Kos's comments and my reply.
Quick summary: No, I don't think exit polls are necessarily better and, in any case, I don't yet have the raw exit poll data. But I did have the raw data from the Pew polls, so that what I used. Pew polls are good. But Kos did get me on New Hampshire. As we continue to analyze further data, I don't think our maps will change by much but they will certainly change in some details (notably, whites in N.H.). Also, now I'm convinced that it's better to use some range of shading to indicate the close calls, so that people don't think I'm expressing 100% confidence in the color of each state.
3.26.2009
Pre-election Polls and Exit Polls
@ 1:14 PM
Subscribe to:
Post Comments (Atom)
25 comments
Kos also showed NM 49-48 McCain?
That might be exit poll data but it is also not the way we voted.
FWIW
Kos.....shut up. Please try to know what your talking about before you post. Andrew, thanks for the analysis, it is all very interesting. Analysis is partly a matter of interpretation, and I believe the readers of this site understand that, and know how to look at these maps. Keep up the good work.
To me, more fascinating than race is the issue of wealth.
On that note, it would really be useful to somehow illustrate the SIZE of the pockets of wealth/poverty in each location, that is, the number of wealthy in California versus Arkansas. As it is, it looks from the graph like wealthy white folks are HUGELY against President Obama. I very much doubt that is true.
Trying to find a simple parameter to normalize data is perilous in this case.
The 04 elections are explained by ancestry, urban living, age and education.
Wealth is a misleading category because wealth changes hands.
So, let me get this right: the problem is that you didn't have raw data yet?
Why isn't the problem that you jumped the gun, had an assertion you wanted to make, but didn't have the data to back it up, so you knowingly used data that was insufficient to support your claims, and you admitted that you knew it was insufficient to support your claims only after you were called on it?
If you didn't have the data, you should have waited to make the assertions rather than use faulty data that you didn't admit up front was poor or inadequate. This smacks of "oops, I got caught, so I better have something to say."
Your unstated but inescapable implication, btw, is that you also don't have legitimate data for your previous assertions about Appalachia vs the Deep South.
Sheesh.
Your numbers for Washington state also look fishy. The CNN exit poll has Obama carrying Washington 57-41 among voters with incomes between 75K and 150K. You have these voters going for McCain. Given that Washington as a whole voted 58-40 for Obama I know which one I would guess is more accurate.
Another case of Andrew cherry picking data to support some pre-determined assumption (see his "In defense of Appalachia" post).
This kind of thinking is why I don't like many Republicans. They decide something, such as "Global Climate Change is a Hoax", and then find data which supports their assumptions. This simply is not the right way to do things.
If this is what FiveThirtyEight is now about, I will no longer read this site. It appears that, for the moment, it's only about what Andrew is about. I will therefore not read his attempts at using polls and statistics to mislead people. It's a shame that his crap has to clutter up the great analysis and commentary by Nate.
@ Chris Hoffman and others...
Exit polls are laughably inaccurate. If they were accurate Bush would not have been elected in either '04 or '00.
Meanwhile as a blogger posted on Kos (though I've not been to CT and don't know if I agree):
Actually, I think that a good case exists for Obama losing the <$20 group in CT. In CT, I suspect that most white male voters with that level of income are retired. Further, CT also has a number of mid-old and old-old (which makes them members of the "Greatest Generation" and WW2 veterans) voters still living in it. That group would have had a very difficult time voting for Obama. Throw in the situation that younger white males making <$20K also tend to have unsettled living situation (read: not registered to vote) and the electorate becomes much more interesting. Since retirees make up a pretty decent slice of the CT electorate, a very good case could be developed that the <$20K electorate selected McCain, a fellow veteran.
It is hard to believe Gelman would say this:
> No, I don't think exit polls
> are necessarily better
That invites the retort,
"Well, who cares what YOUR ignorant ass thinks?"
EVERYBODY knows that exit polls are THE MOST reliable tool we have.
This is not to say they're "always" right, but we did base
the overthrow of an election in Ukraine on one. They had historically been CONFIRMED right at least until Bush/Kerry 2004.
People opining about what they think " a good case exists for" are
being ridiculous. Every last precinct total is available.
There is no "case" when THE FACTS ARE AVAILABLE.
Must Consider the parameter too..|
I appreciate having a site where this kind of discussion actually takes place.
totally off topic, but did anyone read the Republican response to the President's budget? It contains absolutely no detail and has tidbits of wisdom like:
"Instead of returning to double digit levels of inflation and the failed economic policies of the 1970s, Republicans support maintaining the cost of living after witnessing the booms and busts triggered by loose monetary policy."
um, ok, how? This is the lamest thing I've ever seen. It's like they are saying, "you can't say we don't have anything, because here's this 18 page book that is the precursor to the ideas we have. It's a teaser, and it means no more no! We have all the answers, and we'll share them, just not till next week."
http://www.gop.gov/solutions/budget/road-to-recovery-final
in case anyone is interested. Don't worry, it won't take long.
This is amazing to me. Gelman is an astoundingly credible academic who has drawn conclusions based on available data. No data is perfect, but as an academic, Gelman has done what any respectable researcher would do. He gave his sources, his methods and analysis based on the data. Does that mean he is claiming some biblical truth? No. He is saying what the evidence says. Is there other data? Of course. Is it better? In some parts yes, in others no. The point is Gelman has put himself out there to exchange ideas. He cannot think of everything, which is why the dialogue is important. To attack him as Kos, Houle and others have has is exceedingly lame.
