Articles and Analysis


POLL: Pew National Survey

Pew Research Center

Obama 49, Clinton 39 (was Obama 49, Clinton 40 in late February)

Obama 49, McCain 43 (was Obama 50, McCain 43)
Clinton 49, McCain 44 (was Clinton 50, McCain 45)

The survey also tested reactions to Rev. Wright, the economy, and the traits associated with each party and candidate.



The Pew Research results once again highlights just how far Rasmussen results seem tilted toward GOP candidates (or everyone else is tilted toward Democrats.) Frankly, I cannot believe this is likely to be due solely to their IVR-based polling techniques (especially since SurveyUSA uses the same technique and doesn't appear to have the same consistent tilt.)

I've been told that Rasmussen weights their results according to pre-determined quotas for partisan affiliation, a practice that I would think is, ahem, questionable at best.

In any event, the GOP tilt is so consistent, and lately so extreme, that I'd think some questions would be raised about their methods or results.


Mark Blumenthal:


Rasmussen does weight by party. I wrote up their procedure when they adopted it two years ago.

They regularly update their national party ID results here.



Hmmm, after all this, nothing has changed. Let's see how this looks next month.

It seems Obama's numbers were unaffected by the Wright nonsense. Let's see if Hillary comes out unscathed after the "Tall Tale of Tuzla".

What is really surprising is that Clinton's fund-raising hasn't completely dried up. Who is still donating to her? If any of you are, I got a great business venture for you too. I'm selling this toll-booth, brand new and everything - you just sit there and people give you money. It's a win-win.

Anyhoo, I really pity Hillary, Chelsea and poor Bill. It's so sad to see the whole family go at Barack and get beaten down like this. Yikes. Bill & Hillary just look tired and worn down. I feel for them. They tried everything including throwing the kitchen sink, dishwasher, hot water heater and even the sump pump at Obama, and he just deflected them away like he was dusting off his suit. He then straightened his tie, and kept on message, delivering a historic speech on race in the process. This spells bad news for Repubs in the fall. If this Wright mess didn't derail him, well then, I imagine nothing can. It also negates their "he's a muslim" smear tactic. Of course, with their base, it may still work. Half of them will believe he is a muslim and the other half that he's a black panther christian. Still, it won't be enough. Obama's message is just starting to spread. Not everyone follows the news. Once all these college kids really get involved - watch out.

There are rumors on the internet, that this whole wright thing was Hillary's secret weapon - and that she planned to unleash it a week or so before the Penn primary. If that is true, well then, what is left to do but concede?

Another unconfirmed theory is that the Obama campaign actually released the wright videos themselves, timing it perfectly to do the least amount of damage. I must say, it worked perfectly. Supporting this theory is the fact that Wright was on vacation during the most intense media coverage of it. If this is all true, then the type of political savvy the Obama camp has displayed is rarely seen - and must be feared by the GOP. For then you have the most unique of creatures - a new, fresh, inspirational phenomenon combined with the political acumen of a talented veteran politician.

Let the swift-boating commence, GOPers, for you will need an unending stream of such attacks to take Obama's movement down. Just ask Hillary.



My impression is that Pew is supposed to be very good and all. But their description of the survey results - "among 1503 adults... Obama 49-39" - seems shaky at best.

The "about the survey" section says 1503 total voters - presumably, the "Obama or Clinton" question was posed to only the 618 Dems/Dem-leaning RVs, and for this sub-group the MOE is 4.5%. Which would mean the 49-39 score is only just outside the MOE!

Am I missing something here?



Same question here as RS.... is this why Pew's results appear at least on the surface to be out of line with what Gallup and Rasmussen have (~ 45/45)?


Go to their full report, and look at page 20, the first on the general election. You'll see the incredible split by generations, where 18-25 year olds narrowly favor Clinton over McCain but go with Obama over McCain by 28 points.

That, to me, is the big overlooked variable of the campaign - and why I think there are large differences in polls.

I remember once figuring out that Rasmussen had the total Gen-X and Millenial vote at about 35%, but I can't tell if that's fixed in all of their tracking polls or not. Those two together will vote at a rate of 37-45% in this election. Pew didn't release crosstabs, at least not where I could find them, but they appear to have them much closer to the correct answer.

That's why you see so much variation from one poll to the next, from what I can tell. Some go the extra distance to include younger people, or at least adjust appropriately.

I couldn't find cross-tabs for this Pew survey (Grrrrr!) but from the data they have it appears they are at least close.



