Chutzpah is the quality of audacity, for good or for bad. The word derives from the Yiddish khutspeh…
(from Wikipedia)
So I’ve been following the diminishing aftershocks from RFK Jr.s pathetic swing and whiff on voting at Rolling Stone (my original dismissive comments on his efforts are here).
Now please note that I believe that electoral integrity is critically important, and at risk. I don’t believe that recent elections are significantly more at risk than elections have been here in the US (in, say Chicago or other machine cities), but they are more at risk than is acceptable and that needs to change.
But my argument keeps getting undercut by these clowns.
Now I get led over to Salon (where I launch a lame Flash ad while checking my son’s homework), in order to read a defense of Jr’s claims by Steven Freeman, who authored a book that – Freeman claims – supports Kennedy’s claims.
I hope like hell his book does a better job than his Salon piece, though – but I doubt it.
Here’s Salon:
Are exit polls usually accurate?
Yes, they are. On Nov. 2, 2004, Manjoo’s source Mark Blumenthal, the Mystery Pollster, had this to say: “I have always been a fan of exit polls. Despite the occasional controversies, exit polls remain among the most sophisticated and reliable political surveys available.” Properly done exit polls are highly accurate. Given the large sample size in U.S. exit polls, they ought to be accurate within 1 to 2 percentage points of the official count.
Now, I don’t know about you, but I have nasty habit of actually clicking on the links people put in their web writing. So I click on over to Blumenthal, and read a post that opens with this:
Is RFK, Jr. Right About Exit Polls? – Part I
Late last week, Rolling Stone published an article by Robert Kennedy, Jr. that asks provocatively, “Was the 2004 Election Stolen?” While it covers many topics involving alleged suppression and fraud in Ohio, the article disappoints in its discussion of the exit poll controversy, because on that aspect of the controversy Kennedy manages to dredge up nearly every long-ago discredited distortion or half-truth on this subject without any acknowledgement of contrary arguments or the weaknesses in his argument. It is as if the exit poll debate of the last eighteen months never happened. With this two-part post, I want to review the article’s discussion of the exit poll controversy in-depth, for it provides a good opportunity to learn something about what exit polls can tell us — and mostly what they cannot — about whether fraud was committed in the 2004 elections.
But then goes on to say this:
And yes, if you look back at my first post on exit polls on Election Day 2004, I too described exit polls as “among the most sophisticated and reliable political surveys available.”
However, I have certainly learned a great deal about exit polls since then, and calling them the “most reliable” of surveys ignores a host of other practical challenges. Exit polls generally sample a larger number of voters than telephone polls, but they do so because the “cluster sample” technique used on exit polls– which first selects sample precincts and then voters at those precincts — has more sampling error than comparably sized telephone poll samples. Exit polls also miss the growing number that vote by mail or cast absentee ballots.
[emphasis added]
…and this…
…one of the most blatant omissions from the Kennedy article: U.S. exit polls have been wrong before. In fact, according to the Edison-Mitofsky report, they have shown a consistent discrepancy favoring the Democrats in every presidential election since 1988. And while the 2004 discrepancy was the highest ever, they were almost as far off in 1992. More specifically, the “within precinct error” (WPE) reported by Edison-Mitofsky showed differences favoring the Democrat of 2.2 points on the margin in 1988, 5.0 in 1992, 2.2 in 1996, 1.8 in 2000 and 6.5 in 2004 (see p. 34).
Go back and watch the classic political documentary, The War Room — or easier, go back and read my post from January 2005 — and you will see that that leaked exit polls on Election Day 1992 provided as distorted a view as those leaked in 2004. The difference was that the leaked exit polls in 1992 were known mostly to insiders and served to exaggerate the size of Bill Clinton’s eventual victory. Clinton won by less than those early exit polls suggested, but he still won the election, so there was little lingering outrage.
So – our intrepid author has – I can’t even use the work Dowdified or cherrypicked – done what generations of movie publicists have done – and simply edited a useful quote out of a review which says quite the opposite. Blumenthal – a professional pollster – clearly set out in his piece an argument for why the exit polls a) were not as wrong as RFK claimed; b) showed error that was not atypical for U.S. elections; and c) were not as accurate in the German elections as claimed.
