76 Republican House members and 74 Democratic House members. Although there were no overlaps between the two parties’ members in ideology, there were several Democrats whose ideology was centrist (that is, they were on the conservative side of the ideological divide). Ten Democrats were between the center and .1 either side of center in ideology. Furthermore the average score for Democrats was 0.342 to the liberal side of the ideological center. For Republicans, there were no members that close to the center ideologically. Only two Republicans were between .2 and .4 on the Lib-Con score scale. The average for Republican members was .663, much more conservative than the Democrats were liberal. There is no doubt that in the 81st Texas legislature, the Democrats were much more diverse ideologically than the Republicans. The chart illustrates this point perfectly.


Here are the results from the Texas House in the 82nd legislature for Republicans and Democrats (the scores are shown as absolute values; the Conservative scores should be positive and the Liberal scores negative):


The chart depicting the ideology of the two parties’ members in the 82nd Texas legislature tells a completely different story. In the 2010 elections, Republicans won 99 seats, and after the election two Democrats switched parties, bringing the Republican total to 101. The Democrats were reduced to only 49 members. In 2011, the Republicans were bunched near the conservative middle of the scale, exemplified by the Republican average score of .55. The median is also .55, and the mode is .47. Meanwhile, the Democrats who remained in the Texas House were much more tightly grouped at the extremely liberal side of the Lib-Con scale. The average Lib-Con score for the Democrats (.89) placed them at the extreme liberal end of the scale. The median is .91, and the mode is .95. How can we explain this transformation?

First, many of the Democrats who occupied the ideological center or were near the center lost their election bids to Republicans in 2010. Twenty-two Democrats lost their re-election bids in 2010 and two switched parties. Of those 24, 12 scored below the Democratic mean of 0.342, and 12 scored above the Democratic mean. The mean Lib-Con score of the 24 was .28, which is lower than the Democratic members average Lib-Con score. Among the low scoring 12 were moderates such as Joe Heflin, Patrick Rose, Jim McReynolds, Mark Homer, Charles Hopson (a party switcher), David Farabee, and Allan Ritter (the other party switcher). Three Democrats who lost their seats were among those identified by Mark Jones (See my previous post) as out-of-touch with their districts’ ideologies—Abel Herrero, David Leibowitz, and Carol Kent.

Also, a number of Democrats who had been moderates in 2009 became much more liberal in their voting behavior in 2011. For example, Sylvester Turner went from very slightly liberal in 2009 to very liberal in 2011. Others moving to a much more liberal position were Helen Giddings, Joe Deshotel, and Ruth Jones McClendon. Perhaps the Democrats, reduced to such a small number in the House, felt that they had to unify and present a more cohesive liberal front in opposition to an overwhelming majority of Republicans.

Whatever the reason, the House Democrats became much more liberal, and the House Republicans became slightly less conservative.

Wednesday, June 20, 2012

On the Institutional Consequences of Partisan Polarization


This blog is devoted to the relationship between political ideas and political institutions. Nowhere in Texas government is the relationship more evident than in the Texas House of Representatives. Created as the “people’s branch,” the Texas House is supposed to represent the people who inhabit the legislative district. With the polarization of the parties, the consequences for representation are great. This is especially true when there is a shift in partisan representation in the House (e.g., a Democrat is replaced by a Republican). Notice what happened between 2009 and 2011 in the Texas House when several Democrats were replaced by Republicans.

