Nov 7th, 2008 at 12:12 pm
Analysis Done Right
People looking to examine demographic subgroups in the presidential election would do well to follow Kevin Drum’s lead and state the sub-group’s swing relative to the overall swing. A group like union members where Obama did the same as Kerry, or a group like gun owners where he did very slightly better, is a group that doesn’t really seem to have embraced Obama’s charms or message to any particular degree.
76 Responses to “Analysis Done Right”
Register
RSSLogin
Advertisement
Visit Our Affiliated Sites
Books By Matthew Yglesias
Topic CloudView topics sorted by popularity
Abortion
Afghanistan
Basketball
Blanche Lincoln
Budget
China
climate
Congress
Crime
DC
Demographics
Economics
Economy
Energy
Environment
EU
Finance
Football
Gender
Health Care
History
Inequality
International Law
International Relations
Iran
Israel
Media
Monetary Policy
National Security
NATO
NBA
Palestine
Political Reform
Public Health
Public Opinion
Race
Regulation
Russia
Sarah Palin
Sports
Stimulus
taxes
Technology
Trade
transportation
Featured
Subscribe to the Progress Report
Contact Matthew Yglesias
Use this form to contact blog author Matthew Yglesias.
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
November 7th, 2008 at 12:16 pm
Not sure that follows. Maybe Obama’s charms and message swayed most people. If that’s true, it would be fallacious to say the median person wasn’t swayed…
November 7th, 2008 at 12:18 pm
Are you kidding?
Of course, they embraced his charms and/or message.
To your average Reagan Democrat union member or gun owner, or probably a combination of the two, (goddammit we need a better term
)
B -
x +
Where B = black dude, and x = something else
For Obama just to maintain, let alone improve, Kerry’s numbers is an achievement. Clearly, he had some kind of personal appeal that transcended his identity as an African American.
Thus, x > or = B. I’d say that’s fucking huge.
November 7th, 2008 at 12:24 pm
First, in Matt’s defense his phrase “a group that doesn’t really seem to have embraced Obama’s charms or message to any particular degree” may be inartful, but I think what Matt is just trying to say is that there are some groups that responded positively to Obama, but no more so than average.
Second, I actually think Drum’s methodology is a little off. What you should really do is take the group in question and compare it to the national swing MINUS that group–in other words, the control group should not include the group you are comparing to the control. That’s not a big deal unless you are considering a very large group, but if so you may understate the magnitude of the difference by including that group in the control group.
November 7th, 2008 at 12:24 pm
Drum’s analytic methodology seems defective. The overall swing is just a kind of aggregate of the swings among particular groups, so there is no basis for discounting particular swings by comparing them to the total swing.
November 7th, 2008 at 12:26 pm
Re Kevin Drum’s comment “Gays and last-minute deciders are the only groups where Obama performed worse than Kerry ”
—————-
So Appalachia is largely populated by Homosexual procrastinators?
Hmmmm. Who knew?
November 7th, 2008 at 12:30 pm
So Appalachia is largely populated by Homosexual procrastinators?
They are clearly taking their time coming out of the closet.
November 7th, 2008 at 12:51 pm
“People looking to examine demographic subgroups in the presidential election would do well to follow Kevin Drum’s lead and state the sub-group’s swing relative to the overall swing.”
This is incredibly true.
——
Obama’s collapse in last minute voters means one of two things:
1) Obama got killed in the last week
2) The Bradley effect is incredibly real
I was arguing option #1 a couple of days before the election</a, but I think the reality is some mix of both options.
Also, the massive underperformance in Democratic House and Senate numbers vis-a-vis the projections would tend to support the idea the “Obama got killed in the last week” hypothesis is a strong part of the mix…
November 7th, 2008 at 12:56 pm
The negative swing by gays and lesbians is bizarre. Obama is against gay marriage, and didn’t promote other gay rights in any significant way in his campaign, but he’s much more gay-friendly than McCain.
November 7th, 2008 at 1:02 pm
incomes of more than 200,000 (+34)
this really suprised me. Were’nt these the people most likely to have their taxes raised? Maybe McCain election campaign should have used the higher taxes scare.