At that point, I stopped reading the Salon piece – it’s dinner time – and I’m not sure I’ll finish. The staggering dishonesty of his quote has left me thinking that this is not the best way to spend my time.
Note that he’ll probably turn this into “Armed Liberal says: ‘…this is…the best way to spend my time.'”
He’s just an ellipsis away.
You know they have invested far too much intellectual capital in this to back down now. Regardless of all the evidence contrary to their “findings”, ignoring all the experts who have poked holes in their “theories”, they won’t ever back down, it’s just not in their nature.
They will continue to stick with this loser story because its part and parcel to their sanity. If they truly accepted that Chimpy McHitler whomped their Vietnam Decorated Nuanced War Hero Uber Candidate, then they would have to accept that their “base” is an easily swayed ignorant mass, just like those Red State Intelligent Design NASCAR inbreeders. And you know, they just won’t ever buy that.
After carefully reading the link, I have to conclude that you have cherrypicked your quotes also. But your purpose was to show that somebody else was shading the truth, not to accurately represent the link yourself.
I want to point out the key points myself, since neither the guy you quote nor you have done it.
The key point: Everyone — including the exit pollsters — agrees that the average discrepancy was statistically significant.
This is the key fact, without this there wouldn’t be much controversy. If there was nothing wrong here, and you bet that the results would be this far off by random chance, you’d lose 19 times out of 20.
So there’s something wrong here, and now the question is what’s wrong. Is it a problem with the elections, or is it somehow a problem with the sampling? The statistical results can’t tell us which of those went wrong. There are various possible problems with the sampling. Maybe people who vote Republican are less likely to admit it to a pollster. Maybe recently-hired data-collectors didn’t follow their instructions.
Well, what about history? If the same exit-poll methods were reliable before, and significantly wrong now, that would show there was something different now. Either more election fraud or more Republicans who refused to admit their votes etc.
U.S. exit polls have been wrong before. In fact, according to the Edison-Mitofsky report, they have shown a consistent discrepancy favoring the Democrats in every presidential election since 1988. And while the 2004 discrepancy was the highest ever, they were almost as far off in 1992. More specifically, the “within precinct error” (WPE) reported by Edison-Mitofsky showed differences favoring the Democrat of 2.2 points on the margin in 1988, 5.0 in 1992, 2.2 in 1996, 1.8 in 2000 and 6.5 in 2004 (see p. 34).
We can expect variation of a point or more from sheer sampling error. But it looks like Democrats consistently do better on exit polls than elections, and the difference was bigger than usual in 1992 and 2004. 1992. Bush senior’s loss. If it was more election fraud than usual, it still wasn’t enough to swing the election. Did they miscalculate?
The 2+ background bias could be because of systematic flaws in the polling methods. (For some consistent reason they’re less likely to ask Republican voters who they voted for, or the Republican voters are less likely to tell them.) Or it could be because of a low level of voter fraud that favors republicans over democrats by that much. Republican voter fraud may be just that much more effective than Democrat voter fraud.
There were also similar problems resulting in overstatements of the votes cast for Pat Buchanan in the Republican primaries in New Hampshire in 1992 and Arizona in 1996.
I haven’t done my homework on this one. We can expect significant results about 1 time in 20 from just random chance. If Buchanan got exit polls for 50 states there’d be nothing unusual about problems in 2 of them. Of course new hampshire is special. Maybe arizona is special too.
It would make sense to do election fraud mostly in states where the result is already close. A little bit could tip it over. It would be interesting to see whether the biggest differences came in the states with the closest votes. But to some extent that could happen structurally too. In massachusets where there aren’t as many Republican voters, there aren’t as many Republican voters to refuse to tell pollsters how they voted. So the difference would be smaller. And in alabama where there aren’t that many democrats there’s less reason for republican voters to deny their vote.
The statistical data says there’s really something going on here, does not say what it is that’s going on.