HD
Name
Party
Lib-Con 2009
Name
Party
Lib-Con 2011
Difference
21
Allan Ritter
Dem
0.15
Allan Ritter
Rep
0.32
0.17
11
Charles Hopson
Dem
0.05
Charles Hopson
Rep
0.34
0.29
69
David Farabee
Dem
0.06
Lanham Lyne
Rep
0.45
0.39
45
Patrick Rose
Dem
-0.02
Jason Issac
Rep
0.54
0.56
12
Jim McReynolds
Dem
-0.01
James White
Rep
0.67
0.68
78
Joe Moody
Dem
-0.31
Dee Margo
Rep
0.38
0.69
52
Diana Maldonado
Dem
-0.31
Larry Gonzales
Rep
0.43
0.74
106
Kirk England
Dem
-0.15
Rodney Anderson
Rep
0.60
0.75
3
Mark Homer
Dem
0.00
Irwin Cain
Rep
0.77
0.77
1
Stephen Frost
Dem
-0.17
George Lavender
Rep
0.63
0.80
33
Solomon Ortiz
Dem
-0.42
Raul Torres
Rep
0.39
0.81
35
Yvonne Gonzalez-Toureilles
Dem
-0.37
Jose Aliseda
Rep
0.47
0.84
134
Ellen Cohen
Dem
-0.37
Sarah Davis
Rep
0.48
0.85
133
Kristi Thibaut
Dem
-0.36
Jim Murphy
Rep
0.52
0.88
93
Paula Pierson
Dem
-0.41
Barbara Nash
Rep
0.48
0.89
85
Joe Heflin
Dem
-0.12
Jim Landtroop
Rep
0.78
0.90
107
Allen Vaught
Dem
-0.36
Kenneth Sheets
Rep
0.59
0.95
101
Robert Miklos
Dem
-0.42
Cindy Burkett
Rep
0.57
0.99
102
Carol Kent
Dem
-0.40
Stefanie Carter
Rep
0.61
1.01
47
Volinda Bolton
Dem
-0.53
Paul Workman
Rep
0.53
1.06
34
Abel Herrero
Dem
-0.70
Connie Scott
Rep
0.38
1.08
96
Chris Turner
Dem
-0.31
William "Bill" Zedler
Rep
0.82
1.13
57
Jim Dunnam
Dem
-0.59
Marva Beck
Rep
0.62
1.21
117
David Leibowitz
Dem
-0.70
John Garza
Rep
0.54
1.24
Italics=Tea Party Endorsed Candidates

Now, let’s consider what happened to the ideology of the representative as the shift from a Democratic House member to a Republican House member occurred. The least change occurred in the two representatives who switched parties (Allan Ritter and Charles Hopson), but their scores were more conservative in 2011 than in 2009. The most dramatic shift occurred in HD- 117, where David Leibowitz was replaced by John Garza. Leibowitz was quite liberal in his voting record, and Garza was quite conservative in his voting record. HD-117, which is located in western Bexar County, is barely Republican in its voting record. One indication of the district’s ideology is the Texas Weekly Index (TWI), which is the difference between the average vote for statewide Republicans and the average for statewide Democrats in each district in contested statewide general elections in 2008 and 2010. The Texas Weekly Index (TWI) for HD-117 was
-4.94 percent. The average HD is more Republican with an index of -17.1 percent.

Another measure is a calculation by Professor Mark Jones that indicates how well matched the ideology of a member of the Texas House is to the voting population in his or her district. According to Jones:

The graph plots the 150 members of the Texas House during the 2009 legislative session on two dimensions. The first dimension (Y-axis) is the representative’s Liberal-Conservative Score based on their voting record in the House, ranging in theoretical value from -1.0 (extreme liberal) to 1.0 (extreme conservative).

The second dimension (X-axis) accounts for the ideological partisanship of the district and is calculated using the average Republican share of the two-party vote in the election for railroad commissioner in 2004, 2006 and 2008. Given the limited information voters possess regarding the railroad commissioner candidates, this vote tends to be based principally on voters’ partisan-ideological preferences. For each district, the percentage of the two-party vote won by the Republican railroad commissioner candidate in the district is subtracted from the percentage of the two-party vote won by this same candidate state-wide. These values are then summed and divided by three to create the District Partisan Voting Index (PVI). The higher the value for the District PVI, the more Republican/Conservative the district is; while the lower the value, the more Democrat/Liberal the district is. Here, the actual values for the District PVI range from -41 to 22.