All this speculation would die if Pollster.com simply analyzed Rasmussen's methodology and told us if there's something wrong with it or not.




Rasmussen is apparently inflating the Republican identifiers, compared to other sources. At the Rasmussen web page you linked, Rasmussen shows only a 5.6% edge for Democrats among identifiers, whereas a Pew Research report last week showed a 9% edge for the Democrats, plus a 5% edge in self-reported "independents" who report leaning toward Ds (15%) vs Rs (10%). See:

The edge for Ds is 11% in the key swing states, with Florida being the only swing state that retains roughly equal strength between parties in partisan identifiers.

Rasmussen's inflated numbers for Republican identifiers are enough to explain the differences between Rasmussen and Gallup (if Gallup doesn't weight the responses in the same fashion as Rasmussen), although the Pew Research Center polling consistently shows the Democratic candidates substantially ahead of McCain, rather than neck and neck. I have no idea why Pew's figures consistently show the Democrats as stronger than the reports by Gallup and Rasmussen - other than house effects that have always led me to suspect Gallup tilts right for some reason.

Frankly, Pew's figures look more plausible to me among the national polls. And I don't think that I have this opinion because I've always been a yellow dog Democrat. How can a 71-year-old politician who says he wants to stay in Iraq for 100 years and who admits that he knows virtually nothing about political economy be polling even or ahead of both Democratic candidates? Two-thirds of the public think it was a mistake to invade Iraq in the first place and a plurality wants U.S. Troops OUT YESTERDAY! Plus, about 50% of the public are concerned about the economic welfare of their own households and almost 90% have been telling pollsters since the Iowa caucuses that they think the national economy is either in poor condition or fair condition? And Bush's approval ratings are still mired in the low 30%s!

Common sense says that a candidate who is explicitly running to head a "third Bush term" doesn't have a chance in this environment. So it should not be surprising that both democrats have had a 5%+ lead for at least the last month. What is somewhat surprising is that McCain is still winning support from more than 40% in the Pew polls.

The punditocracy uncritically accepts the poll numbers on the general election match-ups from Gallup and Rasmussen because the pundits all like McCain personally and they want to be able to say nice things about his campaign standing right up to November.

It will be interesting to see, after another seven months of polling and the actual election, whether Pew's numbers are fairly accurate, while Gallup and Rasmussen consistently exaggerate Republican strength.




I was drafting my own comment when yours posted. I suspect that you are correct that age weighting for historical turnout by age group is contributing more than anything else to the poor performance of many polls this primary season.

After the Wisconsin primary, I was impressed that PPP had the most accurate pre-election results. PPP explicitly reported that its estimates were based on turnout figures by age from the prior primaries this January and February, not from pre-2008 elections. Even then I think PPP was a little bit low on the estimated margin of victory for Obama.

PS I left out of my recent post a question: why does Rasmussen weight its poll reports using party identifier figures that significantly over estimate the number of Republicans? Is it simply inertia? Incompetence? Or is there an attempt by Rasmussen - conscious or subconscious - to make the GOP look more competitive? A more competitive GOP might be able to turn its voters out at higher rates for the general election, and it also might attract a few extra votes from undecided voters who want to vote for the winner, and therefore vote for the candidate who appears to be leading in the polls.

It's true, of course, that pollsters want to have a decent record for accuracy in tracking voter preferences leading up to an election. But as most readers of this blog know all too well, the actual performance of pollsters in tracking voter preferences seems to have little effect on the amount of polling business that the firms attract! So a pollster with even a little bit of a partisan agenda could be motivated to NOT change a model, even if it was performing poorly, as long as it made the pollster's preferred party appear to be more competitive.

I'm writing these thoughts purely in a hypothetical mode - I have no knowledge of the Rasmussen organization or anyone in a position of authority at Rasmussen who has any actual motive or intent to continue using an outdated model in order to exaggerate the support of one party at the expense of the other.



"Go to their full report, and look at page 20, the first on the general election. You'll see the incredible split by generations, where 18-25 year olds narrowly favor Clinton over McCain but go with Obama over McCain by 28 points.

That, to me, is the big overlooked variable of the campaign - and why I think there are large differences in polls."

Young people tend to become hardcore supporters for a specific party for life, if they vote that way the first few times. Obama could tilt the next generation to the Democrats in such a way that the Republicans will be the minority party for a decade, or more.




I agree - you may see Dem control for the next 20 years, or until they screw it up post-Obama. Remember, Clinton created many Republicans with his shenanigans. And so now Bush has created a generation of lifelong dems with his actions.


