Kosuke Imai and Gary King have just published an “article”:http://gking.harvard.edu/files/ballots.pdf in _Perspectives on Politics_ that’s controversial – but quite important. King is a noted methodologist, who’s made very considerable strides in the application of models of ecological inference in the social sciences. On behalf of the _New York Times_, Imai and King applied their methodology to the disputed election results in Florida. The results are eye-opening.
(1) If overseas absentee ballots had not been counted illegally, there is a very small chance that Gore would have won the election outright. In Imai and King’s account (where they admit that there is some room for alternative interpretation), the chance that Gore actually should have won the election on this alone is around 0.2%.
(2) More to the point: if the recounted votes in Miami Dade and Palm Beach had not been rejected by Katherine Harris, Gore would have won with 82% probability. In Imai and King’s words
bq. To put it one way, the massive differences in probabilities from 0.002 to 0.82 for a Gore victory were due to the decisions made by Katherine Harris.
Finally, and most damningly, Imai and King find “strong and independent support” (albeit indirect) for the proposition that:
bq. the propensity of local election officials to violate the law and accept bad ballots was substantially greater in counties where Bush strategists believed there was more absentee ballot support for their candidate and tried to convince election officials to accept bad ballots.
One should note some caveats – ecological models are still as much art as science. Still, Imai and King have done their homework – they present a strong body of evidence to support the contention that Republican efforts to manipulate the count had a decisive impact in Florida in 2000.
{ 26 comments }
gm 09.10.04 at 6:26 am
Although some may find your contention arguable let’s posit, for a moment, that it’s accurate. Is it not reasonable for American voters to anticipate a situation in which the usually-reliable mechanisms of democracy break down? The founding fathers did an exemplary job of anticipating possible problems, but even their best efforts have a limit. Viewed in this light, the fiasco in Florida should move us not to despair, but to enthusiasm for refining a system that has, so far, served us very well.
Brett Bellmore 09.10.04 at 11:09 am
“If overseas absentee ballots had not been counted illegally”
Funny, I could have sworn the courts endorsed that. Are we ceasing to be legal realists, or something? If so, I can think of a lot of technically illegal things the courts let Democrats do in that election…
You can play what if games all you like. In fact, I encourage you to, because if you persist in thinking you really won elections you lost, you won’t concentrate on trying to ACTUALLY win them.
Tom T. 09.10.04 at 1:34 pm
A “noted methodologist”? It’s amusing to contemplate the telephone call home from college:
__________________________________
“Mom, I’ve decided to become a methodologist.
…
What?
…
No, Mom, that’s a pathologist; no, I’ll be dropping out of pre-med.
…
Mom, don’t start, just listen! Methodology is a science, a social science.
…
No, not that kind of social.
…
What? Oh, don’t start that again; I told you, I just haven’t met the right girl yet. Listen, methodology is sort of like systems analysis.
…
You know, an analyst who studies systems.
…
Systems! Any kind of systems!
…
Yes, I’m still seeing my analyst — what does that have to do with… Yes, he thinks I should do what I know is right for me. No, no, I’ve told you before, you can’t talk to him. Listen, methodology is … well, I’ll be studying how people do things.
…
What? Any kind of things!
…
Well, like ecology. And voting rights. And how ecology relates to voting rights!
…
No, I’m serious! Mom, would you just put Dad on the phone? I need some cash.”
y81 09.10.04 at 2:41 pm
I think John Allen Paulos summarized the situation best, early in the Florida recount imbroglio, when he wrote that the “true” margin of victory was clearly smaller than the margin of error in our measuring instruments (the polling process). That being the case, the result was clearly going to depend on which side did a better job of managing (or manipulating, if you prefer) the process, e.g., by convincing officials in each county to count some ballots and reject others, by making broad popular appeals in some cases and technical legal arguments in others, in short by utilizing the normal tools of politics. Both sides, I am sure, did their best.
Ken Houghton 09.10.04 at 3:29 pm
brett b:
Nope; the courts would have had to decide on the late–and in some cases VERY late (as in postmarked after Election Day)–ballots if the Dems had decided to challenge them.
Instead, they decided not to use every legal means available, either on the state or the county level. Which rather gives the lie to y81’s o/p/t/i/m/i/s/m/ strict economist’s perspective.
Kevin Carson 09.10.04 at 3:41 pm
y81,
I agree. The 2000 election should tell us that, despite what the goo-goos say, if an election is ever close enough for one vote to “count,” it almost certainly won’t. I believe that’s what Jesse Walker over at Reason said when it was all over.