Finally, in the figure a dashed line (a Lowess curve) represents where we would expect a representative’s Liberal-Conservative Score to be, relative to their district’s partisan voting profile based on a statistical analysis of all 150 representatives. Representatives located above the dashed line have a voting record in the House that is more conservative than we would expect based on the partisan-ideological profile of their district, while representatives located below the line have a voting record that is more liberal than we would expect. Democratic representatives are identified by a blue dot and Republicans a red dot.

And here is the graph plotting each representative and his or her district’s PVI:

Among the Democratic representatives in the 81st Texas legislature, the outliers are Herrero, Leibowitz, Hochberg, and Kent. Among the Republicans are Aycock, Christian, Harper-Brown, and Kleinschmidt. Professor Jones notes:

In the figure, the eight representatives whose respective Liberal-Conservative Score is most at odds with the partisan-ideological profile of their district are identified by name. The highest degree of ideological disconnect between a representative and their district is found in the case of Linda Harper-Brown (HR-105, Irving), who is substantially more conservative than her district’s level of ideological partisanship would suggest she would be. She is followed in this ranking of representative-district partisan-ideological discontinuity by David Leibowitz (HR-117, San Antonio), Scott Hochberg (HR-137, Houston), Wayne Christian (HR-9, Center), Abel Herrero (HR-34, Corpus Christi), Tim Kleinschmidt (HR-17, Austin), Carol Kent (HR-102, Dallas) and Jimmie Aycock (HR-54, Killeen). All four Republicans and all four Democrats are respectively more conservative and more liberal than their legislative district’s partisan-ideological profile would indicate. It is very important, however, to keep in mind that the replacement of these representatives by either a co-partisan or a rival-party member would not automatically lead to the presence of a representative whose Liberal-Conservative Score would be more in concert with the partisan-ideological profile of the district.
In contrast to the above-mentioned representatives, half a dozen House members possess a Liberal-Conservative Score located right along the dashed line, indicating a near-perfect match between their legislative floor voting behavior and their district’s partisan-ideological preferences. They are Democrats Armando Martinez (HR-39, Weslaco), Marisa Marquez (HR-77, El Paso) and Richard Peña Raymond (HR-42, Laredo), and Republicans Frank Corte Jr. (HR-122, San Antonio), Joe Straus (HR-121, San Antonio) and Rob Eissler (HR-15, The Woodlands).

He concludes that:
A significant disconnect between a representative’s floor voting record and their district’s partisan-ideological orientation does not necessarily indicate a crisis of representation. However, the presence of this type of incongruity does suggest a potential gap between a representative’s behavior on the House floor and the preferences of a majority of his/her constituents that may be worthy of greater scrutiny by these constituents.

So, with the replacements for the Democrats noted in the table above, are the current occupants of the office more in line with the partisan-ideological orientation of their districts? Or, has the polarization of the parties led to a liberal Democrat being replaced by a conservative Republican, resulting in an even greater discrepancy between the elected representative and his or her district?

Let’s just consider HD-117 as an example. According to the TWI, the district is slightly conservative. According to Jones’ partisan-ideological orientation (PVI), it’s slightly liberal (about -.6 by my calculations from the graph). So the question becomes are voters in HD-117 represented more accurately by Jose Garza than by David Leibowitz? I would have to say that they are not if the distance between the district’s voting/ideological preference and the representative’s ideology is an accurate reflection of representation. What do you think?

Monday, June 4, 2012

On Party Identification in Texas



I mentioned in a previous post that I would have more to say about partisan leaners in a future post. Well, this is the post.