JThomas – thx for the further analysis. You are right, though, that AL’s point had to do with the dishonesty of the Salon piece rather than being a discussion of the technique.
The real worry is that a pattern seems to be emerging here that is reminiscent of previous historical attempts – many successful – to subvert democratic processes.
First is the denigration of the voters as stupid rubes unqualified to vote (the post-election headline of the Guardian (I think) accusing Americans of stupidity being the most egregious example but plenty of others spring to mind).
Next is the persistent accusation of fraud even in cases where only the most paranoid of conspiracy theorists believe it possible.
Follow up with copious quantities of actual fraud, harassment and even violence on the part of the Dem supporters (as far as I know every indictment for vote fraud in 2004 was against a Dem and the great majority of violent acts were committed against Republican offices or vehicles).
A small but vocal and sometimes violent segment of the Democratic party seems convinced that they represent the will of the people against all available evidence. They are steadily chipping away at the intellectual foundations of the democratic process itself. While the great majority of Democrats have little to do with the extremists, this is still an extremely worrisome trend.
J Thomas can’t give up the allegations of vote fraud even when a consistent result, of exit polls that favor Democrats, is demonstrated. For some reason he seems to think that Republicans are (or should be) embarrassed to admit that they voted Republican. Is this embarrassment enough to explain discrepancy?
I can imagine plenty of other scenarii that would lead to discrepancies on the order of a 1-6% undercount.
1. Perhaps exit polls are conducted completely unscientifically, by long-haired hippy-activist-type 20-somethings who naturally gravitate to people like them? They hear what they want to hear. The more frightening these poor dears find Republicans, the more their polls undercount them.
2. Perhaps more Republicans than Democrats despise exit polls and they way they are treated as gospel by the national media, and thus more Republicans than Democrats refuse to have anything to do with exit polls?
3. Perhaps the weaknesses of exit polls have destroyed their scientific value, while leaving the propaganda value of exit polls intact? Exit polls are lousy at predicting elections but great at shoring up the Democrats, who are after all the chosen party affiliation of 90% of reporters and media personalities. If exit polls are not scientific at all, then we shouldn’t expect accurate measurements out of them, only measurements that favor the desired party.
It’s possible that all these scenarii are correct. Occam’s Razor would indicate that these scenarios, based on reality instead of conspiracy theory, are more likely to be true than another scenario that alleges a massive, secret, un-leaked conspiracy.
This is definitely one of those Occam’s Razor issues.
Which is more likely? That the method for conducting elections has changed? (fraud) Or that people have changed their willingness to be honest or talk to exit pollsters?
To me, people’s opinions and actions are the greater variability. That’s not to say there is no fraud, only that polling is fraught with difficulties. If people weren’t so fickle, we wouldn’t need polling to begin with. Hence the variability of people’s responses means that polling will continue to evolve. Since polling problems and people’s fickleness are already established variables, why add others? It doesn’t add up to me, although I can see other points of view. I would caution, however, against pre-supposing an answer and then trying to make the facts fit the conclusion.
I have a major problem with this entire scenario. Think about what is ultimately being said here: the sanctity of the secret ballot be damned, we have reporters busy asking people how they voted seconds after leaving the polling place and their responses dont match the results. Hence someone is cheating, and assumedly the assumed will of the people should be overturned and the loser handed the election based on what exit pollers determine is the will of the people. If exit polls are so alledgedly accurate, why have an actual vote in the first place?
Occams Razor indeed! Which is more reasonable: a systematic conspiracy to defraud elections which is something that is feared, watched for, and analyzed in every election in history, or bad assumptions and/or methodology in exit polling, which apparently _no one_ is watching or auditing in a meaningful way… outside of a guy on the internet named Mystery Pollster. Doesnt the fact that a guy who I imagine wears a bag with a question mark on his head must be looked to as the only viable authority on this matter give us pause in and of itself?
J Thomas, I did go back and finish Freeman’s article (and I’d read all of Blumenthal’s). Yes, my point was simply to highlight Freemen’s egregious falsification of what Blumenthal said, not engage is a stats debate.
