Tuesday, November 8, 2011

Let's Go Vote 2011

Well, a little over an hour to go before the polls close in Charlotte and I'm here in the WSOC newsroom. Great group to work with, as usual, and a lot of fun. Stay tuned for some updates on the Charlotte election races coming up.

Thursday, November 3, 2011

Charlotte's Upcoming 2011 Election

With the 2011 election season coming to a close in the Queen City, I thought I would post a map with the precincts in Mecklenburg County color-coded using the Partisan Voting Index (PVI) that I developed for analyzing the voting behavior of precincts. As just a reminder of what makes up the PVI:

  • For each precincts, I used the 2004 and 2008 presidential election returns in that precinct for each presidential candidate and subtracted the Mecklenburg County average. So, if Kerry got a county-wide average of 51% and a precinct gave him 61%, he got a +10; same for Obama in 2008; and then I averaged the two numbers.

So, when I did this for each precinct, I classified them into the following categories:

  • Greater than +10: Likely precinct for the party

  • Between +3 and +10: Lean precinct for the party

  • Less than +3: toss-up precinct (read battleground)

The map is located here.

Notice that out of the seven city council districts, Districts 1-5 are pretty much "blue" and should elect Democrats, while Districts 6 & 7 are pretty "red" and should elect Republicans out of them. If you take all of the outside boundaries of the districts and consider those the Charlotte city limits, you'll see a pretty significant "blue" tint to the city limits.

While this map is not necessarily intended to indicate who will win or loose the mayoral race or the at-large council races next week, it's pretty indicative of Charlotte matching the national norm that urban areas go Democratic, while suburban areas (read outside the city limits within Mecklenburg) go Republican.

What really surprises me is the sense of "sorting" by Charlotte residents into areas that are heavily one party or the other. The notable scant numbers of "toss-up" precincts is one that really surprised me, but only confirms the findings of folks like Bill Bishop in his wonderful book The Big Sort: Why the Clustering of Like-Minded America is Tearing Us Apart.

Wednesday, August 3, 2011

PVI Analysis of Mecklenburg County Precincts

Following the controversy of the North Carolina General Assembly stepping in and finalizing the district boundaries for the Mecklenburg County Board of Commissioners, I looked at the precincts and their voting behavior at the presidential level to see about developing a Partisan Voting Index analysis. Using the 2004 and 2008 presidential election results at the county-wide level, here are the results for the 195 precincts:

  • Likely Republican precincts (those precincts which, on average, voted 10% or above the county average for Republican presidential candidates in '04 & '08): 72

  • Lean Republican precincts (those precincts which, on average, voted anywhere from 3-10% above the county average for Republican presidential candidates in '04 & '08): 19

  • Toss-up precincts (those precincts which, on average, voted anywhere from +2 Republican to +2 Democrat, compared to the county average for presidential candidates in '04 & '08): 13

  • Lean Democratic precincts (those precincts which, on average, voted anywhere from 3-10% above the county average for Democratic presidential candidates in '04 & '08): 26

  • Likely Democratic precincts (those precincts which, on aveage, voted 10% or above the county average for Democratic presidential candidates in '04 & '08): 65
Here's a map depicting the precincts with their PVI classifications.

What is striking to me is how much Mecklenburg County is fitting into a model described by Bill Bishop in his book "The Big Sort" and in Dante Chinni and James Gimpel's work "Our Patchwork Nation." Bishop, in particular, notes that communities are "becoming even more Democratic or Republican. As Americans have moved over the past three decades, they have clustered in communities of sameness, among people with similar ways of life, beliefs, and, in the end, politics" (5).

In comparing the recent Mecklenburg County Commission districts, known as Stetson 5 and LM Plan B, we find that the districts break down as follows in both plans:

Stetson 5 plan (you can access the data here):

  • Proposed District 1: R+10.8%

  • Proposed District 2: D+23.5%

  • Proposed District 3: D+22.5%

  • Proposed District 4: D+10.0%

  • Proposed District 5: R+14.4%

  • Proposed District 6: R+14.6%

LM Plan B (you can access the data here):

  • Proposed District 1: R+11.4%

  • Proposed District 2: D+20.4%

  • Proposed District 3: D+21.9%

  • Proposed District 4: D+13.2%

  • Proposed District 5: R+11.5%

  • Proposed District 6: R+16.9%

Both of these plans basically draw districts designed to elect one party over the other, as was the charge of the current county commission to the redistricting committee.