Of course, the real political battle was won before the election even began, when Harris purged the voting rolls of “felons.” That surpassed by a couple orders of magnitude any of the nickel and dime stuff done by either side during the actual election process.
And Karl Rove had a history of trying exactly the same thing in Texas, so we know it wasn’t just an “accident.” Not that Bush’s cronies could have orchestrated the entire 2000 election. It was sheer luck that the electoral college vote was so close that the difference was made by a state where Bush’s brother could help things along.
Thomas 09.10.04 at 4:29 pm
When one starts with flawed premises, one reaches flawed conclusions. That’s true even if the methodology is interesting.
The fact is, the Times report on which Imai and King rely for the legal conclusion is not accurate. That fact was established long before the Times published its article. (Imai and King apparently rely on the Times article because it hasn’t been widely disputed. Surely they can do better, methodologically speaking, than that.) For example, there was no requirement, despite their assertion, that all military ballots had to be postmarked. Here’s how the Jurist website describes this question: “On December 8[,2000], U.S. District Judge Lacey Collier ruled in Bush v. Hillsborough County Canvassing Board that it is improper to exclude an overseas absentee ballot solely for lack of a postmark (and solely for lack of a previously filed absentee ballot application).”
Dr. Weevil 09.10.04 at 5:45 pm
At the time, I recall reading that some foreign countries (e.g. Costa Rica) do not postmark their mail. Granted that lack of dated postmarks seriously complicates the effort to determine which absentee ballots are valid and which are not, do you really want to tell American citizens that if they move to the wrong country they lose their right to vote?
joem 09.10.04 at 5:52 pm
The main difference, of course, is that with absentee ballots, there is no question that it represents an ACTUAL vote that was really intended. And given the Florida court’s ruling on technicalities like the military postmarking system, there was absolutely no reason to literally throw out ballots with real votes on them.
But for the county recounts, the “votes” in the final results often represented after-the-fact speculation by partisan officials squinting at dimples. It’s not as if these recounts resulted in an official saying, “Gee, here’s a stack of ballots with indisputable votes that we just plain never counted before!” Instead, as we all saw on TV at the time, the recounts involved Democrat-controlled panels taking ballots that had already been counted at least once, and then holding them up to the light, and then announcing (by a party-line vote that coincidentally favored the Democrats) that they had suddenly discovered a brand new vote.
james 09.10.04 at 5:53 pm
I have read that voting across the country contains a +-5% margin of error. This includes fraud, miscounts, voter error, etc. In Florida, that 5% mattered.
Brett Bellmore 09.10.04 at 6:58 pm
“Of course, the real political battle was won before the election even began, when Harris purged the voting rolls of “felons.—
As you should know, Harris didn’t have the power to purge voter rolls. All she could do was provide a list of suspect names to local elections officials, who were supposed to investigate the people on the list, and determine which of them really were felons, and should be purged.
As it happens, many of them did nothing of the sort, and between officials who purged without investigating, and officials who simply threw the list away, it’s hard to say which numbered more: Non-felons improperly purged, or felons improperly non-purged.
What IS easy to say, is that the former could vote by provisional ballot, while there existed no mechanism to remove the laters’ illegal votes from the tally.
jet 09.10.04 at 8:04 pm
Granted I haven’t read the study, but I find the conclusions highly improbable. Several major news organizations recounted the votes after the 2000 election, and even using the rules the Democrats sued to impose, Gore lost in every single one of those recounts.
So if I wasn’t so sure this was a partisan bunch of hacks doing the research, I’d be interested in seeing how the decisions made by Katherine Harris affected anything in spite of these later findings.
I find the continued groping of this issue, in spite of overwhelming evidence of a conclusion, almost pathological. Get a frigg’n grip people.
Katherine 09.10.04 at 9:01 pm
“Several major news organizations recounted the votes after the 2000 election, and even using the rules the Democrats sued to impose, Gore lost in every single one of those recounts.”
I am 99% sure this is false. If I remember correctly, Gore would have lost in the county-by-county recounts he originally proposed, but would have won in a majority of the statewide recount scenarios. It depended on the standard you chose, obviously–the NY Times had a website where you could pick a recount standard and run the results. Gore won in almost all scenarios where overvotes where you could determine the clear intent of the voter were counted (e.g. “Gore” both punched and written on the line for a write in candidate) and lost in most scenarios where they were not. And I believe he did worse if chads that were dimpled and/or hanging (detatched at one corner or two corners) were counted, than he did if only chads were dangling (detatched at three corners) or fully detatched were counted.