As Blumenthal points out, the exit polls have been wrong before, most significantly in 1992, when they overstated Clinton’s vote by 5%.
So we’re faced with a choice; we can believe in methodologically flawed polls – and note that the LoC for the error in 2004 is only 68% – i.e. there’s a 32% chance the error was simply random – or we can believe in a conspiracy tha thas subverted elections nationwide – which would involve thousands of people each election – for over 14 years – without one of those people being discovered, coming forward, or otherwise letting the cat out of the bag.
Hmmm…I’d need higher confidence levels on the numbers to suppert that theory, wouldn’t you?
A.L.
Pangloss, I’ll respond to your allegations when you have some sort of substantive allegation to respond to. What you’ve said is all spin, no substance.
With one exception, you suggest that these particular polls are not run according to standards. I’ve seen no evidence to support the idea that these polls are in any way worse than usual, except that they need a lot of people and some of the people are new at the job and perhaps don’t completely follow instructions — which I mentioned before. The pollsters can and should do spot checks for that. They should not publish the results for the same reasons the police don’t publish their patrol patterns. I haven’t heard whether they do check, but it’s elementary business sense to do so.
I don’t think this Kerry-is-the-real-president strategy is good for the Democratic Party.
I think they should stick with “Vote for Democrats and Prevent Hurricanes.” Less math involved.
Vote Democrat. Put The Crooks Back Where They Belong.
Sorry. Couldn’t resist that one. Nice double entendre, no?
How about “Keep Hope Alive: Leave Your Chad Hanging”?
Folks, it’s a serious issue, even if J Thomas hasn’t yet stepped to the plate and shown himself to be a serious critic.
Multiply his views by five million or ten million, and pretty soon we do have a crisis of confidence in the electoral mechanics.
That’s serious to me.
A.L.
Democrats: You do the voting, and we’ll worry about the counting
Seriously. It’s insane for any one party to somehow try to take the mantle of “we’re the only ones wanting fair elections” — it’s just suicidal. Instead of acknowledging compromise in things like clean water, clean air, business growth, etc, the Dems seem to be taking positions of absolutism: our way or the highway. At some point it’s just all a house of cards.
“The pollsters can and should do spot checks for that.”
Somehow the idea that the poll watchers inside the precincts spot checking ect is scoffed at (hence the reliance on exit polls to prove the corruption), but the idea that the exit pollsters are monitoring themselves is supposed to give us comfort? Who watches the watchers? Any self-monitoring entity dependant on the percieved accuracy of their work should be subject to skepticism. At least the actual poll workers and monitors contain partisand from both parties. Who balances out the exit pollsters and keeps them on the straight and narrow? Who (aside from bloggers) questions their methodologies? Who do they answer to?
I have to disagree. The complaints are all pretty much coming from the losing side and I don’t see why we should (a) assume that this is anything more than grousing because they lost and (b) why we should even believe that they’re being sincere in their complaints.
As far as I can tell, they’re crying “election fraud” much the same way they’ve cried “Bush lied, soldiers died” or “Pflamegate” for the last several years. IMO it’s just another political football to them rather than a serious policy issue and deserves about the same level of consideration.
Yes, my point was simply to highlight Freemen’s egregious falsification of what Blumenthal said, not engage is a stats debate.
Yes, when you’re proving the other guy is wrong there’s no need to be right yourself. 😉
Seriously, yes, when the point is that the other guy got it wrong, there’s no need to give any sense of where the truth is.
As Blumenthal points out, the exit polls have been wrong before, most significantly in 1992, when they overstated Clinton’s vote by 5%.
And the Democrat vote has been below the exit polls every election since 1990. What conclusion should we draw from this?
So we’re faced with a choice; we can believe in methodologically flawed polls – and note that the LoC for the error in 2004 is only 68% – i.e. there’s a 32% chance the error was simply random –
Oops! You got the detail wrong there. It’s 68% for a bunch of single precincts, and there were a lot of them. If it was only 68% for the nation we could just ignore it. We’d have errors that big in one direction or the other a third of the time.
or we can believe in a conspiracy tha thas subverted elections nationwide – which would involve thousands of people each election – for over 14 years – without one of those people being discovered, coming forward, or otherwise letting the cat out of the bag.