Wednesday, July 27, 2011

Updated PVIs for proposed NC House & Senate districts

Since we're now on the 4th version of a redistricting map for the N.C. House and a 2nd version for the N.C. Senate, I thought it might be helpful to update the Partisan Voting Indices for the new proposed districts.

  • PVI's for NC House (Lewis-Dollar-Dockham 4) here. In a nutshell, potential Likely/Lean Republican districts at 75, Likely/Lean Democratic districts at 36, leaving a total of 9 districts as potential "toss-up" districts. Here's a map of the districts sorted by Likely/Lean/Toss-up categories (see the below blog entry for a description of this approach).

  • PVI's for NC Senate (Rucho 2) here. In a nutshell, potential Likely/Lean Republican districts at 31, Likely/Lean Democratic districts at 16, leaving at total of 3 districts as potential "toss-up" districts. Here's a map of the districts sorted by Likely/Lean/Toss-up categories.

Thursday, July 21, 2011

Cook PVI for pre-2010 NCGA Elections

A couple of "followers" on Twitter asked "how would the Cook PVI have predicted the 2010 legislative elections?" Here are the charts, similar to the ones I produced below (looking at possibly classifications of the proposed redistricting maps), using 2004 and 2008 presidential election returns in the House and Senate districts.

A couple of points about this analysis:

In the NC House:

  • In the "Likely Republican" districts, there were 4 districts that were held by Democrats. All four of them changed political hands as a result of the 2010 elections (as noted by being bold and italics)

  • In the "Lean Republican" districts (26), there were 6 districts that were held by Democrats and those seats were captured by the Republicans in the 2010 elections. One (District 61) was won by an unaffiliated candidate.

  • In the "Toss Up" districts (20), there were 11 districts that were rated with Republican PVI but held by Democrats; of those districts, 4 of them changed parties in the 2010 election.

  • In the "Lean Democratic" district (10), one Democratic incumbent was defeated by a Republican (District 41).

In the NC Senate:

  • In the "Likely Republican" districts (15), one went from Democratic control to Republican control (District 45).

  • In the "Lean Republican" districts (11), 5 districts were held by Democrats; only one survived the 2010 election.

  • In the "Toss Up" districts (6), 4 of those districts went from Democratic to Republican control in the 2010 election.

  • In the "Lean Democratic" districts (5), one was captured by a Republican in the 2010 election.

Please remember--these districts prior to the 2010 election are not the same as the proposed districts, so comparisons are not applicable (apples & oranges).

Sunday, July 17, 2011

Redrawing the Rules of the Game in North Carolina

Well, it’s been a fascinating week in North Carolina politics, due mostly to the redrawing the rules of the game of politics. The proposed district maps for the NC General Assembly have generated a lot of commentary and perspectives, along with the pending revisions to the congressional districts. In looking primarily at the state house and senate legislative districts, it would be helpful to have a “reference” point for dissecting these maps and their political importance in next year’s election.


What I have done is provided color-associated maps of the 120 new districts in the NC House of Representatives and the 50 new districts in the NC Senate, based on the approach that Charlie Cook takes with his analysis of U.S. House districts, known as the Partisan Voting Index, or PVI. The PVI uses the 2004 & 2008 presidential elections as a basis (see below) for the districts, and then classify the districts into five categories: “Likely Republican,” “Lean Republican,” “Toss Up,” “Lean Democratic,” and “Likely Democratic.” Based on the PVI model and classifications (see below), we could expect the following:

  • In the "new" NC House of Representatives, 62 districts appear to be “likely” to elect one party over the other; within these 62 “safe” districts, there is a 33-29 advantage to the Republican Party.

  • In the "New" NC Senate, 27 districts appear to be “likely” to elect one party over the other; within these 27 districts, there is a nearly even split between Democrats (14) and Republicans (13).

  • It is within the districts that appear to “lean” to one party or the other that the Republicans could truly make substantial inroads to majorities in both chambers. In the NC House, 39 districts appear to lean to the GOP, while only 7 appear to lean to the Democrats. In the NC Senate, 16 districts appear to lean to the Republicans, with only 2 leaning to the Democrats.

