AP-GfK Poll: Americans split on health care repeal

[underpants]

Source: AP Yahoo

WASHINGTON – First it was President Barack Obama's health care overhaul that divided the nation. Now it's the Republican cry for repeal.

An Associated Press-GfK poll found likely voters evenly split on whether the law should be scrapped or retooled to make even bigger changes in the way Americans get their health care.

Tea party enthusiasm for repeal has failed to catch on with other groups, the poll found, which may be a problem for Republicans vowing to strike down Obama's signature accomplishment if they gain control of Congress in the Nov. 2 elections.

Among likely voters, 36 percent said they want to revise the law so it does more to change the health care system. A nearly identical share — 37 percent — said they want to repeal it completely

In the poll, only 15 percent said they would leave the overhaul as it is. And 10 percent wanted modifications to narrow its scope.



Read more: http://news.yahoo.com/s/ap/20101022/ap_on_bi_ge/us_ap_poll_health_care;_ylt=AqGqANcOZBp_zzEpjzdUUt5paP0E;_ylu=X3oDMTJwYTEwYTBuBGFzc2V0A2FwLzIwMTAxMDIyL3VzX2FwX3BvbGxfaGVhbHRoX2NhcmUEY3BvcwMxBHBvcwMxBHNlYwN5bl90b3Bfc3RvcmllcwRzbGsDYXAtZ2ZrcG9sbGFt

---

The AP-GfK Poll was conducted Oct. 13-18 by GfK Roper Public Affairs and Corporate Communications. It involved landline and cell phone interviews with 1,501 adults nationwide, including 846 adults classified as likely to vote in the November congressional elections. The margin of sampling error is plus or minus 3.3 percentage points for all adults, 4.4 percentage points for likely voters.

[SemperEadem]

1. my view on polls is this

if I wasn't contacted and asked to participate, then the poll does not reflect anything that I should be concerned about.

Gallup's revelation this morning tells me all I need to know about polls and pollsters.

[Lasher]

11. This is a non-issue anyway.

Unless someone can explain how teabaggers expect to get legislation past a Senate filibuster and a Presidential veto.

[totodeinhere]

18. There are others things that they can do to gut HCR such as refusing to fund various parts of it.

If the GOP gets control of the House, that will happen.

[Lasher]

20. That's true, but such a budget would still be subject to filibuster and veto.

In that case I'm assuming Obama and Senate Democrats would posses some intestinal fortitude, which puts my theory on shaky ground indeed.

[on point]

2. The SCIENCE of polling is legitimate

The science of polling is legitimate, even if some polls (and perhaps some firms) are not. It has nothing to do with whether you were contacted or not. Do not confuse a vote (where everyone participates) with a poll where only a random, representative sample is used, with a margin of error to reflect potential error.

The Nut job right is the group that rejects education and science because "Reality has a known liberal bias", and besides "knowin things is so elitist". They are Neo-liths. Don't join the nut jobs by rejecting science because you don't

1. Get it ( so study up on probability, stats and sampling)
2. Are unhappy that some firms are betraying that science to further their political beliefs

[bemildred]

5. Polling is not science, and it is not based on science.

It is based on statistical methods (math) and certain highly questionable and unproven assumptions about the nature and distribution of political opinions.

Let's take an example, "American opinion" of the health care reform law:

1.) How do you know that such opinions follow a normal distribution? Well you don't. Following a normal distribution is mathematically a fairly stringent condition, but it is just assumed, because it creates a "bootstrap problem" if you don't; you have to take another sample, a large one, and do some more statistics to see whether what you want to study seems to follow a normal distribution well enough. All of this is usually just skipped over because it is inconvenient.

The title of the OP (opinion is divided) actually demands that opinion on health care reform follow a non-normal distribution, with two or three peaks, not one, but this is blown off when the statistics are calculated.

2.) How do you know that opinions on that subject fit in one of the categories provided? You don't, but in fact they will try to cram you into one of the categories or ignore you if they cannot. There is nothing scientific about ignoring inconvenient data (though it's common enough anyway). A normal distribution is continuous, not discrete, so to use a discrete set of opinions to sample a supposedly normally distributed opinion population is already fatuous.

3.) How do you know that opinions given to a poll reflect how people will vote? Well, you don't. In fact people deliberately lie to pollsters, and people who will answer pollsters are already a self-selected biased sample. Even then, they are predicting a future event.

4.) How do you know whether someone you call is likely to vote or not? You don't. Etc.

Well, that should give you the idea. Polling is "faith based", a hodge-podge of convenient assumptions and theoretically dubious heuristic methods. I am OK with it as a form of entertainment or speculation, like Astrology, but it annoys me when pollsters pretend that it is either scientific of has any real predictive value.

[on point]

8. Thanks for expanding discussion with worthy points

Stats and polling is a science. Granted practitioners do not always do the prep work, or sample validation they need to do to validate the samples.

However, your points 1-4 are examples of poor quality work, and are not problems with polling science per se.

1. True that not every population will follow a normal distribution on every question. Usually this work is done up front to qualify and profile the population in order to use it later and tied to larger predictive things like income. This pre-work for the US was generally done long ago and is out of date and needs a serious re-work. Even if the population does not follow a normal distribution, there are still probability tests with error margins that can use any of these other type distributions.

2. There is supposed to be an 'other category' for those who don't agree with the questions posed. Really good samples might also phrase the questions in different ways to see if that correlates to changes in the distribution to see if there is introduced bias. They would also do a test of their samples on a 'complete' sub-population to validate the larger sample.