Hmmm…I’d need higher confidence levels on the numbers to suppert that theory, wouldn’t you?
The data doesn’t say. Of course the natural approach is to say that we already knew there was no vote corruption, of course we already knew that, so of course it’s all just ridiculous and there’s no need to give it another thought.
I see no reason to choose between those alternatives, yet. Not enough information to base a choice on, beyond prejudice.
However, entirely apart from the question whether republican voter scams have tended to be so much more effective than democrat voter scams that republicans tend to get a few percent extra votes, I want an election system that’s good at preventing abuse. Even if republicans and democrats have tended to average out their voter scams, I want a system that doesn’t depend on them to average out.
To the extent that the election process needs improvement, it will need to be a cooperative effort from well-intentioned people from all political perspectives.
If one (minority but large) side continues to accus the other of fraud and bad faith, the god faith essential to reform is destroyed.
No political party has a lock on ethics or morality, but both parties often use ethics and morals (or lack thereof) as blunt objects to beat their opponents about the head.
Election process is one such blunt object.
Since 2000, I observe that those most vocal in alleging election fraud are very selective in wich results they wnat “done over,” and which population segments they think are “under-represented” in results. Always.
Al Gore didn’t want a state-wide recount of over and under votes. He wanted a recount of one type of vote in 4 Democratic stronghold counties to gain only one objective: more votes for Gore, not a more “accurate” result.
And so it goes. It must be a tactic being discussed in those “smoke-filled rooms.”
I hate telephone surveys and hang up on them. Polls, at their best, rely on the feelings, opinions, and motivations of those polled towards who or what they perceive as either the polls subject, or its sponsors.
This is no different than media reports about “perceptions” about fact. Facts are facts, polling about perception of fact only tells you how naive, ill-informed, biased, or ignorant are those polled. Provided, of course, you know what the facts are. Which until election results, tallied in ar fair a manner as can reasonably be accomplished, are not known.
Polling in place of facts are a very poor substitute. 1-6% deviation? I’m surprised its not higher.
Maybe the data on the deviation has been manipulated…
the idea that the exit pollsters are monitoring themselves is supposed to give us comfort? Who watches the watchers?
Let’s review the bidding. The media pay for exit polling because they want something to report. They want to make early predictions. But exit polling doesn’t give good enough answers for that. Exit polling only show’s who’s winning early when it isn’t close. The states that are already completely locked up will give an answer you can depend on, and the media can report that. The places where it’s close, the results aren’t good enough for the media to make claims on. Estimate 20 close states and one of them will be wrong, and the media don’t want to be wrong. So I’m not clear what good it is. But it’s traditional, and it’s news. It’s news because the media do it.
They say they also get to look at things the elections themselves won’t show. They can say who african-americans are voting for, and women, and some things like that. More news.
Exit polls weren’t designed to catch election fraud, but the claim has been repeatedly made that they’re good at that. I guess. If the party in power actually gets 40% of the vote but claims 70%, exit polling will show it. But if the exit polls say El Presidente got 48% but the election says he got 50.73% (which actually happened in 2004) it doesn’t prove anything. Maybe more republican voters refused to talk to the pollsters. Maybe not. No possible way to tell. Exit polls won’t detect fraud in close elections.
So OK, if you can do election fraud and win, what do you get? A lot of high-level patronage jobs. A lot of great contracts. The chance to set policy in ways that may benefit your own kind for generations. If you do exit-polling fraud what do you get? You get a bad reputation. Your skills as a pollster are put in question. It hits you in your pocketbook. Would somebody bother to bribe you to do exit-polling fraud, assuming you’re bribable? So after they lose they can use it to complain? Not likely. It would make more sense to bribe exit-pollers to give results that fit the results of election fraud….
I say the exit poll results are not definitive. They’re compatible with a rather small amount of fraud, no more than 3% or so. (Which is often more than enough to swing a US election.) They’re also compatible with a small unverifiable persistent bias in polling methods. No way to tell which is going on. Maybe both. People who say they know which is going on are showing their bias. (Unless they are involved in election fraud. Then they know, but they don’t tell.) 😉
Great discussion!