  • In terms of those districts classified as potential "toss-up" districts, both chambers have only ten percent of their seats within this category: 5 out of 50 senate seats and 12 out of 120 house seats could be competitive, based on their PVI scores.

  • To document the partisan leanings of these five categories, you may find these analyses showing the averages within the five categories for the NC Senate and NC House districts as to registered party affiliation and the 2004 and 2008 presidential and gubernatorial election results in these new districts.

First, some notes regarding how to read the attached maps and the PVI scores (in other words, the “disclaimer” portion of this analysis):

  • Much like Charlie Cook’s analysis, these are not predictions on who will win the districts, but rather an objective measure to compare districts to one another. Charlie Cook’s Partisan Voting Index (PVI) utilizes a district’s presidential voting patterns against an average (his being national, mine being state-level for NC) for the past two presidential elections: 2004 and 2008. Each party’s presidential election result in the district has the party’s statewide average subtracted from it, and then the two figures are averaged for the PVI. The PVI shows the district’s proclivity of voting for each party at the presidential level (D+5 means the district would have voted, on average, plus five points Democratic), which may (or may not) filter down to lower-ballot races (see below for more on this point).

  • The PVI is based on the 2004 and 2008 presidential elections in North Carolina, which has its own set of cautionary tales. The 2004 presidential election was the classic retelling of a ruby-red presidential state, with George W. Bush winning the state by 13 percent. But when Democrat Obama came to make North Carolina competitive, he turned that ruby-red into a deep purple, winning the state by just 0.32 percent. When you have two elections of such extreme ends, the measurement probably will reflect that shift as well. My “political gut” leads me to believe that 2012 will probably be more like 2008 than 2004 (see the DNC in Charlotte, the Obama campaign “hinting” of NC’s importance to their re-election), with both political parties utilizing a massive ground-game of voter mobilization (the GOP learned from the Democrats in 2008, and brought out their ground game in 2010, thus winning the North Carolina General Assembly).

  • Other analyses of these maps will most certainly utilize a combination of other factors—such as voter registration, down-ballot elections (auditor, lieutenant governor, etc.)—beyond what this analysis does. I certainly do not discount these factors, particularly voter registration, in analyzing these districts and their potential. Again, these maps that I present are not predictions, but indications of where these districts may perform in future elections. A lot of other factors—mobilization, candidate qualities, the role of campaign finance—will certainly make the equation much more difficult, and I readily acknowledge those factors.

  • As noted above, North Carolina voters have a tendency to split their tickets: meaning, voters are willing to vote for one party for president, the other party for governor, switch back to the first party for U.S. Senate, and then back to the second party for state house, etc.

  • Conversely, the 2008 exit poll for the November General Election showed that 89.9% of self-identified Democrats voted for Obama, while self-identified Republicans voted 93.3% for McCain. For both parties’ candidates for U.S. Senate and NC governor in 2008, at least 85% of self-identified partisans voted for their party candidate.

  • Finally, when incumbents run for re-election, they are often hard to beat. The power of incumbency most often is due to name recognition, which challengers often must spend a disproportionate amount of money on to achieve some level of parity with the current officerholder.
So, with those caveats in mind, let me discuss each set of maps and their overall patterns, along with some other observations.


  • Lewis-Dollar-Dockham 1 Map indicating the PVI categories for districts (here is the map posted on the NCGA website that only show the different districts)

  • Listing of districts by the PVI categories: this one is sorted by district number within the category, while this one is sorted by the PVI from most Republican to most Democratic within the categories
The PVI analysis of the new legislative districts indicates that 33 districts will most likely go Republican, while another 29 districts will most likely go Democratic, providing a fairly equal base of support for each party to build upon. Within almost all of these districts, the margins of victories at the presidential level were translated to the parties’ gubernatorial candidates, and the winning party’s legislative candidates benefited as well (some more than their presidential & gubernatorial candidates). In addition, the percentages of registered voters appear to benefit the winning party as well. The one exception is the Democratically-held District 69, which has a VPI of +12 Republican. It is hard to imagine that Democratic incumbent Representative McGuirt would hold on in this plurality Republican-registered district, but again, incumbency may prove beneficial.