I think all of the above assumed no change in the counting of absentee ballots. And that is all from memory, so I probably got some of the particulars wrong, but that is my best recollection.
The difference between the number of Buchanan votes in Palm Beach County and the next highest county easily exceeded Bush’s margin of victory in any recount scenario–by at least 100%, I think. The number of felons wrongfully excluded from voting exceeded Bush’s margin of victory in any recount scenario by several orders of magnitude.
(No, I never had an unhealthy obsession with this. Why do you ask?)
But there was no legitimate way to remedy this error after election day, no legitimate way to remedythe instrinsic stupidity of the guy who wins the popular vote losing the election, and no legitimate way to remedy one of the truly lousiest Supreme Court decisions I have ever read. We should have changed the electoral college, but there was never a prayer of that and there probably never will be. No one will care unless the popular vote winner loses the electoral college, but if he does the electoral vote winner’s party will not agree to a change because it’s a tacit admission that they should have lost. Also, the same low population states that are overrepresented in the electoral college are also even more overrepresented in the constitutional amendment process.
So in a way agree with the “get over it” crowd, but it’s not because they’re right about Bush being justly and legitimately elected–there’s just no use in this except to add another to the long list of reasons to work to defeat Bush this time. (It wouldn’t hurt to vote out Katherine Harris, either.)
The Navigator 09.10.04 at 9:07 pm
Jet,
That’s simply false. If a recount had been done that included counting overvotes, Gore would have won, as Mickey Kaus explains here.
CNN also reports that use of the Palm Beach County standard would have resulted in a slim Gore win – see here.
Katherine 09.10.04 at 9:47 pm
Here is the NY Times’ little vote counting machine.
According to this little thingy, which I can’t believe still exists, if the Florida Supreme Court ordered recount directed by county officials had gone forward, and the Supreme Court done nothing, Bush would have won. (Kaus says that’s incorrect though, because the Florida judge in charge of the case says he “would not have ignored” overvotes.)
If Stephen Breyer and David Souter had been able to get O’Connor or Kennedy to sign onto the view that there was an equal protection issue but the proper remedy was to order a statewide recount by a universal standard–Stevens and Ginsburg probably would have agreed too based on their dissents–it depends on what statewide standard would have been chosen.
There are 24 possible standards for a statewide recount on the Times website. Bush wins in 12 scenarios and Gore wins in 12 scenarios. Gore’s average margin of victory is slightly higher. I was correct in stating that the single most important factor was to count only properly marked Op-Scan ballots, or all Op-Scan ballots where you could clearly discern the voters’ intent (improperly marked ballots where the voters’ intent is clear are called “overvotes” as shorthand.) Gore wins 10 out of 12 scenarios where overvotes are counted. Bush wins 10 of 12 scenarios where overvotes are not counted. Since Florida law says that the standard is the clear intent of the voter, it seems like you should definitely count overvotes as opposed to the much more debatable issue of chads, but who knows what standard actually would have been chosen.
So if the Supreme Court had done nothing Bush would probably have won, but if had intervened without writing the truly indefensible section about the remedy, Gore would have had somewhat better than even odds won.
And there is no way to know what a Democratic or, say, human secretary of state would have done if they held the office instead of Katherine Harris, so it’s impossible to tell whether her partisan approach to the recount was decisive.
All of this is assuming the NY Times correctly categorized every vote, and there was no change in the way absentee ballots were counted.
katherine 09.10.04 at 10:03 pm
(sorry, to head off a tiresome liberal media bias debate: it was an independent research group hired by a bunch of newspapers, not NY Times reporters, who examined and categorized each ballot.)
(and all of this is living in the past, but at least it’s four years ago instead of 30….)
katherine 09.10.04 at 10:18 pm
Final note: if you click on the right tab on the NY Times vote counter, they estimate the number of votes probably lost to each candidate due to lousy ballot design….the net result is -8512 votes for Gore, which is about 16x Bush’s actual margin of victory and about 16x the largest margin of victory for Bush based on any scenario in the Times calculator. Obviously, it would be totally and blatantly illegal to try to add those votes back in, so the only function of this information is to give Democrats ulcers and convince them God hates them. Mark your ballots carefully in November, people!!!
adam scales 09.10.04 at 10:25 pm
I agree with the overall tenor of the comments here – under any plausible recounting system, it’s still extremely close, and the margin of victory likely less than the margin of error. One must also consider the certainty of a Bush win in the House if Florida’s electoral votes remained too contested to allocate prior to the meeting of the EC.