Yeah, I’m being sarcastic
November 7th, 2008 at 1:05 pm
Exactly right.
Kevin’s analysis shows that the group which swung in favor of Obama the most is the extremely rich. As I noted yesterday, this severely undercuts Matthew’s contention that the rich are the Republican “base”. The rich are now the Democratic base, together with the poor and lower middle class.
November 7th, 2008 at 1:08 pm
Re Led’s comment “They [Appalachian Gay Procrastinators ] are clearly taking their time coming out of the closet.”
———-
Hmmm. Ever see the movie “Deliverance”?
Dune dune a dune dune…
November 7th, 2008 at 1:10 pm
The negative swing by gays and lesbians is bizarre.
I don’t think so. Gays and lesbians were particularly against Bush in 2004 because of the Marriage Amendment thing, which was a novel idea and a big deal. By 2008, the novelty of that had worn off, so I’d think that gay opposition to Republicans reverted back to its historical norm.
November 7th, 2008 at 1:17 pm
Apparently some Yes on 8 ads used Obama’s image and quotes; perhaps that had something to do with it.
http://www.pamshouseblend.com/showDiary.do?diaryId=7928
http://laist.com/2008/11/04/despite_obama_saying_no_yes_on_8_co.php
November 7th, 2008 at 1:20 pm
Re Al’s comment “I’d think that gay opposition to Republicans reverted back to its historical norm.”
——-
Or maybe it was Joe Biden saying on national TV that he and Obama agreed with Sarah Palin that Gay Marriage shouldn’t be allowed.
Or it could be the well known “Fuck David Geffen and the horse he rode in on” effect.
November 7th, 2008 at 1:23 pm
Seriously, is Petey ever right about anything these day?
First, we now know beyond any question that there was no Bradley Effect, because the polls were in aggregrate right on the money. End of story.
Second, he seems to have forgotten Drum was comparing 2008 and 2004 numbers. Obama actually won same day deciders 50-45, and barely lost last week deciders 48-50. Kerry won same day deciders 52-45, and lost last week deciders 48-51. There is probably an interesting point to be made about the stability of these numbers, but it likely has far more to do with the nature of undecideds than anything in particular to do with this contest.
November 7th, 2008 at 1:42 pm
DTM,
First, we now know beyond any question that there was no Bradley Effect, because the polls were in aggregrate right on the money. End of story.
Er, no. The RCP average for the period immediately prior to the election had Obama up 7.6%. The lastest popular vote tally has Obama up by 6.4%. There may have been a Bradley Effect of 1 percentage point or more.
November 7th, 2008 at 1:57 pm
I weighed myself on two different scales before and after the election. There was a 1.2% difference, most likely due to the Bradley effect.
November 7th, 2008 at 1:57 pm
Or maybe it was Joe Biden saying on national TV that he and Obama agreed with Sarah Palin that Gay Marriage shouldn’t be allowed.
But as I recall, John Kerry also said that he opposed gay marriage. So in that respect, Obama and Biden were no different than Kerry.
The relevant difference, as far as I can tell, is the gay marriage constitutional amendment, which Bush came out in favor of in 2004. To my mind, that was an enormous factor in the gay opposition to Bush in 2004 (although, to be fair, I could be overestimating the effect, given that in 2004 I lived in the most heavily gay area in the country east of San Fransisco).
McCain voted against the amendment, which was old news by 2008 anyway. And so I think it pretty logical that gay opposition to McCain in 2008 should be somewhat less than gay opposition to Bush in 2004.
November 7th, 2008 at 1:58 pm
Mixner,
No one who is serious about polling uses the national polls for this purpose, nor do they use RCP for their aggregations. Rather, the accepted benchmark is Pollster’s state by state aggregations, and as Pollster wrote:
[V]otes are still being counted in some states, so the numbers in the table may still change, but one thing seems unambiguous: There was no “Bradley effect” yesterday — no hidden McCain vote lurking among the undecided. In the states that were polled most heavily, the trend estimates came remarkably close to the actual result. The undecided vote did not appear to “break” decisively toward either candidate.
