Friday, August 27, 2010

Republicans Best Targets

In any election a party wants to win seats. For the most part it doesn’t matter which seats you win, since any combination can get you to the majority. The seats that are most valuable are those that you can hold for years. Republicans picked up those type of seats. So when they got the majority in 1994, they were able hold it for 12 years. The chart below shows Republican 1994 pick-ups, what type of districts they were, and how many terms they held them.

The first thing that jumps out at me is that more than half the pick-ups were very Republican districts. These were districts that were opportunities in most years, but in a favorable Republican year they were able to pick them up. Of the 36 Dark Red or Light Red districts Republicans held 32 of them for at least six terms until the Democrats had their wave election in 2006. Any seat you only lose when the environment is favorable to the other side is a good pick-up. While the Republicans held 89% of all red districts, and still hold 61% of them, only 15% of the purple to blue districts were held for even four terms. We don’t need to look at analysis to tell us that the more Republican a district is, the more likely the GOP is to hold it. While the sample size isn’t great, the difference is far greater than I’d expect it to be.

Here’s are the districts the Democrats won in 2006:

While 64% of Republican pick-ups in 1994 skewed Republican, only 8% of the Democratic pick-ups did. Wave elections swing back when circumstances are more normal. Because Republicans picked up Republican seats they were able to hold onto most of them. The Democrats are unlikely to be so lucky. Here are the 95 seats that pretty much everyone agrees are in play:

Democrats picked up 20 red seats in 2006 and 2008. The maximum Republicans should pick up this year is 7. That bodes much better for Republican chances to hold the seats. In 2012 there will be a new census and new congressional districts. There are some large states (California, New York, Illinois, Massachusetts) where Republicans have the possibility of unfavorable gerrymanders that’d make the seats difficult to retain. In addition, there are several states (Iowa, New Jersey, Ohio, Pennsylvania, Michigan, Minnesota) which will lose congressional seats. The district won might be lost regardless of the redistricting.

Thursday, August 26, 2010


There are approximately 113 Democratic districts which are considered in play. I’ve seen polls on 77 of those. Republicans were ahead in 38, Democrats 37, and 2 were even. You’d think the remaining 37 districts would be ones that are unlikely to switch. Yet there are districts in that group (LA-3, NY-29, KS-3, TN-6) that almost certainly will switch parties. While Democrats lead in half the districts, they’re all Democratic districts. Winning half in those means big losses.

Of the 22 Republican districts that might be in play, Republicans lead in 10, Democrats lead in 3, and 9 haven’t been polled.

Extrapolating these results for the remaining districts, you’d have 57 districts switching from Democrat to Republican and 5 going the other way.

Many of these House polls are Republican campaign internals or done by We Ask America and American Action Forum. American Action Forum is a Republican pollster. I don’t believe We Ask America is. Some people want to dismiss these polls, especially a campaign’s internal poll. You shouldn’t dismiss a poll simply because it was done for a campaign. The campaign wants to get an accurate poll. If a pollster fudged the numbers, they wouldn’t get hired again. If it’s a reputable pollster, you should take that into account.

Of course, a campaign can fudge the numbers they release. A pollster might not contractually allow that. If they get a reputation of being a poor pollster they won’t get a lot of business. Study the methodology, look at the questions (if available), and check to see if they’re using registered or likely voters.

The most telling indicator of a poll is a campaign’s reaction to an opponent’s poll being released. If they don’t release their own poll, you have to assume it says the same thing. After independent pollster Reuters/Ipsos released a poll showing Michael Bennett down nine points in the Colorado Senate race, he immediately came out with a poll having him up four points. You can make your own judgement of which one to believe, but all the districts that don’t answer bad poll numbers are fairly telling in their silence.

Certainly the likely voter could change. The Democratic hope that Republicans won’t be highly motivated is a false one. Republican enthusiasm has increased since Scott Brown won in January, and Democrats should know from 2006 and 2008 the job isn’t done until you vote.

Turning around the independent vote could work in some cases. Harry Reid’s “be afraid of Sharron Angle, she’s going to steal your medicare and social security” may genuinely make people afraid. Never underestimate the politics of fear. Very few Republican candidates are Sharron Angle and few Democrats have the skill and resources Reid does. It seems unlikely that Democrats can turn independents in most cases.

Democrats could motivate their base, but their efforts so far have met with failure. Many on the far left feel let down and believe that if congress had just closed Guantanamo, gotten out of Afghanistan, and instituted single payer healthcare everything would be fine. Their disappointment may rival Republican disappointment that kept them home in 2006 and 2008.

Tuesday, August 24, 2010

Sorry, Charlie

Two months ago, I predicted a Rubio Florida senate win, partly because it’s so difficult for an independent to win. Here was my prediction:

I painted a bleak picture for Crist, maybe a little too bleak. To run even with Rubio he was going to need something more like this:

At the time, however, Crist was doing better with Republicans and Democrats. Well, PPP has a new poll out of likely voters.