The most Republican districts in the House appears to be Districts 73 (Alexander-Wilkes-Yadkin) and 78 (Randolph & Moore), with R+20 PVI ratings. The most Democratic district appears to be District 29 (Durham), with a PVI rating of D+37.

So both parties start with solid bases, but it is in the “lean” categories where the apparent success of redrawing the lines has paid off for the Republicans: 39 districts are classified as "Lean Republican," with 9 of these districts currently being held by Democratic incumbents. Of these 9, four districts (35, 45, 55, and 59) that Democrats currently hold appear as “open” seats, presenting viable opportunities for the opposing party (Republicans) to pick these seats.

Conversely, only 7 districts "Lean Democratic," with two of them being open seats—another challenge for Democrats to hold on to these seats.

Within the twelve competitive “Toss Up” districts, most of the districts have slight Republican advantages, but some are truely battleground competitive: districts 66 (that includes Richmond, Scotland, Hoke, and Robeson counties) and 41 (Wake) will be the ones, most likely, by which to judge next year’s elections by, as their PVIs indicate an even split between the two parties.

A few more observations: in the districts that are held by Democrats but indicate a “lean Republican” status, half of them share a common characteristic: voters who like to split their tickets. In both 2004 and 2008, four proposed districts voted Republican at the presidential level and voted Democratic at the gubernatorial level. How this will translate even further down ballot is to gaze into the political crystal ball.

As noted above, one House District—69 (Union)—currently held by the Democrats will almost inevitably go Republican, due to the past voting history at the presidential and gubernatorial level (no less than 16% victories for Republicans) and registered Republicans making up 41% of the electorate, compared to 32% registered as Democrats.

Among the Democratic-held districts that are classified “Toss Ups,” they again share the characteristic as their “lean Republican” brethren—that of split ticket voting: Republican at the presidential level, Democratic for governor.

If it’s a consolation to the Democrats, two Republican-held districts (6 (Washington, Beaufort, and Craven) & 46 (Columbus & Robeson)) that are “lean Republican” show the split ticket phenomenon as well. But these two districts also have their Republican incumbent double-bunked with Democratic incumbents, so these should be ones to watch next year, if all four incumbents decide to run within these new districts.

In the Republican-held “Toss Up” category, these districts appear to be ones where the Obama grass-roots mobilization had a significant effect, going from an average 20% Republican advantage in 2004 to, in one case, 0.05% Democratic advantage.

The end result: it appears that the Republicans have truly made their base of winnable districts much stronger than the Democrats. If Republicans were to only take those districts in the “Likely” and “Lean” columns, they would have a safe majority in the NC House of Representatives, starting with 72 seats. The map detailing these seats and classifications are here (same as the above link; deep red is a district with a Republican PVI greater than 10, while districts in green are “toss-ups,” and dark blue are districts with Democratic PVIs greater than 10).


  • Rucho 1 Map indicating the PVI categories for districts (here is the map posted on the NCGA website that only show the different districts)

  • Listing of districts by the PVI categories: this one is sorted by district number within the category, while this one is sorted by the PVI from most Republican to most Democratic within the categories
Across the marble hall on Jones Street, the Senate shows some commonality with the House districts, but also some oddities as well. Both parties have a strong list of “likely” seats, with Republicans appearing to have 13 to the Democrats’ 14 districts. But again, like in the House, it is in the “Lean” columns where Republicans have made substantial inroads: 16 “lean” GOP districts (with four of those being held currently by Democrats) to only 2 “lean” Democratic districts (and no GOP districts are “lean Democratic”). Starting at a potential 29 seats, the GOP would, like in the House, already begin the 2012 contest with a significant majority in the 50-member upper chamber. The five districts that are truly “toss-ups” are just that—they could conceivably break either way.

The districts that appear the most Republican in the Senate are 29 (Moore & Randolph) and 30 (Stokes, Surry & Wilkes), both at R+15, while the district that appears to be the most Democratic is 28 (Guilford) with D+31 PVI.
There are some oddities, however, in the Senate. For example, Senate District 13 is held by a Democratic incumbent, but listed as “lean Democrat,” simply because it went only 2 and 6 points for the Democratic presidential candidates in 2004 and 2008, respectively. More importantly, however, it went 38 and 39 points for the Democratic gubernatorial candidates, so even though the PVI shows a limited average for Democrats, this should be a safe seat for the party.