Expanded commentary from my comments here, over at http://dadmanly.bl*gspot.com/2006/06/chutzpah.html.
J Thomas… If election poll results are not definitive – and they can never be, for a number of well-known reasons – then they can’t be an argument for anything. Certainly, the dishonesty one sees in articles like the one at Salon, and in the whole 2000 election trope (complete with al Gore’s methods and a MSM recount in detail of original materials that showed Gore still losing) tips me toward Thorley’s conclusion in #15.
For people who are interested in creating a better system, there are certainly ways to tighten it. Requiring photo ID while voting, for instance. Making federal elections a federal responsibility (as we do in Canada), to remove a lot of the unevenness in the system. Etc.
My recommendation would be to pay attention only to concrete reform proposals, debate those on the merits, and ignore the rest as either partisan blather or outright lies.
I’ll add that if one side or the other manages to convince itself that elections are rigged whenever they don’t win, over the longer term you end up with one of two possibilities, which eventually converge on the second:
[1] Cheating and fraud ramp up on both sides in an action-reaction cycle, aided by laws passed to aid different modes of same depending on which party is in power; or
[2] The breakdown of democratic legitimacy and its eventual replacement by the alternatives. At which point, it’s a good idea to know how to shoot.
You know I think we highlighted the flaws with “election reform” in the last discussion, at least from a Presidential Election standpoint.
There is no “right to vote” in the Constitution, only the guarantee of franchise for minorities and women.
That said, the Constitution provides the major hurdle for nationwide election reform since voting standards are left to the states. This negates the mandate of a Federal Voter ID, which in my view would clear up much of the voting irregularities.
Since each state is tasked with determining their electoral votes, the solution cannot come from the Feds, it is the purview of the states, and attempting to get all 50 states on the same page for something as complex as voting standards has about as much chance of succeeding as Ann Coulter and Al Franken swapping political affiliations, and then subsequently getting married and taking over for Jim and Tammy Faye Baker’s failed religious kingdom.
Al Gore didn’t want a state-wide recount of over and under votes. He wanted a recount of one type of vote in 4 Democratic stronghold counties to gain only one objective: more votes for Gore, not a more “accurate” result.
Dadmanly, this is a fine example of the problem. Gore asked for a recount where he saw obvious problems. The answer was that since he hadn’t asked for precisely the right thing, he’d get nothing. Some florida official should have taken the initiative to call for the recount that should have been done, and laid the issue to rest. Was it a flaw in florida’s state government that kept it from happening? Or partisan politics?
I see the argument repeatedly that Gore prevented an accurate recount. This is an argument created by political partisans. Given the known failures in the voting system in florida, it should not have been Gore’s responsibility to come up with the correct procedure to correct them.
Somehow we’ve let the partisans take over the system, and they’re tearing it apart. We shouldn’t be choosing between the partisans on one side versus the partisans on the other side. Sheesh.
Since 2000, I observe that those most vocal in alleging election fraud are very selective in wich results they wnat “done over,” and which population segments they think are “under-represented” in results. Always.
gnore the moonbats who say all republicans suck. Stop being their opposite number who says all democrats suck. The answer is, don’t suck. Get together with everybody who’s willing to do a cooperative effort, and do it. Provided there are enough of us. If the partisans are too strong then better pack it in. Emigrate, or learn to shoot and find out which of your neighbors are willing to form an apolitical militia.
And the Democrat vote has been below the exit polls every election since 1990. What conclusion should we draw from this?
Easy. Institutional bias in exit poll sampling. Keep making the same errors in how the sample is obtained, and of course you get the same bias on a consistent basis.
I would guess the biggest problem with exit polling is the population of pollsters. Think about it: who is more likely to blow a day getting in people’s faces and accepting the consequences? Hint: it probably isn’t your average small business owner, soccer mom, middle manager, or even union stiff. It’s probably some idealistic young college student or recent college grad wanting to make a positive contribution to the system.