One caveat on moving to a popular vote. Had this been in effect in 2000, there is no reason whatsover to believe the overall count would have been the same. The EC likely depresses turnout in uncontested states. It certainly depresses campaiging behavior in those states. For example, neither Bush nor Gore spent much time in New York or Texas – becaue the winner-take-all system meant this was time and money wasted. I know many highly educated and politically aware people who wonder whether they will bother to vote in obvious “Red” or “Blue” states.
This is not an argument for retaining the EC, but a caveat that the EC alters voter behavior. Had direct election been in place in 2000, Bush might have been ahead by 500,000. We can’t know.
Jet 09.11.04 at 2:54 am
And that, ladies and gentlemen, is why I love this blog. This thread has to be one of the most rational commentaries on the Florida 2000 fiasco I’ve seen online from the other team.
We can now stop debasing the debate with pointless slander about illegitimate presidents and move on to improving a highly flawed voting system.
Mike Perry 09.11.04 at 4:28 am
I quote: “More to the point: if the recounted votes in Miami Dade and Palm Beach had not been rejected by Katherine Harris, Gore would have won with 82% probability.”
Yeah and if you limited yourself to “recounted votes” in the Florida panhandle, Bush would have won with 100% probability.
What Gore wanted to do was clearly illegal. As the media-sponsored recount of hanging chads showed, the lost votes were a random event. It was a close race. Only count them in Democratic counties and Gore wins. Only count them in Republican counties and Bush wins.
What matters is that the Supreme Court (including Court liberals) said what Gore wanted to do, recounting only in heavily Democratic areas, was clearly unconstitutional. Recount the same everywhere or not at all.
We went with not at all and Bush won. And as the media-sponsored chad recount shows, Bush would have also won if the recount was done statewide. So in any fair scenario, Bush wins.
There’s an additional factor that often gets ignored. The media predictions that Bush had won Florida weren’t a mistake. But their calculations assumed an uninterrupted vote in the heavily Republican Florida panhandle which is on CST. When the networks called for Bush an hour before those polls closed, one study indicated that Bush lost between 10-15,000 votes as people heading home from work decided their votes weren’t needed. So if you hinge the election on what voters intended to do (as with butterfly ballots in Palm Beach), then Bush won quite handily.
What is most disturbing about this whole affair is the immature behavior of Gore and his kindred. Contrast Gore’s willingness to subject the country to enormous trauma so he could get into the White House to Nixon’s behavior in the also very close 1960 election. Immediately after that election there was evidence that the Mafia and stuffed the ballot box for JFK in Illinois and LBJ had fixed the election in Texas. (We now know for a fact tha JFK did make a deal with the Mafia and LBJ always cheated.) But Nixon rejected calls for him to fight the election results, saying it would be too trumatic for the country.
And that demonstrates that, for all his flaws, Nixon was ten times the leader Gore is. Gore ‘wants’ to be president too badly. And that is why the country is fortunate he lost.
–Mike Perry, Inkling blog , Seattle
lazy 09.11.04 at 5:23 am
If Al Gore had won his home state he would be President.
Katherine 09.11.04 at 5:57 am
Did you even read the rest of the thread? Almost everything you state is exactly wrong.
1) Actually, Gore wins in the most likely scenarios for a statewide recount and loses if the recount is restricted to Democratic counties.
2) Gore also does better if hanging and dimpled chads are omitted than if they are included.
3) Clearly illegal? The Florida Supreme Court thought not. The U.S. Supreme Court made an argument about the Equal Protection Clause that is totally inconsistent with their other opinions, particularly the justices that were in the majority, and explicitly instructed other courts to ignore it as a precedent in all other cases.
4) The Supreme Court also did not say “recount all the same way or not at all”. It was more along the lines of “recount the same way or not at all even though you didn’t vote the same way or count the same way the first time. And gee whiz, look at the time, it’s nearly December; we’re sure the Florida Supreme Court didn’t want to recount this late and we better not remand and ask them even though it’s their job and not ours to interpret state law because….we say so.”
5) The media called the state for GORE while the polls in the panhandle were still open, not for Bush.
6) They called the election for Gore 12 minutes before the polls closed in the pandhandle. It was an irresponsible thing to do, but I really wonder how many people were planning to leave for the polls with exactly 12 minutes to get there and get on line, when they saw the networks call and decided it wasn’t worth it.
One of the main studies purporting to show that Bush lost thousands of votes in the Panhandle is based entirely on a finding that turnout in the panhandle was several percentage points lower than in the rest of the state. The study simply assumes that turnout would have been equal to other counties in Florida if the networks had not called the election for Gore. Weak, weak, weak analysis.
