Sorry, but I don’t consider the premises on which the necessity to go to war in Iraq are based to be “a rather small handful of issues.” Add to this the additional questions regarding a number of other international issues (the Kyoto Protocol, the International Criminial Court, environmental and labor standards in trade agreements, international public opinion regarding Kerry and Bush, …) and I don’t think the set of issues is small at all.
The PIPA report did not poll people on their likelyhood to vote. It only issues it’s statements relative to Bush/Kerry supporters, not to projected Bush/Kerry voters. And of course it was done before the election, so it doesn’t contain data on actual Bush/Kerry voters.
Why is that relevant? Well, in pretty much every quoting of the report, here or elsewhere, the statement is along the lines of “Bush won because the people were misled/ignorant/stupid”. I.e. every angry pit thread that contains reference to PIPA talks specifically about voters, quoting a report that only gave data on “supporters”.
Now, approximately half the possible voters actually voted. So, assuming PIPA used a reasonable sample, it’s fair to assume that half of their sample wasn’t valid to any argument that contains the phrase “Bush won” or any other phrase that involves voters. Half is a big number.
More importantly, where did the ignorant people fall? equally into the voting half and the non-voting half? or were they more inclined towards one side?
Well, seeing as how less educated people tend to vote less and more-educated people tend to vote more, it’s not crazy to propose that more of the ignorance was in the non-voting half.
Sorry about the cherry picking, but life is busy. Just cut and paste if there is something you really want me to respond to that I have neglected.
you with the face
I suppose the feeling must be mutual. I don’t see how it is useful to know this. Like I said before, it is nice to trot out in SDMB debates, perhaps, or it might be fodder to fuel liberal angst.
Other than that, this statistic can only be dubiously tied to any observed behavior. Do we know anything about people who actually bothered to vote? No. Can we make any claims about ignorance as a heuristic used by Bush supporters? No. Do we know anything about the possible stochastic disturbance that might condition this correlation? No.
This might have been interesting if the researchers had controlled for variables like education. It would have been very informative if they could have reported that holding education and income constant, Bush voters were still more ignorant about the Duelfer report than Kerry voters. That could have put some real meat on the cognitive dissonance hypothesis, or at least shown us that ceteris paribus, Bush voters are more ignorant. It didn’t.
If I were one of those mysterious values & morals voters, I might not care very much about the contents of the Duelfer report, since Bush would still be the better candidate for me. It would be perfectly rational for me not to invest the time in researching the issues if the probability that my vote would actually change is very low. People use heuristics like this all the time. They are not necessarily ignorant or stupid, though many very well may be. Consequently, I do not give a dirty red cent about researching the president’s faith-based initiatives (save to debate them on the SDMB), because the likelihood of my supporting them is pretty low. Feel free to call me ignorant. I’ve been called worse before.
The real point is, I believe that when you apply even a little bit of critical thought to this report, the utility of its findings evaporate entirely. I have said as much. It is not much better or worse than the standard reports of Gallup or other such polls. Those are often uninformative, too. Go figure. I suspect you actually have to pay them to get the interesting results.
Bill H.
Amazingly enough, this is not true. The relationship between education and turnout is very tenuous, to say the least. You might find it hard to believe, but it is true. A lot of the problem is how one codes education as a variable, but that is a separate issue.
Suppose the following. Educated people, who tend to make more money, have a higher opportunity cost to voting. Missing an hour or two of work, or six hours if you life in Florida, means an awful lot if you can bill, say, $500/hr as opposed to making $5.15/hr. Because their opportunity costs are higher for missing work, educated people place greater weight on their utility calculations for voting. They consider the probability that their vote will make a difference. If you are living in a locked up blue state, you might assume that taking an hour or two to vote just isn’t worth it, no matter which party you support. Everyone in your state is making a similar calculation. They consider their preferences over policy outcomes, how big the opportunity cost of voting is, and the expected behavior of everyone else.
The long and short of it is, given how little most people actually care about politics, it is amazing that anyone ever votes at all, especially educated people, whose probability calculations are perhaps more refined and their opportunity costs are higher.
This weakens the value of the PIPA findings as well. A random sample will very likely poll people in red states. There are a lot of red states. Suppose these people are uneducated. Suppose they also believe that Bush will win their state. Since they don’t plan on voting, they aren’t going to bother educating themselves on the issues. Why should they?
Hence it is essential not to poll just registered voters, but likely voters. I believe there are fewer ignoramuses in the set of people who actually care enough to vote and who are not motivated by a single issue (abortion, gay marriage, etc).