I'd like to second Stephen A. Nuno's post. If you think Pew Polls are worse than Exit Polls (which again had Kerry destroying Bush in 2004...), that's fine. But you can't accuse Mr. Gelman of intellectual dishonesty for using a *credible* polling source (Pew) and stating quite clearly what his source was (i.e. he didn't hide it or claim it was some other source).
Intellectual dishonesty is using an inherently biased source of data and/or hiding your source. Mr. Gellman did neither.
Attack the data and methods, not the man. No need for ad hominem attacks when the author is kind enough to have a comments section for you to directly post *constructive* criticism.
Considering how many of us Dems got burned by exit polls in 2004, I'm surprised there's so many fanatical fans here...
I already did a major analysis of this, which is in my post-election analysis at google docs. Look at point "E:" This is a comparison of the average of the last 7 polls to the actual results from election night, down to 1/100th of a percentage point.
It can be that specific pollsters were way off in some states and more on in others, but the average came very, very close - closer in this cycle than in 2004 and 2000.
Of course sometimes exit polls are lousy just like sometimes pre-election polls are lousy. Remember the 2008 New Hampshire Democratic primary. Sometimes multilevel logistic regression fits by esteemed academics are lousy as well. (I haven't looked at these closely enough to know how good this model is.)
The results from my current home in Washington state look way out of touch with reality. (My old home of Washington DC doesn't even get a mention.)
But it is unfortunate that Dr. Gelman choose to only present graphs and didn't link to the estimated percentages that he calculated. It also seems strange that he didn't (as far as I can tell) use the final vote totals from the election. While that is not broken down by race or income it seems like it would be useful. Hopefully with some tweaks the next iteration will be better.
The exit polls were right- G T Bush won.
@JFisher: "Exit polls are laughably inaccurate. If they were accurate Bush would not have been elected in either '04 or '00."
Bush WASN'T elected in '00. And I am agnostic as to whether Bush was elected in '04 either. I have seen arguments both ways, whether the discrepancies between the exit polls and the reported totals were errors in the polls or dishonesties in the tabulations; it should have been looked into officially, not left to stats professors on the intertubes.
Khalil: Attack the data and methods, not the man. No need for ad hominem attacks when the author is kind enough to have a comments section for you to directly post *constructive* criticism.
Ah, but then the World Wide Web was created with the notion of providing a low-cost platform for such chickenshit antics in mind --or so, after spending fifteen years observing what passes for discourse on it, I must conclude.
A thoroughly nice sentiment all the same, heh.
EVERYBODY knows that all caps are THE MOST reliable tool we have for communicating THE TRUTH!
酒店喝酒,禮服店,酒店小姐,酒店經紀,制服店,便服店,鋼琴酒吧,兼差,酒店兼差,酒店打工,伴唱小姐,暑假打工,酒店上班,日式酒店,舞廳,ktv酒店,酒店,酒店公關,酒店小姐,理容院,日領,龍亨,學生兼差,酒店兼差,酒店上班,酒店打工,禮服酒店,禮服店,酒店小姐,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,台北酒店,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,禮服店 ,酒店小姐,酒店經紀,酒店兼差,寒暑假打工,酒店小姐,經紀 彩色爆米花,經紀人 彩色爆米花,酒店傳播,酒店經紀 彩色爆米花,爆米花,童裝,童裝拍賣,童裝大盤,童裝寄賣,童裝批貨,酒店,酒店,童裝切貨,酒店,GAP童裝,酒店,酒店 ,禮服店 , 酒店小姐,酒店經紀,酒店兼差,寒暑假打工,招待所,酒店小姐,酒店兼差,寒暑假打工,酒店上班,暑假打工,酒店公關,酒店兼職,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,
ugg shoes
ugg boots
ugg
ugg sale
fake ugg
ugg store
ugg discount
men's ugg
womens ugg
buy ugg boots
ugg cardy
replica ugg
cheap ugg
ugg classic
Ugg Amelie Suede
UGG Bailey Button
UGG Classic Cardy
UGG Classic Crochet
UGG Classic Mini
UGG Classic Short
UGG Classic Tall
Ugg Gypsy Sandal
Ugg Halendi Sandal
UGG Hammond Slipper
UGG Infant's Erin Baby
UGG Layback Slipper
UGG Locarno
Ugg Matala Sandal
Ugg Napoule Sandal
UGG New style
UGG Nightfall
Ugg Persephone Sandal
Ugg Skimmer
UGG Sundance
Ugg Tasmina
UGG Ultra Short
UGG Ultra Tall
UGG Women's Coquette
UGG Women's Knightsbridg
UGG Women's Mayfaire
Your blog is wonderful, I like it very much, thank you!
By the way, do you like Burberry Polos, which are very chic, especially the Ralph Lauren Polo Shirts, I love them very much. I also like playing sports, it can keep healthy, what do you like to do?
We are the outlet of Wholesale Polo Shirt, Polo Ralph Lauren shirts on sale,these products are best-seller in our store cheap polo shirts Cheap Polo Shirtsour products also have a corresponding discount activitiesWholesale Polo Shirts -50% OFF,cheap polo shirts Cheap Polo ShirtsSo if you love these, you should not miss our store, we can meet what you want, and you can find many surprise in our store
araç sorgulama
sorgulama
ehliyet
açıköğretim
bağkur sorgulama
ssk sorgulama
emekli sandığı
cinsel bilgiler
radyo dinle
korku
evlilik
ssk hizmet dökümü
ssk emeklilik sorgulama
hikaye
bağkur hizmet dökümü
gazeteler
ilan
ssk
evlilik sitesi
ehliyet sınav sonuçları
iş ilanları
bağkur borç sorgulama
gazete oku
Post a Comment