First, let’s agree on what party identification is because there are several competing definitions. I am a traditionalist, perceiving party identification (PID) as a psychological attachment to a political party. This definition was originally offered in The American Voter by Professors Campbell, Converse, Miller, and Stokes at the University of Michigan, which was published in 1960. They maintained that this is one of the stronger political attitudes, which means that it is less likely to change than other attitudes. They also maintained that it was usually adopted from one’s family. Because of its strength and relative permanence, they argued that it is a filter through which other political attitudes have to pass. They referred to a “funnel of causality,” tracing a person’s other political attitudes to one’s partisan identification. Voting behavior, for example, is considered to be strongly influenced by one’s partisan identification.

The also developed a method of determining a person’s PID. They asked a basic question: “In politics, do you consider yourself a Democrat, Republican, or Independent?” For those responding either Democrat or Republican, they asked a second question: “Are you a strong or not-so-strong Republican (Democrat)?” For independents, the second question was: “Are you a pure independent, or do you lean toward one of the parties?” The result is a seven-point scale, ranging from strong Democrat on the left to strong Republican on the right. The diagram below illustrates the scale.

      SD              WD               DLI            I             RLI             WR          SR
        
Where do registered voters in Texas fall on the PID continuum? In other words, what percentage of Texans falls into each of the seven categories? The results of a May 2012 UT/Texas Tribune poll are represented in the chart below:


If you consider partisans as strong and weak identifiers (SD, WD, SR, WR), then 36 percent of Texas registered voters are Republicans, and 33 percent are Democrats. Independents make up 30 percent of Texas registered voters. However, political scientists have argued about where the “leaners”—Democratic and Republican leaning independents—should be placed. Since the publication of Bruce Keith et al.’s The Myth of the Independent Voter in 1992, political scientists have generally agreed that “leaners” are “closet partisans,” who behave like partisans even though they don’t claim an allegiance to one of the political parties. I have always found accepting this view difficult. If party identification is an emotional attachment to a political party and a person claims no such attachment, why should they be considered partisans? The fact that “leaners” are more supportive of a party’s positions on issues than weak partisans and are more likely to vote for the party’s nominees for public office than weak partisans does not make them partisans.

To some people, the distinction may not seem important, but it has consequences for campaigns and how campaigns target potential voters. If leaners are really partisans, then only about 10 percent of the Texas electorate is independent. If leaners are independents, then about 30 percent of the Texas electorate is independent. If it’s only 10 percent, then they aren’t as consequential to election outcomes, and campaigns should concentrate on mobilizing partisans with partisan appeals. If however, it’s 30 percent, then the campaigns must appeal to independents with messages that are different from their partisan appeals. In other words, it matters!

So, what’s the basis for each side’s view? For those who consider leaners as “closet partisans”—Keith et al., John Petrocik, Alan Abramowitz, and others—the principal reason is that leaners vote for the party’s candidate at a higher rate than weak partisans. Thus, leaners’ behavior (voting) is more like partisans than like independents. However, as Samuel Abrams and Morris Fiorina (2011) assert, the causal relationship could be the other direction: voting causes independents to declare themselves as leaners as they recall their most recent voting decision. In other words, I voted for Obama in the 2008 presidential election; so I must lean toward the Democratic Party. Or, I voted for McCain in the 2008 presidential election; so I must lean toward the Republican Party. Philip Shively made this point in 1980, and this is what I have always believed. Keith et al. and Petrocik also maintain that leaners’ voting decisions demonstrate temporal consistency so that leaners do not vote for one party’s candidates in one election and then switch to another party’s candidates in a subsequent election. Leaners are also no more likely than weak partisans to split their voting decisions during a particular election, voting for some Republicans and some Democrats in the same election.