The four other Democratic-held seats that “Lean Republican” (new districts 1, 18, 25, and 27) show strong Republican performances in 2004, with three of them again splitting in 2008 (voting for McCain while electing a Democratic state senator).

Five senate districts are classified as "toss-ups," with the potential to go to either party, based on 2008 election performances.

The end result in the Senate: like in the House, if Republicans only gain those seats listed in the "Likely Republican" and "Lean Republican" columns, they will have a majority of 29 and control the upper-chamber as well. The map detailing these seats and classifications are here (same as the above link; deep red is a district with a Republican PVI greater than 10, while districts in green are “toss-ups,” and dark blue are districts with Democratic PVIs greater than 10).

One last comment: for those seeking truly competitive elections, it may be disheartening to have only 5 out of 50 senate seats, and 12 out of 120 house seats, where either party could claim victory. So, for 89 out of 170 seats up for grabs in the North Carolina General Assembly next year (those "likely" seats), we may know in the primary election who will hold the seats come the general election. Welcome to the world of partisanship and gerrymandering.

I'd like to thank the NC Free Enterprise Foundation for their very useful comparison charts on the NC House & Senate districts.

Friday, May 13, 2011

Is it a problem to have too many voters?

This week (May 12), the North Carolina House of Representatives passed, on second reading, Bill 658 entitled “Change Early Vote Period.” Since 2000, North Carolinians have had the opportunity to cast their ballots before the Tuesday following the first Monday in November, contending that giving voters more opportunities than just a twelve-and-a-half hour window to cast their political preferences is good for a democratic republic. State law currently allows for nearly three weeks of voting to occur, with voters having the opportunity to both register and cast their ballot with "one-stop" voting.

Bill 658 reduces the period of “one-stop” voting by a week, with an estimated savings of $1,945 per site of early voting for the counties, according to estimates by the General Assembly’s Fiscal Research Division. In this time of severe budgetary constraints, at both the state and local level, it seems wise to save limited tax dollars wherever it may be found.

But the possibility of narrowing the opportunity for popular participation in our governing system seems harsh in a once every two-year period. Who might benefit from this expanded window of voting opportunity, and who might this harm? Well, interestingly enough, both parties benefit, and unaffiliated voters (those who do not align their voter registration with either the Democratic or Republican party) may ultimately be the loser.

In 2008, David Plouffe, campaign manager for Democratic presidential candidate Barack Obama, noted that Democrats “had to expand the electorate or we were cooked.” And so they mobilized and got people, particulary registered Democrats, out to vote early. In North Carolina’s 2008 election, registered Democrats made up 45.7% of the statewide registered voters; but among early votes cast, 51% were from registered Democrats—meaning Democrats “outperformed” their state-wide registration percentage in casting early votes.

But when it came to Republicans and unaffiliated voters, both groups showed an “underperformance” at casting early ballots in 2008. Republican registered voters were 32% of all statewide voters, they made up only 30% of the early votes cast, while unaffiliated voters (22% of registered voters) were only 18% of early votes cast.

So it’s no surprise that House Majority Leader Skip Stam, a proponent of HB 658, remarked that “October 2008 was the longest month in the history of the world, I think,” and went on to remark that the “election’s just too long, and of course that favors incumbents and it favors wealthy people.”

But what Democrats used in 2008, Republicans learned to use in 2010. In last year’s election, registered Republican voters, who were 31% of statewide voters, were 36% of all early votes cast, while Democrats (44% of statewide voters) were 46% of early votes cast. The GOP learned their lesson from 2008, and had a better "outperformance" in early votes than their Democratic voters.

And still, unaffiliated voters underperformed their statewide total in 2010 (only 17% of early votes cast were from unaffiliated voters, who were 23% of registered voters).

So what does this mean? Political parties and candidates learn, and they learn very quickly when one side uses the rules of the game to their advantage. Democrats used it in 2008, and Republicans used in 2010. The election of 2012 will most likely be the all-time battle of which party can “bank their ballots” early.

If politicians are worried that elections are too long, then consider adopting measures that many other nations around the world do—for example, have a set period of “legal campaigning” which lasts only two to three weeks prior to election day. No television campaign ads, no yard signs, no bumper stickers allowed before the designated time—that would more properly address the issue of “long campaigns,” but I dare say the response to that would be “it violates free speech!” And you’d be right.

