Articles and Analysis


US: Generic Ballot (Gallup, Rasmussen 6/21-27)

Topics: Generic House Vote , poll


6/21-27/10; 1,600 registered voters, 3% margin of error
Mode: Live telephone interviews
(Gallup release)

2010 Congress: Generic Ballot
46% Democrat, 45% Republican (chart)

6/21-27/10; 3,500 likely voters, 2% margin of error
Mode: Automated phone
(Rasmussen release)

2010 Congress: Generic Ballot
44% Republican, 38% Democrat (chart)



Gallup's article last week admitted that their "registered voter" model underreports Republicans by 8% so these two results are nearly identical - for the second time in two weeks.



This is a switch. The GOP has been too obstructionist and has fought Obama's policies much of the time, not because they are radical and far left, but because they know they can win political points by postponing important legislation. Financial reform is unforgiveable; energy reform in constrast, is more complicated and increased taxes could affect productivity. That is why I think both sides can find something that they can find common ground on.



Tightening in the rasmussen numbers. Now 6 points, used to be 10.



I think the best indicators of the midterms will be if Independents either favor Democrats, which isn't likely, or favor the GOP by 5 points or less. I think with 2006 turnout from the Democratic base, and moderate turnout from Independents, Democratic losses in the house could be about 20-25 and in the senate 4 or 5 seats; If they lost Lincoln, ND, DE, and IN, and even NV, they could hold on to the others, and pick up at least 2 of the following; Missouri, NC, KY, NH, or OH. Even if they picked up just one of those and lost the other 5 that would be 55 Dems, 44 Republicans, and hopefully Charlie Crist.



There's no real tighteing going on. its just the normal ebb and flow of these types of routine polls. Similar to the ebb and flow of the daily trackers of Obama.

Looks about right though, the GOp probably has a nice solid lead of about 5-8 points.



The Gallup generic ballot in November 1994 was tied at 46%-46% and the Republicans picked up 52 seats. If the GOP goes into the 2010 elections either tied, slightly ahead or slightly behind in the Gallup generic ballot question, they will be in good shape. If they suddenly fall behind by 4% or 5%, they probably won't pick up but somewhere in the 15-20 seats range.



In 1994, Democrats had a huge cash disadvantage to what they have now, and they really didn't know what was coming. Perhaps Gallup's polls were tied, and it was like the train the Democrats didn't see coming.

Not as many polling firms competed like they do now and it wasn't as commercialized.

I tend to like to look at polls from the past 8 years.


Field Marshal:

Good points JMS. To add to that, in Sept, Gallup changes to a likely voter model. When that occurs, look for the GOP to gain a few percentage points. Thus, if by labor day the Dem are up by 1, it would probably translate to a 5 point GOP advantage going into the fall. At that level, 1994 will look small compared to the year the GOP will have in 2010.



Obama's now at 49% approval in Rasmussen among likely voters.

Democrats now lead the generic ballot in Gallup (+7 point improvement from 2 weeks ago.)

Dems are up 2 in Rasmussen from last week, up 4 overall from two weeks ago.

This week, Republicans lost 10 points in Gallup among Independents while the Democrats gained 3.



"The Gallup generic ballot in November 1994 was tied at 46%-46% and the Republicans picked up 52 seats. If the GOP goes into the 2010 elections either tied, slightly ahead or slightly behind in the Gallup generic ballot question, they will be in good shape. If they suddenly fall behind by 4% or 5%, they probably won't pick up but somewhere in the 15-20 seats range."

I'm not sure there's a true correlation between the two elections. The message in 1994 appealed to middle class suburban moderates. The current message appeals to rural whites and the affluent. Those are groups that the republicans already dominate. It won't do any good if they win rural districts by 40 points instead of 25...

I really think republicans are over playing their hand, and a lot of races in november will end up like the spec. election in the PA 12th. That's a really shame...someone needs to smack the Big O down a notch (just, you know, NOT someone in the military chain of command...)



@ Farleftandproud

The word out there in the land is that the Democrats are going to blow tens of millions on trying to get young people and minorities to the polls by reminding them how much they love Obama. If that is what they spend their money on, the money advantage will get wiped out quick. Oh, and if unions keep wasting millions trying to defeat Democrats in primaries, that won't help either.



As I've said before, people shouldn't get all exorcised one way or another about the week-to-week changes in the generic ballot. Especially since it's still June.



@ rdw4potus

You could be right, but in just about every election cycle where the Republicans do well in House and Senate races, they either run behind or, at best, even on the generic ballot. Nate Silver pointed this out a few months ago.

Here are some more examples other than 1994.

In 2002, Democrats led on the generic ballot according to Gallup, 49%-43%, but Republicans actually picked up a few seats in both the Senate and House.

In 1990, Gallup had the Democrats up 50%-42% on the generic and there was very little change in the strength of either party.