It’s worse for Crist than he was doing, but not nearly as bad as I predicted. Rubio’s strong showing with Democrats is a surprise, but I’m not surprised that Crist now trails by 49% among Republicans. He trailed by 31% in July. Crist needs to beat Meek with Democrats and a larger lead with independents if he’s to win. I was counting on a 39% R/35% D/26% I electorate. PPP has a 42% R/42% D/14% I electorate. Since Crist is picking up the most votes with independents that’d be a killer. With my electorate he trails by 5.5% instead of 8. Still, it’s hard for me to see a road to victory for Crist.

Thursday, August 19, 2010

It’s Where You Start That Matters

538 has been trying to correlate the generic ballot with how many seats a party will pick up. Where do the Republicans have to be to pick up the 39 needed seats? I’d read that 2004 was a good Republican year in House votes. That was perplexing. The GOP actually lost 3 seats that year. How can that be a good year? I decided to plot Democratic gains and losses with their share of the House vote.

In 1976 the Democrats got 56.9% of the House vote and gained 1 seat. In 2004 they got 48.9% of the vote and only lost 3 seats. While there is some correlation between percentage of the vote and gains/losses, the chart makes more sense if we look at how many seats the Democrats ended up with after the election.

Now that’s more like it. Since every seat is up every election, the amount they previously had is irrelevant.

There aren’t enough data points to say definitively, but every time the Democrats got 50.9% of the vote or more, they got at least 233 seats. Whenever they got 50.2% or less, they never got above 212. We can extrapolate and predict that Democrats need somewhere between 50.4% and 50.6% of the vote to likely get a majority.

In 2004 the Democrats got 48.6% of the votes and 202 seats. In 2006 they got 54.1% and 233 seats. They needed to increase that share to 55.6% in order to get to 257 seats. Thus, even if Democrats beat Republicans overall by 7 or 8% they’ll likely still lose seats.

What’s interesting is that no matter how bad the Democrats did, they never dropped below 202 seats. Even is 1994 when they were beaten 53.5% to 46.5%. I’m not sure what to make of that. Republicans had over 285 different districts since 1994. So it’s not like all 202 are safe. In fact, in the 2002 redistricting, the Republicans picked up more new districts and lost fewer old districts than the Democrats. In addition, there are a number of districts (e.g. AR-1, AR-2, WV-1, TN-6, TN-8) that Republicans are almost certain to win this year that weren’t ever among the 285+ they had before. It seems conceivable that Republicans could pick up 70 or 80 seats this year, especially with them possibly beating 1994’s 7 point win. The high water mark, however, has been 232. That’d still be a healthy net of 53 seats.

The biggest reason the Democrats can lose so many seats this year is because they have 256 and they’d need another landslide to retain that many. Since the Democrats haven’t held a House majority with less than 50.9% of the votes, it’d be a complete anomaly if they retained the House with less than that.

Wednesday, August 18, 2010

Washington Thoughts

Based on current totals Republicans beat Democrats by about 12,000 votes in the Washington Senate race. About 15,000 going to other parties. These could matter, since they can't vote for other parties in November.

Around 34,000 more people voted in the senate race than the congressional races. The congressional votes for third parties were almost triple those in the Senate vote. This isn't surprising. Since a number of the congressional races are unlikely to be competitive, people feel more freedom to vote for a candidate of choice or not at all. Democrats got more votes than Republicans in the congressional races. Overall, however, 41,000 more people voted Republican in the senate race and only 19,000 voted Democratic. It's likely that more people voted for a Democrat for congress, but also voted for a Republican for Senate than vice versa.

This close to the election I doubt you'll see much change in the electorate. It'll likely come down to a few thousand votes and a few mysterious "new" boxes of ballots from Democratic areas.

The Washington primary in 2008 showed only small changes from the primary vote to the general election vote. That year Democrats had around a 215,000 vote advantage in the primary. This year the Democrats only had a few thousand more votes than Republicans. In a normal year, it’d be in the middle. The chart below lists Democratic primary margins in 2008 and 2010

There is no Republican this year in the 7th. The first thing that jumps out at me is that there wasn’t much drop off or even a Democratic gain in the two heavily Republican districts. It’s likely the 3rd has the biggest drop-off because the Democratic incumbent retired. Every other Democratic district had anywhere from a 15% to 21% shift from 2008 to 2010. That’s fairly consistent. Races that were narrow Republican victories in 2008 should be safe. Close losses should go Republican. Races that weren’t close two years ago will be toss-ups. Any Democratic incumbent that didn’t get 60% of the votes in 2008 needs to look in his or her rear view mirror. Competitive open seats like WA-3 could see a larger change.

In Washington, however, that may not be a big deal. They had big victories then. Dave Reichert, WA-8, had a close win last time. He probably has nothing to worry about this time. Jaime Herrera will very likely pick up WA-3 for the Republicans. While WA-2 and WA-9 are close, they aren’t close enough that the GOP could steal those also. Most of the forecasters didn’t have these two districts on their lists anyway.