by 52 to 40.
Here’s a link …http://newsbusters.org/node/3496
Why do they do that? Is there a legitimate rationale?
(Please resist the temptation to attack the source for a little while anyway. The link above has a link to the actual report by the pollsters if you’d prefer)
Note that a lot of that 52% is “moderately democrat,” which may have been independent/neither a few years ago. I’m going to assume it’s a random sample, and NewsBusters is trying to spin it as crooked polling because they don’t like the results. I’ve seen the argument made for at least a dozen years in attacks on “liberal media.” When polls favor their candidates and causes, they are on the pulse of the American people. When they take the other side, they are the consequence of media bias.
What amuses me is that a slate of polls show the same tilt towards Democrats (http://www.pollingreport.com/2006.htm), but that the readers of that website can’t believe there would possibly be consequences to the past several years of mismanagement, scandal, and ineptitude by their party. No, it HAS to be the liberal bias.
Anyone got a link to the results broken down by states, and/or specifically confined to the states that went for Bush in the '04 elections?
I didn’t know that representative sampling had to include an equal number of republicans and democrats. Wouldn’t that be kind of a stupid poll, where you decide beforehand what the results would be. Is there any real evidence the poll was non-representative sampling?
I mean, if you took a poll to find out whether or not homosexuality is a sin, and made sure you had an equal number of fundamentalist Christians and homosexuals, isn’t that, in and of itself, skewing the results?
It’s been awhile since I’ve studied statistics, polling, and representative sampling, so I need a refresher.
Up until the latter part of last year, polling routinely oversampled Republicans. If I searched What the’s posts, would I find a thread questioning the polling methodology employed back then?
So what explains the shift for polls to have more democrats in it now? One possibility is the nefarious “them” who conduct the polls shifting their sampling strategy from oversampling one side to oversampling the other. Alternatively, perhaps more people are identifying as Democrat now because of current events? Maybe more conservatives are refusing to participate because of current events?
As I’ve said before, polls skewed one way or the other are not good things. Group representation should proportionally reflect the larger population. If it doesn’t, one must regard the results with some suspicion. On the other hand, since appropriate caveats weren’t put on prior polling when it supported conservatives, and since folks like What the weren’t taking time to express their concerns about polling methodology then, I’m not spending a lot of time shedding tears for them now.
Various polls differ on how one should collect data. Some feel that a representative sample should match the percentage of people who define themselves as Democrat or Republican. For example, if a large poll is taken on party affiliation, and it turns out that 52% of the people identify themselves as Democrats and 48% define themselves as Republican, then each sampling should contain roughly the same percentages of Republicans and Democrats to be representative of where the country is.
Others believe that you should just ensure a good random sample, and let the chips fall where they may. Checking party affiliation may still be a good way to see if you sampling is truly random. For example, if the people in your sample identify as Democrat vs Republican to the tune of 70-30, then you’ve probably got a skewed sample and should try again.
Some polls will simply correct their results based on the stated party affiliation. Some simply report it and leave the results the way they are, letting the consumer use all the information provided to draw his own conclusions.
And some polls are simply biased, either in the questions they ask or the way they sample people. I’m not accusing this poll of that, but you can certainly find them - usually conducted by an organization with an axe to grind.
Well, it depends. If there are an equal number of fundamentalist Christians and homosexulas in that population, then your random selection should reflect that. If it doesn’t, then maybe something is wrong with your selection process.
Same thing goes with political sampling. Is your sample skewed toward urban voters? That might make it lean more Democratic. It’s not so much that you want to force your sample to have a certain mix, but if you find that your mix is signficantly different from the population as a whole, then you need to recheck your sampling methods.
I’m a bit surprised at this poll because generally when the pollsters ask for political affiliation, they also break down the results by that affiliation. That way, if you find a large number of Republicans dissastified with the administration, you can better gauge how vulnerable that party is to losses in an election.
The cite is hardly neutral. Show me an objective source and I’ll consider it.
Did you read the AP poll site? That’s the key. The fact that it was a righty blog is besides the point. The AP story was reported widely-- I read it in my own news paper a few days ago. The blogger links to the source material. I don’t see a problem with that. Go to the source materal and give your opinion.
Why? The population is skewed toward urban voters – that is, more people live in cities and suburbs than in the countryside. That’s just how it is.
Regarding the political implications of that, see The Emerging Democratic Majority, by John B. Judis and Ruy Teixeira – http://www.amazon.com/gp/product/0743226917/qid=1136752753/sr=1-2/ref=sr_1_2/104-0764394-7859165?s=books&v=glance&n=283155.
Do you know that Bush carried almost all of the fastest-growing districts in the last election? The conventional wisdom I’ve read is that the population is growing in the ‘red states’ much faster than it is in the ‘blue states’.
There’s also a growing trend to move out to the suburbs and out of the cities, which also favors Republicans.
The point is that if the population overall is 52/48 Republican/Democrat, then your random sample should also have roughly the same split, or you oversampled some areas. Whether they are urban areas is irrelevant.
I don’t understand your point. I specifically said urban. And suburban ≠ urban. Hence the “sub” prefix, but more importantly the voting patterns are different.
Cite?
(And it better not include any stats from Ohio! :mad: )
Judis and Teixeira’s thesis is that a metro area such as San Francisco or L.A., where a “postindustrial economy” thrives, is an “ideopolis,” a center of ideas; and this affects the thinking and the voting behavior of its residents in general, urban and suburban, mostly towards the “blue state” values and away from the “red state” values; and that demographic trends are towards a future in which “ideopolis” residents will be a numerical majority overall.
How different, exactly? Stats?
The Amazon listing for the book excerpts a review from Publishers Weekly:
See also Teixeira’s weblog, “Donkey Rising” (and stop that snickering! 
) which has ongoing analyses of demographic/political trends: 口臭予防に青汁を飲む