Case closed.
November 7th, 2008 at 2:01 pm
I would think numbers on “last minute deciders” are inherently suspect, because that’s the kind of self-analysis that is going to be both inaccurate and imprecise.
As for the Bradley effect, Nate’s numbers — a far more sophisticated analysis of the polls than anything at RCP — pegged the popular vote almost to a tenth, with (if anything) Obama outperforming Nate’s numbers. You’re gonna have to come up with more solid evidence than the RCP numbers to prove a Bradley effect.
Finally, on the LGBT numbers. Let’s remember it went from 77% for Kerry to 70% for Obama. Even assuming that’s statistically significant (and with only 4% of the exit poll sample, I’m not sure it is), these numbers are still higher than almost any other demographic (African-Americans being significantly higher; Jews slightly so). The idea that these numbers represent some sort of problem for Obama with LGBT voters is kind of silly.
November 7th, 2008 at 2:13 pm
DTM,
No one who is serious about polling uses the national polls for this purpose, nor do they use RCP for their aggregations. Rather, the accepted benchmark is Pollster’s state by state aggregations,
Er, first, pollster.com is not “the accepted benchmark” and second, pollster’s results show an even greater disparity than RCP’s. Pollster found a final poll margin for Obama of +7.7% vs. a vote return of +6.1%. Meaning a possible Bradley Effect of 1.6 percentage points.
And the comment you quote mispresents the Bradley Effect. It’s not just false reporting by people who tell pollsters they are undecided, but false reporting by people who tell pollsters they are likely to vote for the black candidate.
November 7th, 2008 at 2:18 pm
I don’t think so. Gays and lesbians were particularly against Bush in 2004 because of the Marriage Amendment thing, which was a novel idea and a big deal. By 2008, the novelty of that had worn off, so I’d think that gay opposition to Republicans reverted back to its historical norm.
This is a very good point Al.
However, we should remember that Senator Clinton was conspicuously to the left of Obama on LGBT issues during the primaries (if only symbolically). There was also the impression by some in the LGBT community that Obama was tolerated homophobia in order to win over African Americans and the religious during the primaries.
November 7th, 2008 at 2:22 pm
I think it’s clear from a variety of data sources that Democrats lost the last week of the campaign for both President and Congress.
The question is why.
I don’t think you can rule out either Bradley Effect or a lousy last week of the campaign, but given that I thought the last week was going very badly in real-time, I’m inclined to lean towards that as the primary explanation rather than the Bradley Effect.
November 7th, 2008 at 2:22 pm
Er, first, pollster.com is not “the accepted benchmark” and second, pollster’s results show an even greater disparity than RCP’s. Pollster found a final poll margin for Obama of +7.7% vs. a vote return of +6.1%. Meaning a possible Bradley Effect of 1.6 percentage points.
Most polls have sampling errors well in excess of 2%. A difference of 1.6% isn’t statistically significant. Furhtermore, as Nate Silver at Fivethirtyeight.com has noted, candidates with significant leads in the polls often see slight tightening come election day. Bill Clinton experienced the same thing in 1996. As was noted up thread, Silver’s more sophisticated modeling which DID NOT consider a Bradley Effect got the popular vote right to with in 0.5%.
November 7th, 2008 at 2:26 pm
The idea that these numbers represent some sort of problem for Obama with LGBT voters is kind of silly.
Whether it’s a “problem” for Obama remains to be seen, but the loss of support by gays and lesbians as compared with their support for the last Democratic candidate is striking.
It may have something to do with the Marriage Amendment issue. Then again, it may be that gays and lesbians are becoming less willing to support Democratic candidates who are only nominally or marginally more supportive of gay rights and equality than their Republican rivals.
November 7th, 2008 at 2:27 pm
I think it’s clear from a variety of data sources that Democrats lost the last week of the campaign for both President and Congress.
It doesn’t seem clear to me. Could you link to some of those sources please?
November 7th, 2008 at 2:32 pm
Willie, it’s “clear” to Petey because he’s decided it’s true. Of course, to most rational observers that fact alone suggests that the opposite must be true.