In fact, before this year, Gallup had only shown the Republicans leading on the generic ballot once in the last 60 years and that was in Sept 1994.

And this phenomenon is not just happening with Gallup. You look at the generic ballot numbers for pretty much any of the mainline polling firms and they almost never show the Republicans beating the Democrats, even in years that they do very well on election night.

That is why I think if you are a Democrat and want to see the Republicans fall on their faces in November, you probably shouldn't feel good about things unless the Democrats are up by 5%+ on the generic.


Field Marshal:

Looks like PPP will show a tight race in Wisc tomorrow confirming Ras' poll last week.



If PPP shows it, I believe it. They have been the far and away most accurate pollster in state-wide races in 2009 and 2010.



The democrats have lost almost all of their lead with hispanics (down to 58%). They completely lost the independents. They have demoralized the liberals. They have lost all of the conservative democrats.

My son is out there among all of those excited young voters that were enthused about Obama in 2008. He said that he doesn't have a single friend that supported Obama in 2008 that still supports him today.

The people that propelled Obama into office are the ones that have no enthusiasm left and traditionally have a very poor showing in mid-term elections.

There is a lot riding on this Friday's jobs report. If it shows a job loss, the momentum toward republicans might become too strong to stop. Especially when the government will be getting rid of 500,000 census workers before the election. If the private sector doesn't add at least that many jobs, we could see unemployment go over 10% before November.

The democrats might plan to spend a ton of money to rebuild the enthusiasm but you can't sell a fad a second time. People already called in and ordered Obama during the informercial. They sure aren't going to call in again now that they got the cheap piece of crap plastic made in China thing.

Obmama is now the Popeel Pocket Fisherman of politics.



JMS Tiger: THe bottom line is that Obama has gone through hard times, and his approval is down with many of his supporters, but the GOP also has huge unfavorables as well. I don't see where they have a clear message other than to "Obstruct Obama".

Democrats probably will spend a lot of money on these elections, but I believe that this year will have a better than average minority turnout to the polls. I think turnout will be better in some states than others, but I predict Democratic turnout will be greatest in the swing states and those with competitive races.

I predict that they will be out in droves to defend candidates like Murray and Feingold, and will be out in huge numbers in states like NV and KY with polarizing opponents.



As unpopular as the GOP is now, just imagine how universally hated they would be if they controlled congress!




"Obama's now at 49% approval in Rasmussen among likely voters."

I'd have to go back to look at comments, but aren't you one of those liberal rass-bashers?

These polls will swing back and forth a few points each week for a couple of months (do people understand what margin of error is?).

What they show us is the trend and the trend is that the democrats are and have been down by at least 6 points for the past 2 months. Go look at last week's gallup article.

Gallup shows a combined total of 91%. That leaves 9% undecided. They also said that their registered voter model runs 8 points too low for republicans. That would make it 53/46 republicans over democrats. If the election was right now, it would be 53/47 and republicans would pick up 55 seats.

The democrats don't have to convince undecided voters - because only 1% are undecided. They have to convince republican supporters to change their mind and vote for democrats instead. Convincing undecided voters is one thing - getting people to change their mind at this point will be nearly impossible.

Notice that I didn't say impossible. Stranger things have happened. But if you look back in history, mid-term elections almost never move more than a couple of points after the 4th of July. They aren't the same as presidential elections. People make their minds up earlier and stick with it. If there isn't significant movement for democrats in the next month, they are toast. If they can gain about 8 points, its a whole different story.



There is no correlation between generic ballot and seat pickups. None.



Gallup and the generic vote:
As you may be able to see from the table below pasted from Gallup, their likely vote, final poll prediction was pretty good in almost all years, including 1994.

Note that the LIKELY vote was accurate. As I recall from elsewhere, the adult and registered vote predictions sucked badly.

As is pointed out from time to time here, most of the difference in the generic vote between poor Rasmussen and others is that he uses likely all the time whereas they start after 1 September.

Note that a 48% Demo share of the actual vote on the average translates to 218 seats. It is not clear how much of that is innocent chance and how much is the result of good old-fashioned demo gerrymandering and, where races are very close, strategic ballot stuffing. I suspect the latter based on numerous clear cases from LBJ (Congress) to the last election.

Ballot stuffing has been strongly associated with big-city machines (e.g., Chicago's assist to Kennedy in 1960, King County's gift that kept on and on giving to Gregoire 6 years ago) and the rural south and west. You might have noted that every single Florida county with "problem" ballots in 2000 was controlled by democrats. It is possible that things have changed drastically in the last few years in the south (e.g., SC), but in the past those who controlled ballot counting in the South were overwhelmingly democrats.

Hopefully, despite recent experience in Minnesota by the same players who gave us King County, ballot-stuffing is becoming rarer outside of big cities (which will always be hopeless).