Thanks to hardheadedliberal (hey! that's me, too!) and the rest. What I like about this site is that it's a place to discuss demographics without the hard-core politics getting too heavy. Most of the time.

This is my thesis about the election:

First of all, I don't get into the traditional age categories like 18-25 year olds because each election has a different generation. I like to follow generations, and so I break it down as Millenials (born 1980-1999), Gen-X (1965-1979), Boomers (46-64) and Traditionals (earlier). I think each generation has different values and attitudes based on when they were born and what they experienced. If you click on my name, then select my blog, you'll see going back my "Generations" series of posts explaining some of this.

In 2004, the turnout overall was 58%, with turnout among Millenials 42%, Gen-X 50%, Boomers 64% and Traditionals 69%. Overall, Millenials made up 9% of all voters and Gen-X 24%, for a total of 33% born after the Voting Rights Act.

If these patterns hold, we will have 37% of the electorate born after 1965 this time. But - and here's the kick - if you have nothing more than Millenials voting in the same percentages as Gen-X, that number goes up to 40%. If both of these vote as often as Boomers, that number reaches 45% (and overall turnout climbs to 65%).

Why might this happen? It already has in several key primary states, including South Carolina and Wisconsin. Barack Obama talks like a Gen-Xer (he is, narrowly, a Boomer) and speaks to a different generation of Americans.

I'm not saying that this is an appropriate way to break things down because I have a theory. I'm saying it's already happening. Look at the Pew poll and its breakdown by age - Obama beats McCain by 28 points in the 18-25 (Millenial)! An error in the voting rate among the generations produces a decent error in the final tally when they are as skewed as they are by age.

I'd like for pollsters to take a hard look at this - not by age, per se, but by generation. There is every reason to believe that if the younger generations are enthused, there will be a massive generational change. The polls show it's at least a possibility. And I, born in 1965, could wind up being nearly the Median Voting Age.

If nothing else, I think that polls should be considered invalid if they do not include 17% Millenials and 23% Gen-Xers. I think that is a reasonable median to start with. And I don't care how hard we are to reach on the phone, either!



hardheadedliberal, the fact that Pew shows Dems with a 9% party ID advantage over Republicans is not by itself proof that Rasmussen is "inflating" Republican participation. Why not invert the actors and assume that Pew is "deflating" Republican voters?

Even if it were true that Rasmussen is inflating Republicans, the difference is very small to account for the difference between McCain's 10% lead in Rasmussen versus the Obama 6% lead in Pew. That is a 16% discrepancy, while the Party ID discrepancy between Rasmussen and Pew is only 3% (9% for Pew, 6% for Rasmussen, according to the numbers you gave us).




Your post is a little beyond me at the moment, but here's a question regarding age vs. generation...

When you look at voting performance by age, should the trend follow the generational catagory or the individuals? By which I mean (totally hypothetical) if 18 - 25's voted in the last two elections at 20%, does one assume that in the following election 18 to 25's will vote at 20%, or does the 20% keep pace with that cohort as they age (do you assume the 25's who vote in 2000 at 20% will be the 33's who vote in 2008 at 20%; or do you assume each successive group of 25's will adopt the previous group of 25's rate?) If I understood correctly using generational terms like Gen-X point to the voting behavior following the individual, but age cohort like 18 - 25 following the catagory. Is the former right and the latter wrong?

God I hope I didn't just embarrass myself.

This is neither here nor there, but per Obama talking like a Gen-Xer - no way! Speaking as one who is as utterly, tragically classic Gen-X as they come - god help me, I still listen to Gen-X - Obama is anything but that. His image is a blend of classic Boomer (or projection of Boomer fantasy -'fulfilling the promise of the 60's'; the Kennedy-MLKesque speechifying) and Millennial (the high gloss package; the celebrity/popularity = value/virtue sensibility). He's the perfect marriage of Boomer credulity and Millennial commercialism. Skepticism, irony and authenticity weren't even invited to the wedding. :-)



Rasmussen was very accurate in the 2004 presidential election. http://www.slate.com/id/2110860/ I wouldn't be too quick to rule them out.



Well, if you're a Gen-Xer like me, we can talk. :-)

I say that Obama is more of a Gen-Xer because I see his central theme being that all the old divisions and ways of looking at things are just BS. The fact that he hinges it on hope is indeed more Millenial, but he's too old to be counted in that group. His life experience is one of navigating two worlds, which I think is a Gen-X experience. I can see your point clearly, and it was my take at first. I decided there was more to him than that. No matter what, however, I hope you can see my point that Millenials are excited by him for some pretty good reasons.