November 7th, 2008 at 2:35 pm
“It doesn’t seem clear to me.”
- Performed below the lower end of the range in everybody’s projections in House and Senate.
- Lost late deciders as shown by exit polls.
- Performed under national polls at the top of the ticket.
I really don’t think it’s controversial that we lost the last week of the campaign. The interesting question, again, is why.
November 7th, 2008 at 2:38 pm
Most polls have sampling errors well in excess of 2%. A difference of 1.6% isn’t statistically significant.
Yes, that’s why the results I reported are based on averages of many polls taken over several days.
As was noted up thread, Silver’s more sophisticated modeling which DID NOT consider a Bradley Effect got the popular vote right to with in 0.5%.
So, a single blogger and amateur pollster using a home-brew analysis got lucky in one election (and not so lucky on the electoral college vote, either).
November 7th, 2008 at 2:44 pm
So, a single blogger and amateur pollster using a home-brew analysis got lucky in one election
Heh. How many elections do you use in concluding that there might have been a Bradley Effect?
(and not so lucky on the electoral college vote, either).
He got every state except for Indiana (and possibly one of 4 electoral votes in Nebraska) right.
November 7th, 2008 at 2:51 pm
However, we should remember that Senator Clinton was conspicuously to the left of Obama on LGBT issues during the primaries (if only symbolically). There was also the impression by some in the LGBT community that Obama was tolerated homophobia in order to win over African Americans and the religious during the primaries.
Interesting. If correct, that’s a second point of carryover from the primaries to the general (Matthew noted the carryover in Appalachia yesterday).
November 7th, 2008 at 2:52 pm
Oh ho, that’s rich, Mixner getting all condescending about Nate Silver. On his electoral college prediction, he basically only miscalled Indiana, which he had a slight McCain favorite (1.5%) instead of the actual result, Obama by less than 1%. Yeah, man, he really sucked.
November 7th, 2008 at 2:52 pm
“is a group that doesn’t really seem to have embraced Obama’s charms or message to any particular degree”
I think that the last couple of months of comments suggesting with the existing political and economic fundamentals, a Democratic candidate ought to have been polling better were correct. Ergo, a lot of Anglos did have a problem supporting an AA candidate.
But, the AA candidate was able to pull together a winning combo. GWB had zero mandate coming into his first term, and yet he – like any other – approached the job like he was a 100% President, not a 50.01%. When you win, you win. Whether to act like a punani afterward is a personal decision.
November 7th, 2008 at 2:54 pm
Lost late deciders as shown by exit polls.
The exit polls Linked in Drum’s post show that of the 10% of voters who decided in the last 7 days of the election:
Obama won 48.5% to McCain’s 48.6%.
This doesn’t seem like a significant difference to me.
- Performed under national polls at the top of the ticket.
Sigh!
November 7th, 2008 at 3:12 pm
If one were to look at the national polls, then Nate Silver’s approach would easily be the best take available.
But again, if you are serious about polling then you aren’t going to look at the national polls, but rather the state by state results. That is true if for no other reason than that the combined sample size of the state polls dwarfs the combined sample size of the national polls. It also gives you a much wider range of sources and methodologies, including pollsters who specialize in certain states and know how to get representative samples.
But Mixner isn’t a serious person, so it doesn’t surprise me he isn’t serious about polling.
November 7th, 2008 at 3:15 pm
cmholm,
Actually, a group of political scientists compiled twelve different models predicting the two-party vote share using various factors. The final count is still ongoing, but it looks like Obama outperformed seven or eight of those predictions.
November 7th, 2008 at 3:33 pm
WillieStyle,
Heh. How many elections do you use in concluding that there might have been a Bradley Effect?
Heh. Exactly as many as I need to use to support that claim: one. You do realize that you need more evidence to support a claim of the superiority of a poll analysis methodology than to support a claim of a possible factor in a single election, don’t you?
DTM,
But again, if you are serious about polling then you aren’t going to look at the national polls, but rather the state by state results.