3. A poll is a snapshot in time and does not indicate past or future behavior. Mis-understanding of what a poll is. Also, unless the poll is somehow anonymous, there is also peer pressure bias introduced as people may say what they think is politically correct.

4. You missed a major population bias which is very important now. 'Calling' people is no longer a random sample as people who have land phones, answer them to unknowns, and are willing to talk to pollsters are not representative of the population as it may have been in the 50s and 60s. Other population studies have shown this 'called' population to be older, whiter, more rural and more conservative. For political purposes, more 'republican'.

As I said, the science of polling is good, even though some of the practices of the firms done today are not. Work on pointing out the flaws and bias in the practice, not in the science itself unless you have worked out some new quantum fractal alternative reality probability theory deserving of the Nobel prize. Otherwise, the progressive wing begins to resemble the know-nothings of the ignorant and delusional right wing.

[bemildred]

10. Polling is not science, saying that does not make it so.

1.) A normal distribution is not normal, as I said it is a stringent mathematical condition, it ought not ever be assumed. All of those error interval calculations are based in most cases on assuming one has either a normal or Poisson distribution of some form. Most of the real world is chaotic and messy, it is like oatmeal, it has lumps in it, it does not have well-defined central peaks and tails that quickly taper off to "nothing".

2.) You have to show that the categories fit the population to even have a fig-leaf for using them, the population comes first, the categories, if there are any, follow. This is a major beef of mine, polls uniformly fail to show that they are discussion something real, they make some questions up and assume they make sense and reflect reality. In fact good questions are hard to come by, and lots of questions are useless or misleading in themselves.

3.) A poll is an attempt at a snapshot in time, whether it is that in fact or not is another question, other than that I am OK with #3.

4.) I did not "miss" it, I ignored it, I was not concerning myself with the many methodological issues of polling itself, which - as you point out - one can also easily criticize. I was addressing the fatuous claim that it is "scientific".

[RaleighNCDUer]

15. There is no such thing as 'normal'.

Using the exact same poll, the exact same wording, the exact same method of response gathering in the NW, the midwest, the NE, the SE, and the SW, and drawn equitably across all five regions, and you will come up with 6 distinctly different 'norms'. Do the polls on Saturday and on Wednesday, and you will get still other results. Do the polling in the morning and in the evening, again, different results.

Change the size of the poll, from 500 to 1500, to 15000, and you will get different results again. The larger the poll, the better reflection of reality. The smaller the poll, the greater the chance for wildly incorrect responses.

Pollsters know very well how to target predictable results by setting such parameters - simply by determining the day and hour of the calls, they can skew the results. Wording that has one connotation in one region may produce completely different connotations in another.

Polling IS a science. So is propaganda.

[soryang]

3. Employment based health care is a fraud

Unless you have a net worth in the millions or die a sudden death, it is guaranteed to bankrupt your family.

[ramapo]

6. Inefficient and discrimanatory yes but a fraud?

Many countries use employment based health insurance as a means of delivering on universal coverage. The difference between our 'system' and those found in the civilized world is that the civilized world has strict regulations on requirements for basic coverage, premiums, and portability. Employees have a choice of plans and this choice is not limited to what your employer feels like providing. The mechanism for premium payment just happens to be through the employer.

Our hodgepodge of coverage creates great inefficiencies and inequities. For the most part, employment based health care provides decent coverage although there are plenty of examples where inadequate policies do result in large financial burdens on policy holders. Theoretically, limits on coverage go away when the new law is fully implemented but there is already a lot of push back from companies like McDonalds who provide very limited coverage.

[soryang]

16. You lose your insurance when you need it most

When you are so sick you lose your job, you lose your health care. The problem, contrary to popular belief hasn't been solved. It's a deliberate catch 22 built into the system. The system takes it's exorbitant profits throughout your working career but when the biggest claims come due, when you can't work, they skip out of the contract and put the risk that you insured for on the beneficiary. That is a systemic fraud.

[AlbertCat]

4. "split"?

Like, in half?

No... a majority want it to remain the same, be tweaked, or overhauled to be even more effective.

[Historic NY]

7. They question they should ask is do they know whats in the current plan that passed?

[bluestateguy]

9. So

51% either support the law as is or favor revisions, while 37% support full repeal.

Morons in the media can't even read a poll.

But by all means, please, go ahead and pass a repeal. I'll then look foward to President Obama's Rose Garden ceremony as he vetoes it.

[and-justice-for-all]

12. Roper Public Affairs and Corporate Communications who?

:eyes:

[HughMoran]

13. So a majority - 51% want it unchanged or expanded

61% want it expanded, unchanged or reduced in scope, but not repealed.

Republicans are wrong. Again.

[underpants]

17. Yes that is what I took out of it too

but no we have to have STRIFE

Pitting the people against each other- nothing new

[RaleighNCDUer]

14. So, according to this the healthcare is a success.

47% want it to do less or go away. 51% want to keep it as it is, or expand it further.

Sounds like a win to me.

(edited for mid-afternoon tiredness addition)

[totodeinhere]

19. 51% is hardly a ringing endorsement. n/t

[RaleighNCDUer]

21. Stronger than the endorsement of its elimination.

not to mentioin that if you get into details, and ask about specifics, the antis have no fucking clue what they dislike, while the pros are far better informed of whats in the bill, who is helped, and how much the overall plan saves.

There's times I'm torn about the messaging from the WH - I don't know if they just suck at getting the message out, or if they've simply come to the probably correct conclusion that the antis are not going to listen no matter what they say, so why spend resources saying it.