Sam Wang over at the Princeton Election Consortium does a very nice job of analyzing polls. Mr. Wang's forecast differs from a lot of other work, including the fine work the people here at Dkos are doing.
One of the first questions he asks is what's up with the Dems?:
Senate Democrats are doing surprisingly well. Across the board, Democratic candidates in the nine states above are doing better in the polls-only estimate than the mainstream media models would predict. This is particularly true for Alaska, Arkansas, and North Carolina. In these three states, Democrats are outperforming the expectations of the data pundits (The Upshot’s Leo, Nate Silver, Harry Enten, John Sides, etc.). Why is that, and will it last?
He also presents this chart to demonstrate the differences between some of the different people looking at the polls and other factors.
The first data column is the current PEC poll median. The next two columns show what a polls-only win probability looks like. Finally, the last three columns show the media organizations’ win probabilities. All probabilities are shaded according to who is favored, the Democrat (blue) or the Republican (red). “sum6″ is the sum of probabilities (converted to seats) for six key races: AK, AR, CO, IA, LA, and NC.
Mr. Wang goes on to talk about the differences between his analysis an some of the larger media analysis.
The mainstream media organizations are a different story. They show a general tendency to be more favorable to Republicans.
...
Longtime readers of PEC will not be surprised to know that I think the media organizations are making a mistake. It is nearly Labor Day. By now, we have tons of polling data. Even the stalest poll is a more direct measurement of opinion than an indirect fundamentals-based measure.
It appears Mr. Wang leans more heavily on poll results and doesn't feel that the fundamentals have the impact others do, especially the closer we get to the election. He offers this chart of his polling analysis result history to show how the polls are beginning to settle down a bit.
I have to give credit to all the people who take the time and effort to legitimately try to make a forecast of poll results. It would be a really tough job if the public was making rational choices based on facts, but given the public's willingness to change their mind based on the most petty events in the latest 24 hour news cycle, the forecasters do very nicely with their very cloudy crystal balls.
IMHO we have some very important events coming up that may very well lend to a slight shift in upcoming polling. When president Obama announces his immigration executive actions, it's going to be almost impossible for Boehner to keep the lid on the Tea Party nuts who want impeachment. Then there's the constant leaking of Republican scandals from New Jersey to Texas, to Wisconsin, and now a new one developing in Kentucky. It will only take a shift of 1 or 2 % to keep Dems in control of the Senate and to keep Obama's judicial nominees alive.
So Mr. Wang is going against the forecasting current in giving the Dems a 70% chance of holding the Senate, but you'd have to be bit crazy to be in the forecasting business without saying something like:
Finally…I note that this is all a work in progress. I’m using PEC as a sandbox for kicking around ideas.
The one thing that will make Mr. Wang's forecast come true is: GOTV! GOTV! GOTV! GOTV! GOTV!
UPDATE: I just read an article from the HuffPost's pollster that they are Changing their model. Some good explanations as to why and worth the read if you like the geeky stuff. Here's what their new model says as of today.
And here's some of their reasoning as to why, and I must say I'm glad to see them looking at what I consider a real problem in polling.
A lot has changed about public polls since Pollster.com first launched eight years ago. In 2006, just under 3 percent of the 872 polls we tracked that measured U.S. Senate contests had partisan sponsorship (a campaign, party committee or PAC). Of the 480 Senate polls we have tracked so far in 2014, more than half either have partisan sponsors (25 percent) or were conducted by a polling firm with a partisan affiliation (35 percent).
Although the shift toward automated telephone and Internet polling was well underway by 2006, just under half of the surveys we tracked that year (49 percent) used live interviewers. This year, only 29 percent of the polls are live-interviewer telephone polls.
Perhaps most striking, just 14 percent of the Senate polls we have tracked this year were conducted by nonpartisan organizations using traditional live-interviewer phone methods.
But be careful of the above results that are leaning Dem, because a lot of the polls conducted today are registered voter polls rather than likely voter polls.
At this point in the electoral year, many of the polling organizations used to calibrate our model are still reporting their registered-voter results, which traditionally lean Democratic. That means this additional house effect adjustment nudges the estimates in the Democratic direction for now -- but the difference is slight. As of this writing, the calibration increases the Democratic margin in the 16 Senate races for which we have poll charts by an average of 0.4 percentage points, although the effect is not uniform. The margin grows slightly more Democratic in 11 races and slightly more Republican in five. As the remaining polls switch in the next weeks from reporting results of registered voters to reporting results of likely voters, this pattern will probably fade.