Typical DTM nonsense. State by state polls (or results) and national polls are both “serious.” State polls are more useful for some types of analysis and national polls for others. If you think you have a, er, serious argument that state polls are more useful than national polls with respect to the Bradley Effect, then make it. If you think you have a serious argument that state polls suggest a very different conclusion about the Bradley effect in this election than national polls, then make it.
November 7th, 2008 at 3:37 pm
“but given that I thought the last week was going very badly in real-time”
You just failed to tell anyone. You are a dick for many reasons, Petey, but your post-hoc “I Told You so’s” top the list.
Petey, let us remember, is the dipshit that predicted electoral doom due to the Palin Effect.
Oh, and how many more electoral votes would Hillary Clinton have won than Barack Obama in 2008, Petey? Or, better yet, John Edwards?
November 7th, 2008 at 3:40 pm
Mixner,
I already explained to you why state polls are vastly superior in the aggregate to national polls for this purpose (”[T]he combined sample size of the state polls dwarfs the combined sample size of the national polls. [Using state polls] also gives you a much wider range of sources and methodologies, including pollsters who specialize in certain states and know how to get representative samples.”). And I already linked you to Pollster for an analysis of what the state polls show.
But as usual, we have reached the point where you are complaining about people not meeting a burden you couldn’t possibly meet yourself, so we are done here.
November 7th, 2008 at 4:38 pm
DTM,
I already explained to you why state polls are vastly superior in the aggregate to national polls for this purpose (”[T]he combined sample size of the state polls dwarfs the combined sample size of the national polls. [Using state polls] also gives you a much wider range of sources and methodologies, including pollsters who specialize in certain states and know how to get representative samples.”).
You need to produce evidence to support your claim about sample sizes. Then you need to provide an argument to support the claim that the differences in these sizes, if any, is large enough have a significant effect on the analysis. Then you have to do the same thing with respect to your claim regarding “sources and methodologies.” Since you tell us you have abandoned faith-based analysis and wishful thinking and are now an “empiricist,” you shouldn’t have any trouble producing the empirical evidence and rational argument to support your disputed claims.
And I already linked you to Pollster for an analysis of what the state polls show.
You have provided no links to any analysis of anything by anyone.
November 7th, 2008 at 6:53 pm
Fuck this, I have the data on the actual margin in every state (based on votes included in the count as reported by the NYTimes at 7PM on 11/6) as compared to 538’s projected margin for every state (ignoring 538’s stated margin of error for their estimates). Here. Identify the Bradley effect.
November 7th, 2008 at 7:19 pm
Just to help, the first you do is say D.C. doesn’t count, which means that Obama only slightly over-performed 538 projections rather than massively over-performed, and then say that while it might be true that there was no national Bradley effect (since Obama overperformed his polling in more states than he under-performed) it might nevertheless be the case that some states showed Bradley effects (even though an equally good explanation of this is that polling errors are nearly randomly distributed). So here are the states where Obama under-performed 538’s final projection by by 2% or more, from a better presentation of the same data referenced above:
Tennessee
Delaware
Mississippi
Iowa
Kentucky
Utah
West Virginia
Arizona
Idaho
Oklahoma
North Dakota
Wyoming
Louisiana
Arkansas
Alaska
November 7th, 2008 at 7:21 pm
“You just failed to tell anyone.”
I actually provided a link upthread to me telling folks. You’ve never been particularly good at reading comprehension, brewmn…
November 7th, 2008 at 7:38 pm
By the way, I’m fine with Republican diehards insisting the Bradley Effect is alive and well, if that means in 2012 they once again fall for something like the Pennsylvania rope-a-dope Obama and Rendell ran on them this time.
November 7th, 2008 at 7:46 pm
Identify the Bradley effect.
The Bradley Effect may have been present in every state where Obama underperformed. In fact, the disproportionate number of southern states in the list of underperformers suggests that the effect was real.
November 7th, 2008 at 8:00 pm
Those states were also polled less often than the states 538 got right, but I’m sure that doesn’t matter.