But to take it out on voters to cast their fundamental right as citizens to participate by decreasing their chances to be heard? Is that a civic value North Carolinians want to promote when it comes to the fundamental right of voting?

And does it do either party any good to limit that opportunity? Ironically, the GOP may be passing a bill that enabled their success last year just to get back at the Democrats for the ‘08 victory. And the ultimate loser may be the individuals who we need the most: the voters of North Carolina.

Sunday, February 27, 2011

Should NC return to partisan judicial elections?

In the February 26 edition, the Salisbury Post gave a “dart” to NC House Bill 64 (Senate Bill 47) and co-sponsored by Rowan County’s state senator Andrew Brock (R-Davie), to return judicial elections in North Carolina to a partisan election, rather than the current non-partisan election system. In the study of judicial politics, it might be helpful to know what researchers have found when judges are subjected to the “will of the people,” both in terms of partisan and non-partisan elections.

Twenty-two states use elections to select judges, with a majority of those states utilizing non-partisan elections—North Carolina being one of them. Surrounding the idea of judicial elections are contending values: making the third branch of government responsive and accountable to the people, while ensuring the idea of fairness and impartiality in the administration of justice.

In an analysis of judicial elections across the states, two political scientists (Bonneau and Hall) explored the arguments and “myths” of judicial elections, one of which is that nonpartisan elections “depoliticize” campaigns and decrease the amount spent on judicial elections. In fact, the researchers found that nonpartisan judicial elections increase the costs of campaigns, whereas partisan elections decreased the costs of elections.

In defending judicial elections, the scholars noted that partisan elections reduce the cost of gathering information for voters, and the researchers found that partisan elections “provide a relatively rational basis upon which to select” judicial candidates by voters.

Some argue that money may buy justice, and in West Virginia, money did appear to buy recently a seat on that state’s supreme court and influence a major decision there—which the U.S. Supreme Court struck down subsequently. According to the Council of State Governments, each state Supreme Court candidate in 2000 across the nation raised, on average, over $430,000, with sixteen candidates raising more than $1 million each. But Bonneau and Hall found that money is “a necessary condition for educating and mobilizing voters” and judicial elections are no different than legislaive and executive contests.

Some argue that “politics and the bench don’t mix,” that justice shouldn’t be beholden to the party affiliation of the judge and those appearing before the court. In a host of other studies which examined the argument that nonpartisan elections improve the quality of the bench, researchers have found that partisan elections do not necessarily produce less qualified judges.

Before 2004 in North Carolina, judicial officers had their partisan affiliation listed on the general election ballots. One facet that judicial politics scholars investigate is the decrease in the votes cast down the ballot compared to the top-line office on the ballot (otherwise known as “run-off”).

In four elections that were partisan (1996, 1998, 2000, and 2002), the run-off from the top of the ballot (U.S. president, U.S. Senate, or the combined U.S. House races) in those years would range from 2.83 percent for the chief justice to 8.68 percent for the court of appeals judges.

After non-partisan elections were introduced with the 2004 election in North Carolina, the run-off percentages dramatically worsened. In the 2004 election, the NC Supreme Court associate justices contests saw a run-off of 23 and 26 percent, with the court of appeals justices 27 percent off the presidential votes cast.

In 2008, in a highly competitive election and with an astounding 70 percent registered voters showing up to the polls, there was anywhere from 31 percent to 44 percent run-off for the court of appeals judges. Just last year, the decrease in votes cast for the lone court of appeals judgeship, which featured the first state-wide instant run-off system, was nearly 27 percent from the U.S. Senate race between Burr and Marshall.

While those who advocate non-partisan judicial elections argue that the electorate is better served, the electorate may think otherwise. Some contend that partisan affiliation serves as a critical “cue” to voters, along with the fact that those voters who wish to cast a “straight-ticket” ballot don’t impact the judicial elections.

The third branch of government is a critical policymaker in many regards—most recently, North Carolina education policy has been greatly impacted by a lone judge in the Leandro case. While politics may never be fully removed from those who wear the black robe, the system of judicial elections—whether partisan or non-partisan—is still in the hands of a Tar Heel electorate who seems, in recent non-partisan judicial elections at least, to not really care about these important offices of our state government.