Political Geekery 101...

Trying to find a neutral pollster this election season is like trying to find another conservative here on campus to talk with (first day of classes, I had to say something) but time and again Larry Sabato does this with amazing accuracy. Here is his (long) and wonkish article from RCP

September 2, 2010
The Crystal Ball's Labor Day Predictions
By Larry Sabato

For decades I've advised students to let the facts speak for themselves, while avoiding the indulgence of shouting at the facts. In other words, we should take in all the available, reliable information; process it; and let the emerging mosaic tell its story-whether the picture pleases or not. The human (and partisan) tendency to twist facts into pretzels in order to produce a desired result must be avoided at all costs.

We've been patient and cautious here at the Crystal Ball as a year's worth of facts has accumulated. We've sifted the polls, cranked up the models, and watched the candidates and campaigns closely. All political observers have "gut feelings" about an election year, but feelings make for good songs and lousy predictions. Forecasting is an imprecise art. People who get too far ahead of the facts or are too insistent about what will happen are usually partisans-openly or in disguise.

The Crystal Ball's predictions are clinical. We are fond of people in both parties. We cheer for no one.

2010 was always going to be a Republican year, in the midterm tradition. It has simply been a question of degree. Several scenarios were possible, depending in large measure on whether, or how quickly, the deeply troubled American economy recovered from the Great Recession. Had Democratic hopes on economic revitalization materialized, it is easy to see how the party could have used its superior financial resources, combined with the tendency of Republicans in some districts and states to nominate ideological fringe candidates, to keep losses to the low 30s in the House and a handful in the Senate.

But conditions have deteriorated badly for Democrats over the summer. The economy appears rotten, with little chance of a substantial comeback by November 2nd. Unemployment is very high, income growth sluggish, and public confidence quite low. The Democrats' self-proclaimed "Recovery Summer" has become a term of derision, and to most voters-fair or not-it seems that President Obama has over-promised and under-delivered.

Obama's job approval ratings have drifted down well below 50% in most surveys. The generic ballot that asks likely voters whether they will cast ballots for Democrats or Republicans this year has moved increasingly in the GOP direction. While far less important, other controversies such as the mosque debate and immigration policy have made the climate worse for Democrats. Republican voters are raring to vote, their energy fueled by anti-Obama passion and concern over debt, spending, taxes, health care, and the size of government. Democrats are much less enthusiastic by almost every measure, and the Democratic base's turnout will lag. Plus, Democrats have won over 50 House seats in 2006 and 2008, many of them in Republican territory, so their exposure to any sort of GOP wave is high.

Given what we can see at this moment, Republicans have a good chance to win the House by picking up as many as 47 seats, net. This is a "net" number since the GOP will probably lose several of its own congressional districts in Delaware, Hawaii, and Louisiana. This estimate, which may be raised or lowered by Election Day, is based on a careful district-by-district analysis, plus electoral modeling based on trends in President Obama's Gallup job approval rating and the Democratic-versus-Republican congressional generic ballot (discussed later in this essay). If anything, we have been conservative in estimating the probable GOP House gains, if the election were being held today.

In the Senate, we now believe the GOP will do a bit better than our long-time prediction of +7 seats. Republicans have an outside shot at winning full control (+10), but are more likely to end up with +8 (or maybe +9, at which point it will be interesting to see how senators such as Joseph Lieberman of Connecticut, Ben Nelson of Nebraska, and others react). GOP leaders themselves did not believe such a result was truly possible just a few months ago. If the Republican wave on November 2 is as large as some polls are suggesting it may be, then the surprise on election night could be a full GOP takeover. Since World War II, the House of Representatives has flipped parties on six occasions (1946, 1948, 1952, 1954, 1994, and 2006). Every time, the Senate flipped too, even when it had not been predicted to do so. These few examples do not create an iron law of politics, but they do suggest an electoral tendency.

The seat switches are probably coming in Arkansas, Colorado, Delaware (but only if the eventual GOP nominee is Rep. Mike Castle), Indiana, North Dakota, and Pennsylvania. We expect Republicans to pick off at least a couple of these states: California, Illinois, Nevada, Washington, and Wisconsin. While it is possible that Republicans will lose one or two of their own open seats, the only 50-50 chance of that right now is in Florida-and it might not happen even there. There can also be unanticipated shockers if a GOP wave develops. While we rate Gov. Joe Manchin (D) the early favorite to fill the late Sen. Robert Byrd's seat, his Republican opponent, John Raese, is a self-funder in a strongly anti-Obama state.

The inescapable conclusion is that the Senate is on the bubble, with only a slight lean at Labor Day toward Democratic retention.