November 7th, 2008 at 8:12 pm
Incidentally, although I am not the first to observe this, there does appear to be a reasonably strong correlation between the two-party share and how likely the polls were to understate McCain or Obama’s support respectively (you can see this by comparing the “Obama Margin” column to the “Difference” column in the spreadsheet above). So one possibility is that we are picking up some sort of localized bandwagon effect (assuming people can perceive in some fashion where their neighbors are leaning, say through newspaper polls, casual conversation, etc.). Anyway, the good news for Nate is this seems like a pretty easy thing to factor into his model.
November 7th, 2008 at 8:17 pm
So one possibility …
…. says DTM, as he continues to backpedal furiously from his earlier “We now know beyond any question that there was no Bradley Effect” nonsense.
November 7th, 2008 at 8:22 pm
Mixner,
I am not backpedaling at all. Mark Blumenthal remains correct when he said:
[V]otes are still being counted in some states, so the numbers in the table may still change, but one thing seems unambiguous: There was no “Bradley effect” yesterday — no hidden McCain vote lurking among the undecided. In the states that were polled most heavily, the trend estimates came remarkably close to the actual result. The undecided vote did not appear to “break” decisively toward either candidate.
Now you may think that the fact that in some states McCain did better than he did in the polls proves there was a Bradley Effect in just those states, but that is because you have no idea what you are talking about. There is an overall pattern, however, which I was supplying a plausible explanation for.
November 7th, 2008 at 8:41 pm
I am not backpedaling at all.
Right, I just imagined you replacing “case closed” with “one possibility….”
Not only have you utterly failed to substantiate your assertion that no part of the disparities between poll data and election results were caused by the Bradley Effect, you haven’t even produced any evidence that merely suggests the Bradley Effect was not present.
And no, quoting a statement by Mark Blumenthal a third time won’t help your case any more than it did the first two times.
November 7th, 2008 at 8:45 pm
Mixner, you don’t have evidence just because the polls were off, and in some cases the polls were off in a way which favored McCain. If you think that counts as evidence, I don’t think we can talk about this.
Among other things which would be true if there were a Bradley effect, it’d be the case that the polls in this election showed the African-American candidate over-performing his actual results by more than the polls in other elections have shown non African-American candidates over-performing his results. Show that, and it’d be worth considering whether it was a Bradley effect which explained it. As it is, there’s no phenomenon which we even need the effect to explain.
November 7th, 2008 at 9:01 pm
washer,
Mixner, you don’t have evidence just because the polls were off, and in some cases the polls were off in a way which favored McCain. If you think that counts as evidence, I don’t think we can talk about this.
In any result in which Obama underperformed a poll, the underperformance may be due, partly or wholly, to the Bradley Effect. If you seriously don’t realize this, I don’t think we can talk about this.
If we plausibly assume that white racism is more common in southern states than in northern ones, the disproportionate number of southern states in the list of underperformers suggests that the Bradley Effect may have been present in those states.
Among other things which would be true if there were a Bradley effect, it’d be the case that the polls in this election showed the African-American candidate over-performing his actual results by more than the polls in other elections have shown non African-American candidates over-performing his results.
I’m not sure I understand what you’re trying to say in this torturous sentence, but if I’m understanding it correctly, no, that is not correct.
November 7th, 2008 at 10:01 pm
I know Mixner won’t understand this, but for the record:
The Bradley Effect can’t explain Obama actually doing better than the polls in many states, and indeed overperforming the polls on average. Nor can it explain why there was a fairly strong correlation between the two-party share in a state and whether the polls underrepresented Obama’s support or rather McCain’s support.
What I did above is suggest one possible explanation for the actual observed pattern, the very same pattern that is inconsistent with a Bradley effect. But since that possible explanation has nothing to do with the Bradley Effect, and indeed seeks to explain a pattern which is inconsistent with the Bradley effect, of course I am not suggesting a Bradley Effect occurred.
Again, though, since Mixner is not very bright and doesn’t understand the basic concepts involved in this issue, he won’t understand any of what I just wrote. Nonetheless, he probably won’t be able to stop himself from demonstrating his lack of comprehension in a response.
November 7th, 2008 at 10:17 pm
Linkman: Do we have the swing at Luton?