Gallup Poll Accuracy Record: Midterm Congressional Elections

Demo Repub DemAdv Err
% % %
2006 Likely voters 51 44 7
Actual 53 45 8 -1
2-party vote 54.1 45.9 8.2

2002 Likely voters 45 51 -6
Actual 45.9 50.5 -4.6 -1.4
2-party vote 47.6 52.4 -4.8

1998 Likely voters 49 45 4
Actual 47.8 48.4 -0.6 4.6
2-party vote 49.7 50.3 -0.6

1994 Likely voters 46.5 53.5 -7
Actual 45.5 52.4 -6.9 -0.1
2-party vote 46.5 53.5 -7

1990 Likely voters 54 46 8
Actual 52.9 44.9 8 0
2-party vote 54.1 45.9 8.2

1986 Likely voters -- -- --
Registered 52 40 12
Actual 54.6 44.5 10.1
2-party vote 55.1 44.9 10.2

1982 Likely voters 55 45 10

Final results based Actual 55.3 43.1 12.2 -2.2
on combined studies 2-party vote 56.2 43.8 12.4

1978 Likely voters 55 45 10

Final results based Actual 53.7 44.9 8.8 1.2
on combined studies 2-party vote 54.5 45.5 8.9

1974 Likely voters 60 40 20
Actual 57.6 40.6 17 3
2-party vote 58.7 41.3 17.3

1970 Likely Voters 53 47 6
Actual 53.4 45.1 8.3 -2.3
2-party vote 54.2 45.8 8.4

1966 Likely voters 52.5 47.5 5
Actual 50.9 48.3 2.6 2.4
2-party vote 51.3 48.7 2.6

1962 Likely voters 55.5 45.5 10
Actual 52.5 47.2 5.3 4.7
2-party vote 52.7 47.3 5.3

1958 Likely voters 57 43 14
Actual 56.2 43.4 12.8 1.2
2-party vote 56.4 43.6 12.9

1954 Likely voters 51.5 48.5 3
Actual 52.5 47 5.5 -2.5
2-party vote 52.8 47.2 5.5

1950 Likely voters 51 49 2
Actual 49 49 0 2
2-party vote 50 50 0



Your data is interesting, but it misses the point. Nobody cares about the actual split among votes for Congress from the two parties. It's only the decisions on individual seats that matter.

What difference does it make if 50% of the voters vote for Democrats nationally for Congress and 50% vote for Republicans? That would never mean that the change in seats would be 0.

In states like Utah and Wyoming, Republican candidates will win 75% of the votes, and in districts in African American areas in big cities, Democrats will win 75% of the votes.
So what have we proved?

Your data is totally irrelevant.



Just stating the numbers, hoosier. Take what you will from them. But 49/50 in Rasmussen is a +14 point improvement from two weeks ago, and his best numbers there in two months. Knowing them I doubt if it will last, but it's still a mistake to completely ignore them.



iVote: The numbers have literally ping-ponged up and down over the past few weeks in a way that I've not seen from Ras....ever perhaps.

I'm not sure what the cause is; perhaps he's been adjusting his LV screen or testing out new area codes (running out of fresh respondents?) It was -9 yesterday and 49-50 one day last week and then, as you said at -14 for a day two weeks ago I think. It's absurd to think that attitudes have shifted up and down so suddenly, and for no compelling reason. Indeed, Gallup's been pretty stable the past week or so, with support at 45 to 48%, generally.

Expect Obama's support to head back down south to 45-46% territory as well in the next few days (certainly not because I want it to).



From the data posted above, note:

In 1998, Democrats lost the generic vote, but gained 5 seats. In 1966, Democrats won the generic vote, but lost 47 seats. Obviously there are a number of factors, including distribution of vote, gerrymandering and quality of candidate that goes into play.



Unless I have somehow misunderstood it, your opinion is almost certainly wrong. As Gallup shows in another table, the percentage vote correlates reasonably well to the division in actual seats. Hence, the predicted split in votes nationally is vitally interesting.

The fact that some districts are lopsided affairs is interesting but only marginally important to the final number of seats won and lost. Liberals will win by lopsided amounts in cities and conservatives will win by lopsided amounts in rural areas. Usually it is close to a wash.

After all, 435 is a lot of contests, so statistically one would expect the binomial distribution to be closely approximated by the normal distribution. That is, the proportion of seats won should hew fairly closely to the percentage vote nationally, as it has in the past. This is very different from the electoral vote where a few states breaking one way or the other can inflate the electoral margin.

Also, if I understand your comments about a "tie", you appear to be confusing "change" and numbers of seats won. EVERY House seat is up for election. Thus, a "tie" percentage vote will not bring stasis, as you imply; it will bring a 218 to 218 tie in the number of seats held by each party, plus or minus 2. That is a change of 39-45!

Dear Nancy would not agree with your assessment at all if that eventuates.


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.