Talk:1981 Burlington mayoral election

Basic math error in vote analysis
I'm removing the vote analysis section because it's simply impossible. "When compared between Sanders and Paquette, voters who had a household income of $0 to $20,000 voted 67.6% in favor of Sanders and 32.4% in favor of Paquette. Voters with a household income of over $20,000 voted 61.2% in favor of Sanders and 38.8% in favor of Paquette." That's impossible, since that includes all voters, and Sanders did not get over 61% of the vote relative to Paquette. I don't know whether the error is in the source, or the error is in misquoting the source, but it just can't be. --GRuban (talk) 14:25, 21 December 2020 (UTC)
 * That said, it's a beautiful section, so if the errors can be corrected, I'd be all in favor of restoring. --GRuban (talk) 14:58, 21 December 2020 (UTC)
 * - please check the math as above. Yes, I know it only compares Sanders to Paquette. It still conflicts with the DYK hook that Sanders won by 10 votes. That's not 61.2% to 38.8% unless only 50 people voted. --GRuban (talk) 17:24, 21 December 2020 (UTC)
 * The irregularities with the data can be easily explain by one of two things. It is either due to it ignoring the other candidates and only looking at Sanders and Paquette. The irregularity could also be explained by its odd ending point of $20,000, which is over $30,000 less than the median income of Vermont residents in 1984 (which is as far back as the US Census Bureau goes) which means that poorer voters, who would be more likely to vote for Sanders, would be lumped in with the richer voters more likely to vote for Paquette. Jon698 (talk) 18:09, 21 December 2020 (UTC)
 * Basic math. Here is what your infobox says: Popular vote Sanders	4,330	Paquette. 4,320. There is just no way that is in any way between 61.2% or 67.6% in favor of Sanders. It doesn't matter where you put the ending point, if you include all the voters below and above it. Either your paragraph is wrong or your infobox is wrong. --GRuban (talk) 18:13, 21 December 2020 (UTC)
 * Agreed, as written it is impossible. I cannot access the paper to see if there was a typo, or a nuance of wording we didn't catch, without spending $10 (I just got my JSTOR account!) . It seems very clear from all other sources that the 10 vote margin is correct, so this part has to be wrong. --Floquenbeam (talk) 19:00, 21 December 2020 (UTC)
 * Agree with GRuban and Floquenbeam. If Sanders' margin over Paquette was over 20% in both groups, it was over 20% overall, no matter how unequally the groups were chosen. If other candidates are considered, that would lower 20% to something in the teens, not to 0.11%. — Preceding unsigned comment added by Art LaPella (talk • contribs) 19:10, 21 December 2020 (UTC)
 * The referenced article says the following: "As is typical of postelection surveys, a larger percentage remembers voting for the winner than that which actually did. Sixty-three percent of the voters claim to have chosen Sanders in 1981, when, as mentioned above, he won by just ten votes". The paper goes on to say that the results of the survey are good for showing trends, and they think "an accurate picture of the Sanders coalition", but the magnitudes of the numbers should be viewed "with caution".  IMHO, I think we need to remove this section, the numbers are not accurate.  A less satisfactory solution, but one I would not necessarily fight against, would be to go into quite a bit of detail with the necessary caveats. --Floquenbeam (talk) 22:26, 21 December 2020 (UTC)
 * I am going to remove it; having voter analysis is a great idea, so we can return it once the caveats are added, however, until then, providing information we know is incorrect is ... not good. Pinging Jon again, because presumably he will be the one doing the work to return it. --GRuban (talk) 14:28, 22 December 2020 (UTC)