In 2008, as we all know, the bottom fell out of the Republican Party in the 5th District. Virgil Goode was defeated by liberal underdog Tom Perriello. But there are some problems with the standard account of how this happened.
Let's start with the basics. In an incredibly close election, Perriello defeated Goode in what was essentially a 50-50 race. Perriello won 50.08% and Goode won 49.85%. But in the Presidential race, McCain still held onto the 5th with a narrow victory of 51% to 48%.
This is the infamous Obama surge that Perriello rode to victory. But while Obama was winning Virginia with 53% of the vote, a surprising 8% improvement over John Kerry, his improvement over Kerry in the 5th was only 5% better. With the high percentage of black voters in the 5th, plus the liberal bastion of Charlottesville, many assumed that Obama would have a more significant impact in the 5th District. In fact, the swing seems somewhat muted. When I think of surge, I think of a big swing to one party or candidate. There was certainly a swing, but more of a nudge than a surge.
Part of Obama's improvement over Kerry was reduced because of McCain's improvements over Bush. The 5th was one of the few congressional districts in Virginia where McCain received more votes than Bush in 2004. Certainly Obama did a better job than McCain in increasing support, but McCain was not nearly as lackluster as some commentators analyzing the 5th have indicated.
Let's look at how Goode compares to McCain across the district.
Albemarle County: 40% McCain, 37% Goode (-3%)
Appomattox County: 64% McCain, 64% Goode (0%)
Bedford County: 66% McCain, 62% Goode (-4%)
Brunswick County: 35% McCain, 38% Goode (+3%)
Buckingham County: 49% McCain, 50% Goode (+1%)
Campbell County: 68% McCain, 64% Goode (-4%)
Charlotte County: 55% McCain, 56% Goode (+1%)
Cumberland County: 51% McCain, 52% Goode (+1%)
Fluvanna County: 50% McCain, 48% Goode (-2%)
Franklin County: 61% McCain, 62% Goode (+1%)
Greene County: 60% McCain, 54% Goode (-6%)
Halifax County: 51% McCain, 53% Goode (+2%)
Henry County: 56% McCain, 57% Goode (+1%)
Lunenburg County: 51% McCain, 50% Goode (-1%)
Mecklenburg County: 52% McCain, 55% Goode (+3%)
Nelson County: 45% McCain, 44% Goode (-1%)
Pittsylvania County: 62% McCain, 62% Goode (+0%)
Prince Edward County: 44% McCain, 46% Goode (+2%)
Bedford City: 55% McCain, 50% Goode (-5%)
Charlottesville City: 20% McCain, 19% Goode (-1%)
Danville City: 40% McCain, 42% Goode (+2%)
Martinsville City: 35% McCain, 39% Goode (+4%)
Across the district, Goode's crossover has been shattered. He performs only 1% better than John McCain in his home county of Franklin! Henry is about the same, with only the City of Martinsville showing any significant crossover for Goode. His best counties outside of his home base are Mecklenburg, the only county with in a North Carolina media market and outside of Perriello's ad campaign, and Brunswick, the most isolated part of the district. Even then he outperforms McCain by only 3%.
A consistent 3% could have saved Goode, but in 2008 he is facing a Democrat with resources able to compete across the district. He's hit hard not only in the northern counties like Greene and Albemarle, but sees a significant crossover of McCain voters to Perriello in Bedford and Campbell County. I have no idea what was going on in these Lynchburg suburbs, but it's something to look for in 2010.
From 2000 to 2006, the Republicans had a good run in the 5th District. This included a counter-cycle performance in 2006 that only narrowly brought down their performance. If 56% was an elevated ceiling for Republican support in the district during good times, it's easy to see them dropping to 53% in what was a solid Democratic wave year. What's impressive for the Republicans is that it didn't drop even more.
Looking back, I think stories about African-Americans, liberals, and college students are a little overplayed. Yes, the 5th has them. But it also has farmers, conservatives, and regular blue collar workers. And even some Hillary voting PUMAs. This district is very polarized and has few swing voters. Which is why the Republican surge in 2004 was weak in the district, and why the Democrats were unable to capitalize on their own surge in 2006 or 2008. At the same time, the electorate is shifting slowly toward the Democrats as Charlottesville grows and Southside declines.