That isn’t bad in-and-of-itself, but couple the likely demographic skew with the practical impossibility of getting this army of workers trained in anything but the most rudimentary polling techniques, and who would be surprised if there was a consistent and detectable bias towards friendly in the crowds?
As anyone who has been through Stats 101 can quote at you, correlation does not imply causation. If you’re going to lean so heavily on third-party statistical analysis, it would be best not to forget the basics.
If election poll results are not definitive – and they can never be, for a number of well-known reasons – then they can’t be an argument for anything.
They can’t be a *proof* of anything. Unfortunately, they are an argument. The results are consistent with voter fraud. It might have turned out otherwise, it might have turned out mostly inconsistent with voter fraud. But it didn’t.
If you’re predisposed to believe the election(s) stolen, then it’s one more brick in the wall. So democrats can use it to inflame their base, like republicans do with gay marriage or abortion. They’ll never convince their enemies they’re right, but they aren’t interested in that conversation. Republican partisans can get outraged about democrats doing that while democrats get outraged about republicans trying to rev up their own base with 9/11.
Yeah, it’s just one more brick in the wall.
My recommendation would be to pay attention only to concrete reform proposals, debate those on the merits, and ignore the rest as either partisan blather or outright lies.
Good! So what we’re doing in this thread is studiously ignoring RFK Jr’s partisan blather and outright lies…. 😉
“And the Democrat vote has been below the exit polls every election since 1990. What conclusion should we draw from this?”
Easy. Institutional bias in exit poll sampling. Keep making the same errors in how the sample is obtained, and of course you get the same bias on a consistent basis.
Bzzzt. Wrong answer.
That’s one plausible hypothesis. It’s consistent with the data. Concluding that it’s true is premature.
There is some evidence that you may be right. There was a tendency for the effect to show up more for younger interviewers. That fits. But then, I haven’t seen a study of where the younger interviewers tended to be. There could be reasons why the precincts that got more voter fraud might also tend to get younger interviewers. As you point out, correlation does not imply causation.
And the Democrat vote has been below the exit polls every election since 1990. What conclusion should we draw from this?
Question: where is this data? Can someone provide a link? Are data points available for earlier years?
I followed the 2000 election aftermath fairly closely and remember seeing some good analysis of spoiled ballots by precinct. I would be interested in seeing if the higher spoilage I recall the Democrats having in the key precincts might tie to a nationwide trend.
Possible hypothesis: Democrats rely more heavily on urban poor, who may spoil ballots more. This could account for the statistical anomaly of exit polls that consistently show higher Democratic results than the election count.
J Thomas, I don’t know if this was your intention, but I detect moving goalposts:
You first ask us to draw the conclusion that elections have been rigged based on consistently higher Democrat results at exit polls, then say that my “plausible hypothis” can’t be considered conclusive, only consistent with the data.
Okay, I’d buy point (b) but that kind of torpedos point (a), doesn’t it?
“…That’s one plausible hypothesis. It’s consistent with the data. Concluding that it’s true is premature…”
There are billions of plausible hypotheses. If you’re not using Occam and his razor you can expound at length about various permutations. Elimiate un-needed variables. Please. Maybe Elvis shot JR? Maybe poor footwear kept republicans from standing around long enough to talk to the exit poller. Maybe the process itself of exit polling presupposes statistical clusters that might not exist in reality. Lots of maybes. You have your favorite, I have mine.
This is a senseless argument, with all due respect. I have been following it, and it sounds like somebody who has already reached a conculsion desperately trying to at least inject uncertainty in a fairly routine procedure. It reminds me of the OJ trial. It’s an election. They’re managed locally. Ergo, they will be performed with thousands of different quality results. Stringing together a national pattern from the noise of thousands of different amateur election processes is so fantastically stupid it beggers the imagination.
Perhaps I should get over my shyness. I’m not sure you’re getting me here.
Maybe I need an assertiveness course.
I have read with interest the comments above.