DING DING DING DING!! We have a winner! Assumption #4 is indeed the primary issue. At least if we’re letting you have your givens. I would quibble about Assumption #1 as well. I know more than one person who votes via coin flip. I think assumption #2 may be problematic as well because I don’t think it is my place to decide how my fellow citizens decide. In an ideal world, and in the classical formulation of a democracy, this would be true and all voters would be making informed decisions. Still it is not my place to decide if someone else’s vote is “defective” or not(Assumption #3, please line up against the wall).
Still, going along with your givens Assumption #4 is the issue. Take a look at the election exit polls to see a list of some of the issues which the PIPA report did not ask about. Taxes, Abortion, Corporate influence, Environmental policy, Gay Marriage, the list goes on and on. This is why I keep saying it would be wickedly difficult to pin down the exact motives in play when someone votes for a presidential candidate. For most people(except a few rabid single-issue voters) it is a matrix of issues for even a single person. Maeglin’s formula had ellipses in it for a reason. That middle term will be HUGEJIGMUNGEUS. So big I had to make up a new word. You just can’t shoehorn people like that, at least not if you want to have a tight confidence interval.
Enjoy,
Steven
On Preview:
That depends on how the randomness is selected. Random among the entire population via a single identifying key(like a SSN, but not the real SSN because that is under lock and key, right, RIGHT?!?) will be blind to geography. More structured randomness, like a percentage of the pool proportionate to each state’s population being randomly selected from a statewide list of residents(Driver’s license possibly?) would not be blind to geography but would take this into account by calling more people in the more populous states.
Maeglin wrote
Everything I’ve read disputes this. I don’t have any cites I’d call solid, but there are these (from a google search):
From http://www.soc.ucsb.edu/projects/ct4/pages/seminars/Rountables/Kamens.doc, a study in Confidence and Cynicism in American Institutions
http://www.spcomm.uiuc.edu/Users/salthaus/althaus%20and%20trautman%20AAPOR04.pdf, a study on the impact of television market size on voter turnouts.
and
I’d enjoy reading more about this negative correlation claim. Cite, please.
But, in the end we agree on my premise in my last post:
Steven,
Okay, I think I see it now.
Perhaps there is yet another hidden assumption in my argument, or maybe a better way to state #4: The questions asked about in the PIPA survey are the important questions.
This is of course an evaluative claim, and I can see that reasonable people could disagree w/ its truth, and so now I can see how reasonable people could disagree over whether the survey has any significance.
But it’s important to note, I think, that it’s probably a safe assumption to say that lots of people who voted for Bush do think those issues are important.
But of course we can’t know for sure that this is the case, and we certainly can’t knmow sure that those people would have gotten those questions wrong, without doing another study. (See, I’m getting it… 
)
-FrL-
Maeglin wrote
Oh, by the way, PIPA didn’t even poll registered voters, let alone likely voters. They polled adult Americans (with telephones)
But they are. Note I did not mention how important the issues were. I only mentioned that their number is pretty small. Specifically, this poll only concentrated on the issues surounding WMD and collaberation with AlQaeda. Surely those are a pretty small number of issues.
Except, of course that some of those issues are not as clear cut as the issues of WMD and terrorism by Iraq. If I recall the Kyoto question was simply …
Oh heck, lets just look it up.
"*Based on what you know, do you think the U.S. DOES or DOES NOT participate in the following treaties and agreements?
The Kyoto agreement to reduce global warming Does Does Not
Bush supporters…36 49
Kerry supporters…31 58
The International Criminal Court that tries individuals
for war crimes, genocide, or crimes against humanity
if their own country won’t try them
Bush supporters…45 42
Kerry supporters…39 48*"
Then there were some question about whether or not the respondant thought each candidate held certain positions about a variety of issues. I’m not sure the way those questions were asked that we have useful information about them. Buit given that most of us could come up with as many issues which were not covered, yes, I do think “handful” is an accurate characterization. I will withdraw the addition of “rather small”. Is that better?
I make it 3 about who the respondant will vote for, 1 about Bush’s honesty, 14 unreleased, 12 about the Iraq war 5 about other issues. Ignoring the first3 catagories, I make that Iraq and 5 other issues. Maybe I will put “rather small” back.
You found the unreleased question in another thread. How do you make the count?
Geezus Christ, people. These are huge effects we are talking about and huge effects will be robust to these sort of things unless you posit a countervailing huge efect. (I.e., there doesn’t have to just be a correlation between knowledge and whether they vote; there has to be a freakin’ gigantic correlation.)
Let’s go through a little example. First off, the turnout rate in terms of eligible voters was nearly 60%. So, let’s postulate that was what it was in our sample and that it held evenly between Kerry and Bush voters. (Since the PIPA poll showed a tie between Bush and Kerry and the actual results were pretty close, this is further evidence that this is probably a good assumption. Also note that 62% of the PIPA sample at least claimed to have voted in the off-year 2002 election, so assuming 60% of them voted this year is not unreasonably high…If anything, it might be a bit low.)
