# Survey Response Error: Variance v. Bias (Another Election Thread) **URL:** \ **Category:** Great Debates\ **Created:** [November 11, 2000, 2:10pm UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787 "2000-11-11T14:10:30Z")\ **Posts on this page:** 10\ **Page:** 7
**Author:** ![jshore](https://avatars.discourse-cdn.com/v4/letter/j/b5e925/32.png) [@jshore](https://boards.straightdope.com/u/jshore)\ **Post date:** [December 5, 2000, 2:23am UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/121 "2000-12-05T02:23:12Z")
> [@](#): > > \*Originally posted by RTFirefly \* > > The [Miami Herald](http://www.herald.com/thispage.htm?content/archive/news/elect2000/decision/104268.htm) had some statisticians do a precinct-by-precinct analysis of the nonvotes in FL… Yeah, that’s a cool analysis. Not sure I’d say their numbers are “conservative” given their attempt to assign all the discarded ballots to a given candidate. But, I think it is obvious that one would be **very** hard-pressed to come up with any reasonable assumptions under which Bush would have won had all the ballots been recorded the way the voters intended. Admittedly, this is academic…But interesting! The problem with the Canadian study (and one before it involving schoolchildren) is that you are looking for a very small effect. After all, on the basis of the successful Buchanan votes, it looks like only about 1% of the PBC voters were confused into voting for him. And, even if you assume that, say, 12000 of the 19000 overvotes were also confused votes intended for Gore, you are still just talking about a 5% effect. So, like you said, they have to do a study involving a lot more than 15 people to get good results. ---
**Author:** ![Monster104](https://avatars.discourse-cdn.com/v4/letter/m/c6cbf5/32.png) [@Monster104](https://boards.straightdope.com/u/Monster104)\ **Post date:** [December 5, 2000, 2:44am UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/122 "2000-12-05T02:44:50Z")
Considering that people don’t vote according to the linear predictions of statistics, I’d have to say that saying “Gore would’ve won by 23,000 votes” doesn’t fly. Also, what do you have to say about the fact that most of Gore’s votes would have come from the 4 Gore-heavy counties, and the fact that the heaviest democratic precincts were counted first in the sample hand recounts of those counties (Especially in Miami-Dade)? The statistics the Miami Herald were made using those numbers (On their assumption that everyone else in each county in Florida would vote to the same ratios compared to sample recounts), so predicting how many votes Gore would have recieved by using biased sample recount numbers would be very innaccurate. ---
**Author:** ![jshore](https://avatars.discourse-cdn.com/v4/letter/j/b5e925/32.png) [@jshore](https://boards.straightdope.com/u/jshore)\ **Post date:** [December 5, 2000, 3:01am UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/123 "2000-12-05T03:01:45Z")
> [@](#): > > \*Originally posted by Monster104 \* > > Also, what do you have to say about the fact that most of Gore’s votes would have come from the 4 Gore-heavy counties, and the fact that the heaviest democratic precincts were counted first in the sample hand recounts of those counties (Especially in Miami-Dade)? The statistics the Miami Herald were made using those numbers (On their assumption that everyone else in each county in Florida would vote to the same ratios compared to sample recounts), so predicting how many votes Gore would have recieved by using biased sample recount numbers would be very innaccurate. What the f- are you talking about **Monster104**? Me thinks you need to go back and read the article again. The statistics used had absolutely NOTHING to do with the sample recounts. What they did was went precinct-by-precinct through the whole state and assigned the rejected ballots to one candidate or the other on the basis of how the unrejected ballots broke down in that precinct. Sorry…you lose…Would you like to play again? ---
**Author:** ![ZenBeam](https://avatars.discourse-cdn.com/v4/letter/z/3ab097/32.png) [@ZenBeam](https://boards.straightdope.com/u/ZenBeam)\ **Post date:** [December 5, 2000, 6:35pm UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/124 "2000-12-05T18:35:46Z")
Overall, I personally wouldn’t trust the final answer they got. Two serious flaws (which they acknowledge in the article) are that they assume all voters within a precinct are equally likely to make a mistake, and that all the under- and over-votes were unintentional. One of those assumptions helps Bush, and the other helps Gore. Still, it’s a starting point. I would expect that the criticisms could be addressed, and will be at some point. A refined study could estimate how likely republicans and democrats were to accidentally under- and over-vote, and also how likely people were to intentionally under- and over-vote. I did find it interesting that in the study, Duval gives **Gore** an additional 1000 or so votes. ---
**Author:** ![nebuli](https://avatars.discourse-cdn.com/v4/letter/n/0ea827/32.png) [@nebuli](https://boards.straightdope.com/u/nebuli)\ **Post date:** [December 5, 2000, 11:34pm UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/125 "2000-12-05T23:34:50Z")
> [@](#): > > I did find it interesting that in the study, Duval gives Gore an additional 1000 or so votes. According to an earlier article in the New York Times, about 16,000 of Duval’s 26,000 overvotes were from _very_ heavily Democratic, inner-city Jacksonville precincts. In some precincts 30% of the ballots were doublepunched. ---