As a side issue I would like to point out that the election system put in place for the 2000 election in Florida was implemented by a Democrat. That alternating ballot idea they used for the Presidential candidates was too confusing for their target voters. Too Bad
You first ask us to draw the conclusion that elections have been rigged based on consistently higher Democrat results at exit polls, then say that my “plausible hypothis” can’t be considered conclusive, only consistent with the data.
Mark, I didn’t ask you to draw the conclusion that the elections were rigged. You read that in. The conclusion I drew myself in that post was:
I see no reason to choose between those alternatives, yet. Not enough information to base a choice on, beyond prejudice.
I don’t understand why people persist in choosing sides about things they don’t know. (Except that some of them are partisans who want to get other people to believe-without-evidence the things that are good for their side.)
Why not wait and see? If actual evidence about vote fraud turns up, it may be compatible with this data. (Depending on the pattern of falsification. Or it might wind up something that isn’t compatible with the exit-polling bias.) On the other hand, if the exit-polling guys hire more retired people to do exit-polling and train them carefully and the bias goes away, that says something too. Why prejudge it?
There were claims leaked results in 2004 could influence the election. Instead of people refusing to talk to exit pollers some could seek them out or leave the polling place more than once, pretending to have voted. They could even call their friends, have them do the same. Vote early, vote often, no risk involved.
Alan, that could happen but it doesn’t make sense. It would take somebody doing it at a lot of polling stations. The pollsters were supposed to interview every fifth voter, So imagine it’s you scamming them. You have to repeatedly sneak in and come back out and arrange to be the 5th guy in line behind the one who’s getting interviewed. You do this enough that you’re about 6% of the people polled. There’s somebody like you at the majority of polling stations. And the result is — the media report that the Democrat is doing better than he really is, and you help rally the Republican vote. But then sometimes things happen that don’t make sense.
Here’s one that makes more sense to me. A little old lady who voted Republican gets accosted by a young man who wants to know how she voted. She thinks he’s a Democrat. Rather than cause any unpleasantness she says she voted Democrat.
Just as an academic exercise, avoiding the math, let’s work out an example. A precinct where 500 voters vote Democrat and 500 vote Republican. One possibility is that through some kind of fraud, 30 votes for democrats get switched to republican. The result is the vote is 470:530. The exit count is 105:95 or maybe as far off as 85:115. There’s a good chance it won’t show up as significantly different from the real poll, but it’s far more likely to be off in the Republican direction than the Democrat.
Now suppose that 6% of the republican voters refuse to talk to the exit-pollsters, but all the democrats do. That gives the same result.
Now suppose that the same number of democrats as republicans refuse to answer, but 3% of the republican voters lie and say they voted democrat and none of the democrats lie. Same result.
The exit-poll data can’t tell the difference among explanations. And it’s statistically significant but the more you break the data down into categories the harder it is to get significant results. Early on some people claimed that the touch-screen votes were far more biased than the rest. That would be serious if true. Some other people claimed the data they had didn’t give a significant result there — which doesn’t mean it’s wrong, it means they didn’t have enough data to tell. I haven’t tracked the most recent claims about that. However, something like that would probably be necessary if the exit poll results came from fraud. If a whole lot of precincts had republicans doing a little bit of vote fraud — switching a few percent of democrat votes to republican, say — then they couldn’t keep it a secret. Some would talk out of self-importance or bragging etc. Some honest republicans would find out. Someone would tell the media. It would have to be a method that most of the republican election workers didn’t know about. (Unless you want to demonise republicans.)
Touchscreen voting fits that. One person puts the wrong smartcard in and the machine is reprogrammed. The one who puts in the wrong smartcard doesn’t even have to know he’s doing anything special. With a proprietary system, maybe all it takes is one crooked employee to result in massive election fraud.
And with a system like that, you could do real sophisticated election fraud. Throw it to one side when it’s real close, and give enough extra votes to the other side to make up for it, in precincts where they’re already winning. Very hard to detect statistically.
I don’t know whether computerised voting machine fraud is compatible with the data. I’m not real interested in that. If it somehow got proven then all we’d get out of it would be a lot of outrage and a witchhunt for the perps and such. Much more important to fix the system than figure out whether the broken system has been getting exploited.