What surprises me is that the few swing voters out there, at least the swing voters in 2008, weren't in Charlottesville and weren't in Southside (at least what most people call Southside). Bedford and Campbell, two of the main suburbs of Lynchburg, stand out for putting Perriello over Goode. Why? The Lynchburg metropolitan area is one of the bastions of the Republican Party, particularly because of social issues. In 2006, the region turned out strongly for the anti-gay marriage amendment. This is the home of Liberty University! Jerry Falwell Jr. has donated to Hurt's campaign, but I have seen little emphasis on social issues from Hurt's campaign. Instead, I've seen Perriello wisely vote against federal funding of abortion and work to downplay potentially controversial social issues. Smart move. This is an area to watch for me on election night.
Before we go, we'll look at the Survey USA polls from 2008. Like 2006, Survey USA had three polls of the 5th District race:
August 10-12: Goode 64%, Perriello 30%
October 6-7: Goode 55%, Perriello 42%
October 30-November 2: Goode 50%, Perriello 47%
Looking at some of the internals, we find that the first poll from August had 17% black voters, equal to their 2006 polling. The October poll also had 17% black voters. But then right before the election, Survey USA's poll put it at 22% and said that Perriello could win with 25%. After the election, as they argue their polling in 2010, they seem to be arguing that the black percent of the vote in 2008 was 25%, far higher than any other source I've heard.
In other words, throughout the 2008 election Survey USA tried to convince everyone that they didn't see ANY increase in black turnout from 2006 to 2008. Then, at the last possible second, they put out a poll showing black turnout significantly higher and Perriello within the margin of error. This was after both parties had internal polling showing the race a dead heat and Goode in risk of being defeated.
Let's look at party. August: 41% Republican, 33% Democrat, 20% Independent. October: 38% Republican, 34% Democrat, 21% Independent. Final: 38% Republican, 40% Democrat, and 20% Independent.
Poll by poll, Survey USA shows an electorate more and more favorable to the Democrats, with Democrats eventually outnumbering Republicans by 2%. This is a 14% swing from 2006, when they had the final poll with 44% Republicans, 32% Democrats, and 22% Independents. Yet the outcome of the 5th district swung to the Democrats by only 5%. Is it possible that Survey USA isn't tracking changes in voters, but changing in party identification? Throughout 2008, people who voted Republican were more likely to call themselves independent, after the anger at Bush's reckless spending? A prelude to the trend that continues today with the Tea Party.
Finally, ideology. August: 39% Conservative, 34% Moderate, and 12% Liberal. October: 40% Conservative, 31% Moderate, and 13% Liberal. Final: 41% Conservative, 35% Moderate, and 17% Liberal.
Ironically, while at the same time showing a surge in Democrat voters, Survey USA sees a slight increase in conservative voters during the 2008 campaign! Liberals are higher, but the breakdown as a whole seems very similar to Survey USA's final poll from 2006. This gives support to my theory that changes in the party breakdown in the poll is due to people changing how they describe themselves, not who is voting. Instead, conservatives started to shift to calling themselves independent, but still turnout and vote Republican.
And the two competing polls from Survey USA right now? One has blacks at 12%, which is less than even 2006. The other has them at 20%. Which is right? I am leaning heavily toward the 20%. Signs point to 2006 being a bad year for Democrats in Southside, particularly in African-American areas. 17% may be the low point that is produced by a Democrat campaign with no resources for GOTV. 20% sounds reasonable, especially if you're putting 2008 at 25% as Survey USA does.
Party? The RDD has it at 36% Republican, 32% Democrat, and 29% Independent. This would have Democrats at about the same turnout as 2006, but with a shift of Republicans to Independents from 2006. This is consistent with the idea that conservative Republicans call themselves independents as they affiliate with the Tea Party movement, but I also don't want to read too much into Survey USA's party breakdown given how large their swing was in 2008 despite few changes in the outcome.
Finally, ideology. 47% Conservative, 35% Moderate, and 10% Liberal. This is a swing of around 7% from 2008 and 2006. If true, it means the electorate in 2010 will be even more favorable to Republicans than in 2006, which was already a sensible year for them in the Fifth District.
It's very hard for me to sort out the twists and turns in Survey USA. That the only way they seem able to change Perriello's % is through changing the racial composition of their sample seems questionable to me. They did it in 2008 and again in 2010. Is it possible that they are bad at sampling white voters in this sprawling, diverse district and they compensate through fiddling with the black %? We'll find out . . .
Next, a look at the polling in 2008 as we try to sort out what's going on.