For those who question the lumping of leaners into one of the partisan camps—Abrams, Fiorina, Todd Eberly, and others—the leaners are not partisans, challenging the now-accepted view that they are partisans. Most recently (March 2012), Third Way released a report by St. Mary's College of Maryland Professor Todd Eberly, using data from a panel study of three successive elections, that made two points about the leaners: (1) leaners were significantly more likely to change their party identification than were Democrats or Republicans, whether strong or weak partisans. (2) leaners switched their vote choice over elections, and Republican leaners were more loyal than Democratic leaners. In response to this study, Abramowitz countered that the study used by Eberly had a small sample size and that there was a larger panel study, conducted during the 2008-2009 period, which showed different results. This survey indicated that very high percentages of Democrats (82 percent) and Republicans (73 percent) retained their position as party leaners from January 2008 until a date more than a year and one-half later. Also, during this period, only five percent of leaners switched to the opposing party. He also points out that the leaners voted for the party toward which they leaned and attributed it to their sharing the dominant ideological orientation of the party toward which they leaned. In response to Abramowitz, Eberly noted that the 2009-2009 panel does allow one to observe partisanship over several elections, unlike the 2000-2004 panel. He added results from a 1992-1997 panel survey. The results of his investigation led Eberly to conclude that partisan loyalty declines over time and is weakest among weak partisans. But he also observed that most defections were by leaners. He concludes that about 20 percent of the electorate are not loyal partisans, and that, “in an era of closely matched political parties and relatively narrow two-party vote shares, winning and maintaining the support of that 20 percent is crucial.”

So, what do you think? Should leaners be considered partisans or are they susceptible to appeals from either political party? Should campaigns concentrate their efforts not only on turning out faithful partisans but also on winning the independents, whether leaners or pure independents?

Friday, June 1, 2012

On the Anemic Voter Turnout in the May Primaries

Voter turnout in the United States is a problem, assuming, of course, that high voter turnout is a good thing. If you don’t believe that, then this post is not for you. The voter turnout in the May 29th primaries was not good, especially in the Democratic Party’s primary. Here are the stats.

First, only 2,030,927 Texans voted in both parties’ primaries. That’s an anemic 15.54 percent of the 13,065,425 registered voters in Texas. Turnout was particularly low in the Democratic Party’s primary, where only 587,146 voted. If 33 percent of Texas’ registered voters are Democrats (strong and not-so-strong Democrats), then there are approximately 4,311,590 Democrats who were registered to vote in Texas. That means that voter turnout in the Democratic primary was 13.62 percent. In the Republican Party’s primary, 1,443,781 voters participated. If 36 percent of Texas’ registered voters are Republicans (strong and not-so-strong Republicans), then there are approximately 4,703,553 Republicans who were registered to vote in Texas. That means that voter turnout in the Republican primary was 30.70 percent. That is much higher than in the Democratic Primary, but it is no cause for celebration.

Second, why was voter turnout so low and how does this compare with previous primary elections in presidential years? Let’s answer the second part of the question first. In 2008, 33.23 percent of registered voters voted in both parties’ presidential primaries. In 2004, 12.45 percent voted in both parties’ presidential primaries, and in 2000, 16.48 percent voted. For the three most recent presidential primaries, the average turnout was 20.72 percent. So, 2012 was below the average for the last three presidential primaries, but it was higher than in 2004 and only slightly lower than in 2000.
On the first part of the question, there are several answers: (1) the presidential nomination for both parties had already been decided by the time that Texas voters voted in their primaries. In 2008, John McCain did not clinch the nomination until March 4th, the date of the Texas primary, and the Democratic Party’s nominee was not decided, and the contest between Senators Hillary Clinton and Barack Obama was very close. As a result, turnout was high in both parties’ primaries, and more Texans voted in the Democratic primary than in the Republican primary. (2) the timing of the election was different, having been moved to May 29th because of lawsuits over redistricting in Texas. Furthermore, holding the primary elections after the four-day Memorial Day weekend did not help turnout. (3) there were few close contests in the Democratic primary elections, reducing interest in the election. There was much more excitement in the Republican primary, especially in the contest to select a U.S. Senate candidate to replace retiring Senator Kay Bailey Hutchison.

One could consider the early voting turnout a bright spot; however, early voting has not increased voter turnout. Nearly half (49 percent) of Texans voted early in the political parties’ primaries this year. Fifty-one percent of Democrats voted early, and 48 percent of Republicans voted early.