**Author:** ![jshore](https://avatars.discourse-cdn.com/v4/letter/j/b5e925/32.png) [@jshore](https://boards.straightdope.com/u/jshore)\ **Post date:** [December 6, 2000, 3:11am UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/126 "2000-12-06T03:11:09Z")
> [@](#): > > \*Originally posted by ZenBeam \* > Overall, I personally wouldn’t trust the final answer they got. Two serious flaws (which they acknowledge in the article) are that they assume all voters within a precinct are equally likely to make a mistake, and that all the under- and over-votes were unintentional. Well, as for the second flaw, I agree this assumption is extreme. However, the large differences in under- and over-votes between counties using different voting systems seems to suggest that a large fraction of these were unintentional…And, you don’t need the fraction to be all that large to tip things in Gore’s direction. ---
**Author:** ![Monster104](https://avatars.discourse-cdn.com/v4/letter/m/c6cbf5/32.png) [@Monster104](https://boards.straightdope.com/u/Monster104)\ **Post date:** [December 6, 2000, 5:55am UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/127 "2000-12-06T05:55:00Z")
> [@](#): > > **What the f- are you talking about Monster104?** Oops. I did read the article. Twice, in fact. However, after I read it I was unable to post immedietly. I got this article mixed up with another statistic based just on the sample recounts in the contested counties. Sorry. ---
**Author:** ![RTFirefly](https://avatars.discourse-cdn.com/v4/letter/r/c77e96/32.png) [@RTFirefly](https://boards.straightdope.com/u/RTFirefly)\ **Post date:** [December 6, 2000, 5:00pm UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/128 "2000-12-06T17:00:07Z")
> [@](#): > > \*Originally posted by jshore \* > \*\* > > > [@](#): > > > > \*Originally posted by ZenBeam \* > > Overall, I personally wouldn’t trust the final answer they got. Two serious flaws (which they acknowledge in the article) are that they assume all voters within a precinct are equally likely to make a mistake, and that all the under- and over-votes were unintentional. > > Well, as for the second flaw, I agree this assumption is extreme. However, the large differences in under- and over-votes between counties using different voting systems seems to suggest that a large fraction of these were unintentional…And, you don’t need the fraction to be all that large to tip things in Gore’s direction. \*\* Since the Republicans on this board have basically been saying that it’s only the Dems down there who were so dumb as to fill out their ballot incorrectly, the position that Republican voters are _more_ likely to make errors isn’t being argued by anyone here. So saying Democratic and Republican voters are equally likely to make a mistake on their ballot is giving the GOP the benefit of the doubt. With optically scanned ballots, I understand that undervoting and overvoting are both a minuscule fraction of what they are for punched ballots. So it seems fair to assume that the vast majority of under/overvoting with the punched ballots is unintentional, an artifact of the voting system. Artifacts of voting systems shouldn’t determine elections; the people should. ---
**Author:** ![jshore](https://avatars.discourse-cdn.com/v4/letter/j/b5e925/32.png) [@jshore](https://boards.straightdope.com/u/jshore)\ **Post date:** [December 8, 2000, 4:03am UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/129 "2000-12-08T04:03:23Z")
> [@](#): > > \*Originally posted by Monster104 \* > > Oops. I did read the article. Twice, in fact. However, after I read it I was unable to post immedietly. I got this article mixed up with another statistic based just on the sample recounts in the contested counties. Sorry. No problem…And sorry for being a little harsh there. I guess I just get a little testy about these things sometimes! ---
**Author:** ![RTFirefly](https://avatars.discourse-cdn.com/v4/letter/r/c77e96/32.png) [@RTFirefly](https://boards.straightdope.com/u/RTFirefly)\ **Post date:** [January 27, 2001, 2:21pm UTC](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787/130 "2001-01-27T14:21:27Z")
It seems that eight of the FL counties that used punch-card ballots (including Palm Beach County) have computer records showing the various combinations of ‘overvotes’ - the ballots that were punched more than once. The Washington Post did a [story](http://washingtonpost.com/wp-dyn/articles/A53804-2001Jan26.html) about it in this morning’s paper. In the eight counties for which the Post was able to obtain records, Gore’s hole was punched about 46,000 times, and Bush’s was only punched 17,000 times. The article didn’t include detailed county-by-county breakdowns (I’ve already emailed the author of the article, asking if such breakdowns will be forthcoming), but apparently over 7200 of the overvotes in Palm Beach County involved Gore-Buchanan or Gore-McReynolds, the diagonally adjacent candidates to Gore on the butterfly ballot. The records also show how the voters voted in the race for FL’s US Senate seat. The Gore-Buchanan and Gore-McReynolds overvoters favored Democrat Bill Nelson over Republican Bill McCollum, 6645-632. Read the article for more details, but on the whole, it solidifies the case that Bush’s win in Florida was an artifact of the voting systems, rather than a reflection of the intent of the voters. [Previous page](https://boards.straightdope.com/t/survey-response-error-variance-v-bias-another-election-thread/40787.md?page=6)