by Ruy Teixeira
(cross-posted at www.washingtonmonthly.com/showdown06)
The blizzard of polls released over the weekend and today suggest some tightening of the race, but do not appear to fundamentally alter the assessment I offered five days ago in my last update. Tuesday should still be a very good day for the Democrats.
Start with Bush’s approval rating. Taking the latest polls into account, Charles Franklin’s trend-based estimate now stands at 38 percent. Pretty bad for the incumbent party.
Turning to the generic congressional ballot, confusion abounds, so let me try to separate signal from noise as best I can.
There have been seven polls released in the last couple of days. Here are the results for likely voters (LVs):
CNN (Fri-Sun): +20D
Newsweek (Thu-Fri): +16D
Time (Wed-Fri): +15D
Fox (Sat-Sun): +13D
USA Today/Gallup (Thu-Sun): +7D
ABC News/Washington Post (Wed-Sat): +6D
Pew (Wed-Sat): +4D
Quite a spread! And here are the same polls, this time for registered voters (RVs):
CNN (Fri-Sun): +15D
Newsweek (Thu-Fri): +16D
Time (Wed-Fri): +15D
Fox (Sat-Sun): no RV data
USA Today/Gallup (Thu-Sun): +11D
ABC News/Washington Post (Wed-Sat): +10D
Pew (Wed-Sat): +8D
Somewhat closer together, but still a fair amount of variation.
Here are some observations on these data that may help make sense of them.
1. Charles Franklin’s trend-based estimate (which actually doesn’t include the most recent two polls, CNN (+20D) and Fox (+13D)) still estimates the Democratic advantage at 11 points.
2. The average LV Democratic advantage in these 7 polls is around 12 points. The average RV Democratic advantage is around 13 points. Still very good in either case.
3. Note that, reflecting the widely varying methodologies these pollsters use, the relation between RVs and LVs in these polls varies widely. Some (CNN) have the Democratic advantage among LVs actually larger than among RVs; some have it exactly the same (Time, Newsweek); and some have it smaller (Pew, Post, Gallup–interestingly by exactly the same 4 point margin).
4. Of course, it is entirely possible that some of these pollsters’ LV methodologies are better than others. And there are certainly reasons to be skeptical that Democrats will actually manage a double digit lead in the popular Congressional ballot on election day. So let’s say that, for example, Gallup has it about right–and they do have a good track record in the last several offyear elections.
Well, as Gallup points out, a seven point lead ain’t chopped liver. Here’s some of what they have to say:

Gallup has modeled the number of seats a party will control based on that party’s share of the national two-party vote for the House of Representatives using historical voting data in midterm elections from 1946 to 2002. The model takes into account structural factors such as the party of the president and the majority party in Congress entering the elections. The results suggest that a party needs at least a two percentage-point advantage in the national House vote to win a majority of the 435 seats. Based on this historical analysis, the Democrats’ seven-point margin suggests they will win a large enough share of the national vote to have a majority of the seats in the next Congress.
More specifically, taking the final survey’s margin of error into account, the model predicts that the Democrats could gain anywhere from 11 seats on the low end to 58 seats on the high end — with 35 seats being the most likely number. Given that Democrats need to gain just 15 seats to wrest control from Republicans, a Democratic takeover appears likely.

5. Some of the variation in the LV samples is no doubt due to varying estimates of how many Democrats vs. how many Republicans are projected to be in the voting electorate. Pew, which gives the Democrats the smallest estimated lead, has Democrats and Republicans at parity among likely voters. Gallup gives the Democrats a 2 point edge in representation among LVs. And Fox, whose estimated Dem lead among LVs is very close to the average of all these polls, gives the Democrats a 4 point representation edge.
It will be interesting to see on election day who’s got that part of equation right.
6. Turning to independents, even in the Pew poll, independents are giving the Democrats a 10 point advantage. Gallup and Fox have independents voting Democratic by 15 and Newsweek has the margin at 25. This will be an important data point to track and indicates, even at the low end of this range, the Democrats will be well-served by high turnout of independents in this particular election. To remind folks once more of the historical context on the independent vote:

As far back as I can get data (1982), the Democrats have never had a lead among independents larger than 4 points in an actual election, a level they managed to achieve in both 1986 and 1990. Indeed, since 1990, they’ve lost independents in every congressional election: by 14 points in 1994; by 4 points in 1998; and by 2 points in 2002.

So this election could represent quite a turning point in this pattern.
As for the race by race data, not a great deal has changed since my last post five days ago. For example, the Bafumi-Erikson-Wlezien seat shift model, which forecasts the level of seat shifts through computer simulations of the 435 individual House contests, looks like it would produce about the same result today as it did two weeks ago (a 32 seat Dem gain), if I’m understanding the inputs into their model correctly. (This is very similar to Alan Abramowitz’ forecasting model–not based on computer simulations–which calls for a Democratic pickup of 29 seats).
Also, Democracy Corps has released their final survey of 50 competitive Republican House districts and they’re showing a slightly compressed, but still impressive, 5 point Democratic lead in the named Congressional ballot in these districts. Note also, that the DCorps survey shows the Democrats with a 16 lead among independents in these districts.
Majortity Watch has not done any new polls, so nothing to report there. Over at Pollster.com, where Mark Blumenthal and Charles Franklin collect all the available public polling on all the House races their current scorecard assigns 219 seats to the Democrats with 29 tossups. Assuming the Democrats and Republicans split the tossups, that would bring the Democratic total to 233 seats–a gain of 30 seats over where they now stand.
Over in nonpartisan pundit-land, Charlie Cook is holding steady in his prediction of a 20-35 seat Democratic pickup (let’s pick the midpoint and call it 28 seats). Stuart Rothenberg has the Democratic gain between 30 and 36 seats (let’s call it 33). And Larry Sabato has the Democratic gain pegged at 29 seats.
You know, I think I’m beginning to detect a pattern here. It will be interesting to see how it all turns out when the real world talks back.
Turning to the Senate, the latest Pollster.com 5-poll averages show the Democrats up by 6 in NJ and 4 points in MD, the two seats the Republicans have been given some chance of picking up. And they are leading by 1,3,13,6,12 and 2 points, respectively in MO, MT, OH, RI, PA and VA. Thus, figuring strictly on the basis of these data, one would see the Democrats picking up six seats, but with agonizing nail-biters in at least MO and possibly also in VA and MT.
Of course, several of these races are so close in the polls, one can hardly pronounce with a huge degree of confidence that the Democrats will, in fact, get their six seats. But it certainly seems like a reasonable possibility. Checking our nonpartisan pundits, it’s worth noting that both Rothenberg and Sabato see the Democrats getting the six seat pickup. Cook is more circumspect, calling for a 4-6 seat Democratic pickup.
Well, that’s it for the updating. On to the biggest poll of ’em all: election day!