Okay, now let us turn to the question about WMDs immediately prior to the war. 25% of Bush supporters got the generally-accepted-to-be-correct answer that “Iraq had some limited activities that could be used to help develop weapons of mass destruction, but not an active program.” For Kerry supporters, the number was 51%. So, let’s suppose that Kerry supporters who vote are no smarter or dumber than those who didn’t vote and so they also got this correct at a 51% rate. Then, the question we can ask is: “How much smarter would the Bush voters who voted have to be than the Bush supporters at large in the poll in order to match this rate of having 51% of the voters being correct?” Well, the answer is that it is impossible. Because 51%*0.6 is already a larger number than the 25% of the Bush supporters who knew the correct answer. So, even positing that Kerry voters are no smarter than non-voting Kerry supporters while Bush voters are infinitely smarter than non-voting Bush supporters, you still can’t bridge the gap. You’d have to posit a combination of Kerry voters being significantly dumber than Kerry non-voters and Bush voters being significantly smarter than Bush non-voters in order to have it come out that the same percentage of Bush voters and Kerry voters got the correct answer. [For simplicity, I used “dumber” and “smarter” here to stand for “less knowledgeable” or “more knowledgeable” on this particular subject.]
So, do you begin to get the picture why these various things you are throwing out ain’t going to likely make much of a difference?!? This sort of thing is why when polls are taken, people can argue about whether the difference between likely voters might explain a discrepancy between Kerry ahead by 1 and Bush ahead by 2 but it ain’t going to make up for, say, a 65% to 35% margin. It is also probably why Kerry did not contest the vote in Ohio. There were apparently enough absentee and provisional ballots to more than bridge the difference between Bush and Kerry, but lacking any reason to believe they were going to practically all break for Kerry, it was extremely unlikely…read “essentially impossible”…that Kerry was going to catch up once those votes were counted.
Where do you come up with this stuff?!?! Honestly, I feel like the person I used to see posting under the username of Maeglin was much better than this. Why are you even proposing that the way they did their random sample was to poll the same number of people in each state or in some other way that would over-weight the red states? It just boggles my mind why you believe they created their random sample in this way.
And, by the way, they have the demographics of their sample although it is only broken down to the regional level. Still, the fact that Kerry and Bush polled virtually a dead-heat doesn’t leave much room for them to have over-polled in the red states.
None of this negates the observation that more Bush supporters were found to harbor erroneous beliefs than Kerry *supporters *. If people extrapolate this to actual voters, that is hardly the fault of the study.
All right, let’s frame all this in some kind of hypothetical. Out of 1000 randomly selected individuals, less say 500 self-identified as pro-Bush and 500 self-identified as pro-Kerry. Let’s say 375 of the pro-Bush missed the question “Did the US find WMD stockpiles in Iraq post-invasion?”, while only 125 of the pro-Kerry did. What could have caused this discrepancy?
Is it because the pro-Bush group is disproportionately undereducated when compared with the pro-Kerry group? Okay, let’s assume that it is. How would that take anything away from the conclusion that Bush-supporters are more misinformed/deluded about certain issues than Kerry-supporters? The painful truth is that regardless of if Bush supporters are less educated, older, religious, or whatever, the final observation does not change.
But let’s keep going. To keep it simple, we’ll speculate that the less educated folks are solely responsible for the discrepancy. Out of a group of 500 Bushites, let’s say (as you speculate) only half (250) will potentially vote. Then let’s assume that no undereducated folks will vote and that 75% of the potential voter group is undereducated (because that is the percentage of Bush folks who missed the WMD question). That means a grand total of 63 votes will go to Bush out of the sample.
Let’s compare that with the Kerry group. Out of 500 Kerryites, 250 will potentially vote. The percentage of undereducated will be 25% because that is the percent who answered the WMD question incorrectly, and, again assuming the undereducated won’t vote, that means 188 votes will to Kerry from the sample.
That leaves a score of Bush-66, Kerry-188. Hot damn, it looks like Kerry should have at least won the popular vote. But alas (barring any hanky panky at the polls), Kerry did not win the popular vote. This tells us something other than your explanation is at work, Bill.
I note that in other threads, Bricker is now saying that this thread “thoroughly debunks” the results and the study. Is there something I’m missing here??
That’s what I’ve been asking whenever Bricker says this in other threads.
I don’t see any debunking at all here. I see a discussion of what conclusions it’s appropriate to draw from the study, and plenty of convincing arguments supporting the obvious one: Bush supporters were less informed about these topics than Kerry supporters.