Gerald: Well, I’ve worked out the swing, but it’s a secret.
Linkman: Er, well, ah, there…there isn’t the swing… how about the swong?
Norman: Well, I’ve got the swong here in this box and it’s looking fine. I can see through the breathing holes that it’s eating up peanuts at the rate of knots.
Linkman: And how about the swang?
Alphonse: Well, it’s 29% up over six hundred feet, but it’s a little bit soft around the edges about…
Linkman: What do you make of the nylon-dot cardigan and plastic mule rest?
Offscreen Voice: There’s no such thing.
Linkman: Thank you, Spike.
Norman: Can I just come in here and say that the swong has choked itself to death.
George: Well, the election’s really beginning to heat up now.
Eric: I can’t add anything to that.
November 7th, 2008 at 10:21 pm
DTM,
The Bradley Effect can’t explain Obama actually doing better than the polls in many states,
No one said it could. It can however explain Obama underperforming the polls in many states and nationally.
and indeed overperforming the polls on average.
He didn’t overperform the polls on average. Even your own preferred source of polling analysis found that Obama underperformed the polls on average.
Nor can it explain why there was a fairly strong correlation between the two-party share in a state and whether the polls underrepresented Obama’s support or rather McCain’s support.
Yet another utterly irrelevant (and pretty incoherent) claim. The Bradley Effect doesn’t need to explain that.
What I did above is suggest one possible explanation for the actual observed pattern, the very same pattern that is inconsistent with a Bradley effect.
No what you did was to make the utterly ludicrous claim that “we now know beyond any question that there was no Bradley Effect.” Seeming to recognize the absurdity of that proposition, you’ve been running away from it ever since.
November 7th, 2008 at 10:32 pm
As I predicted, Mixner couldn’t stop himself.
November 7th, 2008 at 10:45 pm
As I predicted, DTM tries deploying a smokescreen to divert attention from his absurd claim.
November 8th, 2008 at 10:52 am
In any result in which Obama underperformed a poll, the underperformance may be due, partly or wholly, to the Bradley Effect. If you seriously don’t realize this, I don’t think we can talk about this.
In any result in which Obama underperformed a poll, the underperformance may be due, partly or wholly, to orbital mind control lasers operated by the Illuminati. If you seriously don’t realize this, I don’t think we can talk about this.
November 8th, 2008 at 12:59 pm
I still find it hilarious that misanthropes like Petey are claiming a 6.5%, 100-electoral vote victory for Obama somehow validates their over-the-top demonization of the man over the last year.
And I still defy them to plausibly make a case that there is a single Democratic politician in America who could have achieved a more convincing win on election Day.
November 8th, 2008 at 8:58 pm
Re: However, we should remember that Senator Clinton was conspicuously to the left of Obama on LGBT issues during the primaries (if only symbolically).
A lot of gay guys I know were solid Hillary supporters and were very cool toward Obama as a result. Maybe gays are the true PUMAs? In fact, I was initially cool toward Obama, but never once seriously considered voting for McCain. My partner however never warmed to Obama one bit, and though he too voted Dem, he was mainly voting against Sarah Palin not voting for Obama.
Re: it may be that gays and lesbians are becoming less willing to support Democratic candidates who are only nominally or marginally more supportive of gay rights and equality than their Republican rivals.
Once upon a time, John McCain mused that a gay man could become president and openly dissed the Religious Right. But that was a long, long time ago. This year he toed the SoCon line quite rigidly. Add Sarah Palin to the mix and it’s absurd to claim that there was little difference between the two tickets on LGBT issues. And there are others issues we care about than just marriage you know.
March 11th, 2009 at 4:52 am
If you have to do it, you might as well do it right
March 12th, 2009 at 11:49 pm
I bookmarked this site. Thank you for good job!
March 14th, 2009 at 5:37 am
It is the coolest site,keep so!
xanax
March 22nd, 2009 at 6:23 am
tramadol
Incredible site!
April 3rd, 2009 at 4:18 am
Great site. Good info
cheap brand pfizer viagra
April 9th, 2009 at 6:49 am
Great site, Good info viagra