Now, you asked if it's fair to look at youth as a block, assuming each generation just has other stuff on their minds when they are young, or if generations vote differently as they pass through that "phase". The conventional wisdom is that 18-25 year olds just aren't into politics, and interest climbs as people get older.

Well, the truth is that Boomers voted at rates fo about 50% when they were that age. Gen-Xers were more in the high 30s at best, reflecting our attitude towards "Teen Spirit" (had to have a Cobain ref in there). Millenials were already at 42% in the last election, even though they had little stake.

I'm arguing that Millenials will vote like their parents, the Boomers, when it looks like they have a reason to. If Obama is the nominee (and I think he is already) they will probably have that stake and vote closer to 50%.

So I'm arguing that conventional wisdom about kids is wrong, having been skewed by us Gen-Xers and our crappy attitude towards things (we can still talk, right?).

Grouping Gen-X and Millenials is not an obvious thing, but I do think that in a McCain/Obama race it's a lot easier to do. I had to explain why in my own blog over 4 posts running about 4000 words. Yeesh. The point is that our experience in the world has been very different because the world really did change in the late 1960s and we grew up in a very different place than the Boomers and Traditionals.

So, my point is this: Two things are converging to make this election a referendum on generational change - an age gap between candidates of 26 years and a new generation voting in large numbers for the first time. I think that understanding what the generations value and get excited about is the key to understanding how this will all go down.

I'd like to see the polls reflect that more. I see this change buried in their cross-tabs, when they release them, but I don't see anyone really getting into it. But to me, it's already shown itself to be the most important thing.

More than anything, I love having this discussion. People have already gotten on me for lumping Gen-Xers and Millenials into one group, which I admit is difficult. I hope that through discussion we can all come up with a stronger argument. I just think it's finally our time, but only if we embrace the Millenials and work with them (even if they seem naive and into celebrity worship laced with cardboard "values").



Check out the daily rass and gall. The polls, they are a changin'.



Eric - very interesting. I'll have to nip over to your blog to see more - I don't want to ask you to re-write stuff here that you've already put up over there.

Perhaps I'm utterly dense today, but you're saying the conventional wisdom is to treat 18-25 (for example) as a static category, assuming that the voting behavior is tied to being that age rather than the individuals who spend 7 years in that catagory.... but you prefer tagging the performance to the individuals who make up the catagory? Like - the Gen-X voters who turned out at 30% when they were 18-25 will continue to underperform the previous generation as they march ever closer to the grave?

And - there's no way a real Gen Xer would get on board with all this "Unity." Its a very thin line that separates "unity," "conformity," and "if you're not with us, you're against us." We're not "joiners." Agree?

True Gen-X is much more "take the dog, the pony, and the rest of your circus somewhere else. your shiny shiny words don't impress me. just tell me what you're going to do, and get on with it. Leave me the f. alone. And close the blinds on your way out, goddammit."




Thanks much for the citation about Rasmussen's weighting methodology. It was exactly the source data I was looking for.

I do recall that in 2006 I checked Rasmussen's last pre-election polls in 21 US Senate races and (as I recall) they overestimated the GOP percentage in 19 of 21 contests.

That didn't mean they called those races incorrectly, of course, but the consistent GOP bias in their polls seems irrefutable.

Coming from and old school (SRC at the University of Michigan) I find the idea of weighting political polls by partisan quotas to be appalling.



Yes, I am saying that a lot of people graph 18-25 year olds back to 1972 without caring that it's three whole generations that they're looking at, and I think this is just wrong. I don't know that we Gen-Xers will always "underperform" on turnout, but it wouldn't surprise me. Whatever. nevermind.

Do I sound more like I was born in 1965 now? :-)

Why am I into unity? Because as much as I hate conformity, I'm really tired of what we do around the world. I think it's worth a little conformity if we can gain a little freedom along the way.

So, if you have to, color me more "Tom Sawyer" than "Smells like Teen Spirit", perhaps more Run DMC than P-Funk All Stars. And I'll do anything, up to and including starting a revolution, if it means I never have to hear the Beach Boys again!

Whatever the case, our generation is moving into a position of some power that will last at least until the Millenials really take over. It's at least interesting, and it might even be really cool. Not kewl, but cool.




Cheers to that, Eric :-)

And thanks for the explanation - as it happens I asked a friend about this last night and she insisted that you track the age category (as you said, 18-25 going back to '72) rather than the generation as they age, but I think your approach makes more sense, intuitively.


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.