Citing this thread as evidence that the study has been “debunked” is misleading, at best. One might argue about what further conclusions to draw from the study, or whether the study reveals anything about the reasons for supporting one candidate over the other, but to claim that it doesn’t show a knowledge gap between Bush and Kerry supporters on these particular issues is laughable.
Well, I don’t know about it being ‘debunked’. To my mind, lurking in this thead, what I got out of it is that drawing any kind of hard and fast conclusions is certainly debatable. Generally when people are now throwing around the PIPA report (either way) I take whatever they are saying with a rather large grain of salt.
As to this:
Again, I think it shows a POTENTIAL knowledge gap between Bush and Kerry supporters on CERTAIN issues. Any concusions drawn beyond that break down on partisan lines IMO. And you guy can keep hammering away on this but thats just how I see it having carefully read through the thread. Shame the OP hasn’t been back in a while.
-XT
The word “POTENTIAL” above should be at very least “PROBABLE”. My preference would be “VERY PROBABLE” since a margin of error of 3% on the main thesis is pretty clearly in the realm of the probable versus the merely possible. It breaks down roughly to 30-1 odds(actually worse, but the math would be daunting to write here). How often are you going to lay down money at those odds at a place where you know the bookies are doing their homework? Would you say “The team facing 30-1 odds has the POTENTIAL to lose” or would you say “The team facing 30-1 odds will PROBABLY lose”?
I have been careful, as has virtually everyone else in this thread, about conclusions beyond the very probable disproportionate ignorance of self-identified Bush supporters on these issues versus self-identified Kerry supporters. I have personally poked holes in attempts to link the study results to voting behavior. I have also noted that their analysis, beyond that one primary premise, seems conjectural and politically driven. This does not mean the data is useless or “bunk”. As long as the report is used to support claims of disproportionate ignorance on these issues then it is absolutely legitimate. The report would be of special significance to educators or media as a gauge of their effectiveness in educating/reporting on these issues.
Enjoy,
Steven
Actually, I don’t even know how you are getting these 30-1 numbers. As I noted in a previous post, the probability of flipping greater than 60% or less than 40% heads from 1000 flips of a fair coin is less than 1 in a billion. If you look at the sort of differentials one is getting in this survey on some of the questions, the odds of it being due to chance are at least that improbable.
The sort of differences that are being seen here in a sample size of 1000 people on these issues falls into the realm of being about as close to absolutely certain as you are ever likely to get in a statistical study. I am sure any medical researchers would be drooling at these sort of statistics. Hell, they would practically be drooling if they saw percentage differences this pronounced in a sample size of only 100 people.
Dangit, I’m thinking confidence interval even while I’m reading/typing margin of error. A confidence interval of 97% would indicate that 97 out of 100 surveys would show these numbers(within the margin of error). So that means 3% would show something wildly different. If we were betting on the results of the exact same survey with a different pool of respondents a bookie would say that 97% to 3% chance of getting these results versus different results would simplify to a little worse than 30-1 odds.
Enjoy,
Steven
So, I am still confused. Exactly why is it only a “POTENTIAL” knowledge gap? Is the new standard that a survey’s results have to be statistically significant out to, say, there being only 1 chance in 10^20 that they could be gotten such a result by chance in order to be believed?
I suppose if I let go of the apple I am holding, it will potentially fall down?
Because I don’t think the actual report makes any definitive statements about the wider question of a GENERAL knowledge gap between Bush and Kerry supporters which is what I was commenting on…only a high probability that there are indications of a knowledge gap on CERTAIN ISSUES between Kerry and Bush SUPPORTERS. To me that makes it a POTENTIAL knowledge gap, jshore. You obviously are completely convinced. I’m sorry that I’m unconvinced by a single study, no matter how supposedly low the margin of error is.
I’m just skeptical that way. I’d be making the exact same arguements if the Pubs produced a single poll that indicated that Kerry supporters were elitist snobs who are out of touch with the population at large. To me, if you are going to make sweeping claims (like there is an obvious knowledge gap between Kerry and Bush supporters) its going to take something thats a bit broader as far as the questions asked (i.e. don’t ask questions that are obviously part of the standard Dem meme (I note it was a CORRECT meme btw) for this election only and be surprised at the results)…and its going to take more than a single poll of 900 random folks to convince me. Reguardless of what you are polling for. Perhaps I’m being unreasonable here. I’m like that sometimes.
-XT
Riiiiiiiiight. Because there has been absolutely no evidence aside from this study that conservatives mistakenly believe that WMDs were found in Iraq or ties to terrorism established. And certainly none of them have been cited, repeatedly, on this very board. Nu uh, no way, didn’t happen.
Hey, did you hear about all those WMDs we found in Iraq?