The statehouses will provide the third leg of the Republicans' 2010 victory. We have long suggested the GOP would gain a net +6 governorships. We now believe they will win +8. This boon to the GOP for redistricting will be enhanced by a gain of perhaps 300 to 500 seats in the state legislatures, and the addition of Republican control in 8 to 12 legislative chambers around the country.

Republicans are likely or even certain to gain the governorships in Iowa, Kansas, Michigan, Oklahoma, Pennsylvania, Tennessee, Wisconsin, and Wyoming. We believe the GOP candidates also have an edge in Illinois and Oregon-both of these quite surprising. Democrats will also pick up a few statehouses to cut their losses: Hawaii is near-certain, with fair to good shots in Connecticut, Minnesota, and Rhode Island (though we currently retain the last two as toss-ups).

There is no question that the ratings in some close races will change as scandals emerge in the coming sixty days or as the big primaries on September 14 occur. The contests listed only as "leaning" one way or the other are the most vulnerable to shift in the next two months. Unless an unexpected, extraordinary number of these changes are in favor of the Democrats, it is hard to see how Republicans can fail to do very well.

We still have 6 toss-ups for Senate (California, Florida, Illinois, Nevada, Washington, and Wisconsin-seats held by the Democrats in all but Florida ). The 9 toss-ups for Governor are in California, Florida, Georgia, Maine, Maryland, Minnesota, Ohio, Rhode Island, and Vermont-six currently held by GOP chief executives and three by Democrats. In the House, the Crystal Ball counts 29 toss-ups, 28 seats held in this Congress by Democrats and just one by Republicans.

As always, the Crystal Ball will make a guess in every contest before Election Day. Some will be moved sooner, and a few head-scratchers will only be categorized at the very last minute. We're proud of our record, with more than 98% of the contests called correctly over the decade-long life of the Crystal Ball. In some years, our overall seat changes in each category have been exactly on the button, and in 2008 we were a single electoral vote off the Obama-McCain finish of 365-173. But we fully admit here and now that, as always, we'll get some of them wrong. We truly believe our website motto: "He who lives by the Crystal Ball ends up eating ground glass." That's part of the fun of politics. Moreover, remembering the great Gallup goof of 1948, we know there's an election year somewhere up ahead when the Crystal Ball cracks so badly we will have to export it to Switzerland for expert repairs.

In 2006 and 2008 the Crystal Ball was full of good news for Democrats, while this one may cause a run on Prozac among our Democratic readers. Is there any way back for the Democrats in the eight remaining weeks?

A political historian always thinks of election-changing events from the past, such as the Cuban Missile Crisis in October 1962 which cut Democratic midterm losses a couple of weeks later, or the autumn 1998 GOP push for Bill Clinton's impeachment that backfired at the polls, eliminating expected GOP gains in the Senate and resulting in a drop of Republican seats in the House. We know what we don't know: Something big and unexpected can always drop from the skies. Naturally, it can work the other way, too-say, a substantial increase in unemployment or negative GDP growth just before Election Day.

If there isn't a dramatic development that has overarching political implications, then Democrats will have to depend on their financial edge, tested candidates, and leaders finding ways to motivate the troops to a far greater extent than we see today (or witnessed in the 2009 off-year elections).

Any shift will quickly show up in the generic ballot match-up among likely voters. Gallup's generic ballot question reads: "If the elections for Congress were being held today, which party's candidate would you vote for in your congressional district - the Democratic Party's candidate or the Republican Party's candidate?" In a midterm election, only about 40% of the adults will come to the polls (or vote early), compared to 63% in the presidential contest of 2008. If average Democrats start to demonstrate interest in the election that is comparable to the intensity being shown by Republicans, the generic ballot response among the minority of Americans likely to participate in the midterm election will tighten. While a recent Gallup poll had the GOP leading the generic ballot by a massive 10 percent, on average Republicans are ahead by about 5 percentage points-still quite high by historic standards.

Overall, though, a strong bet is that 2010 will generate a substantial pendulum swing from the Democrats to the Republicans. It is not that Republicans are popular-most polls show the party even less liked than the Democrats. Many observers find it amazing that the less-liked party is on the verge of triumphing over the better-liked party. Nevertheless, in the time-honored American way, voters will be inclined to punish the party in-power by checking and balancing it with more members from the opposition party.

Each week we will update our ratings in each category, all the way to election eve, November 1st. In that sense, the predictions we make below are the general election's starting points, not the guaranteed finish. There is a lot of good politics to come in September and October, so stay tuned.