The other study is by that beacon of integrity and disinterest, John Lott. I can’t find a copy of that study, which I am genuinely sorry about because it sounds like it could be real doozy. Lott seems to claim based on “historical turnout figures” that the networks’ 12 minutes early claim convinced thousands and thousands of Republicans to leave the polls dejectedly, but did not depress Democratic turnout at all.
Needless to say, neither of these is exactly up to the standards of projecting based on actual ballots cast.
MDP 09.11.04 at 10:42 am
kevin carson: “Of course, the real political battle was won before the election even began, when Harris purged the voting rolls of ‘felons.’ That surpassed by a couple orders of magnitude any of the nickel and dime stuff done by either side during the actual election process.”
Isn’t it strange that the Civil Rights Commission couldn’t find a single faux-felon who was actually prevented from voting?
Jim Miller 09.11.04 at 2:44 pm
I’ll have to look at the paper but I must say that, as described, it doesn’t sound impressive.
An analysis of who really won in Florida must consider the following points:
1. As investigations by two Democratic newspapers, the Palm Beach Post and the Miami Herald showed, thousands of felons (a group almost as Democratic as academics) voted illegally. Some had been left on the rolls by Democratic election boards that refused to obey the state law.
2. I looked through the tabulations of overseas ballots by county. Counties carried by Bush (and presumably with Republican election boards) accepted about 70 percent of the overseas ballots, in line with previous elections. (Florida has very complex rules for such ballots.) Counties carried by Gore accepted about 30 percent of those ballots. (They may have been breaking a federal law which protects military ballots.)
3. There is much evidence of fraud, nearly all by Democrats, in registration, voting, the first count, and the recounts. Many of the counties where there were problems — notably Palm Beach, Broward, and Miami-Dade — have long histories of vote fraud, all by Democrats.
The worst fraud problem was probably the voting by non-citizens, which was made easy by the 1993 “Motor Voter” Act — as some of its backers intended. This is so common that, according to an article I read in the last few weeks, Florida prosecutors don’t bother to file charges in such cases — even prosecutors who file other vote fraud cases.
Some will want more evidence than I have provided in this post. Fair enough. I’ll try to get a summary of the evidence out before the election on my site.
Jess 09.12.04 at 1:03 am
Just moved here, nice to have found this forum. Thank you.
Please see this ! I was in PBC furing the 2000 election saw first hand. My vote did not get counted.
Also please consider voting by mail this year, it is one way that this election will not be stolen again. I have a copy of this film if there may be anyone “on the fence” I would consoder loaning it out to you.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Unprecedented: The 2000 Presidential Election is the riveting story about the battle for the Presidency in Florida and the undermining of democracy in America. Filmmakers Richard Ray Pérez and Joan Sekler examine modern America’s most controversial political contest: the election of George W. Bush. What emerges is a disturbing picture of an election marred by suspicious irregularities, electoral injustices, and sinister voter purges in a state governed by the winning candidate’s brother. George W. Bush stole the presidency of the United States… and got away with it. ” …the movie highlights those on the front lines —from the African-Americans who were turned away from the polling booths for assorted reasons. In one memorable scene the filmmakers freeze-frame a ‘protest’ against the ballot recount, identifying participants as staff members of Republican elected officials.” –Elaine Dutka, Los Angeles Times
Dan Simon 09.12.04 at 6:46 am
Clearly illegal? The Florida Supreme Court thought not. The U.S. Supreme Court made an argument about the Equal Protection Clause that is totally inconsistent with their other opinions…
True. Then again, the Florida Supreme Court themselves made an argument about Florida electoral law that is totally inconsistent with their other opinions. In fact, just about everything bad you can say about the Bush v. Gore decision (and there’s an awful lot–much of it said in the dissent) can also be said of the Florida Supreme Court decision (and was said in that dissent).
More generally, as far as I can tell, every time a judge in the Florida election fiasco meddled in the process, all the way up to Bush v. Gore, he or she simply made things worse–more arbitrary, more inconsistent, and less democratic. And of course, each such decision only encouraged both sides to escalate their barrage of legal motions and sophistries.
The second biggest thing wrong with the American electoral process (and American democracy in general, for that matter) is the willingness of the judiciary to tamper capriciously with it in the name of various ad hoc constitutional or legal principles. The biggest thing wrong with the process is how few Americans see anything wrong with this sort of judicial evisceration of democratic sovereignty.
Comments on this entry are closed.