Non-political questions:
What is the biggest deficit that any candidate has ever overcome in the polls, on the eve of the election, to win? (assuming a direct election, not some parliamentary system.) Has any candidate ever trailed by, say, 10% in polling on the eve of Election Day and yet won?
What is the biggest margin that has ever been overcome by a vote recount? Has any candidate ever trailed by more than 1% and yet won in a recount?
The biggest change that I knoe of is much closer than that, only about 8/100th of 1% difference.
This was the Coleman-Franken Senate election of 2008 in Minnesota.
Election night, the preliminary results reported in the news media said that Coleman ® was 225 votes ahead of Franken (D) out of the 2.8 million votes cast.
The initial results reported to the Secretary of State 2 weeks later were Coleman over Franken by 215 votes. This ws close enough (less 1/2 of 1 %) to trigger an automatic recount by state law.
The manual recount was held in public, with representatives of each campaign watching. At the end, results reported to the State Canvassing Board, and their review of challenged ballots, showed Franken ahead of Coleman by 49 votes.
There remained some absentee ballots which had not been counted. Some of them were incorrectly rejected; they should have been counted. The campaigns aregued (Franken wanted them counted, Coleman did not), and filed suit. The Mn Supreme Court ruled that each county should find improperly rejected absentee ballots, and send them to the Secretary of State. They were jointly considered by his office, together with the campaigns, and decided which ones to count. When those were counted, the final result showed Franken leading Coleman by 225 votes, and the State Canvassing Board certified those as the results.
The Coleman campaign then sued to prevent the state from issuing an election certificate to Franken. The trial took 2 months, with hundreds of witnesses, and recounted some of the absentee ballots. The court decision showed Franken leading Coleman by 312 votes.
the Coleman campaign then appealed to the MN Supreme Court. They took 2 months, then rejected the appeal and affirmed Franken’s 312 vote win. He was then awarded his election certificate, and seated by the US Senate.
Not really what you are asking, but… The Quebec referendum on separation from Canada in 1995 was won by the “No” (Stay in Canada) side, 50.58 to 49.42 or 2,362,248 to 2,308,360 votes.
The difference of 53,888 ballots was less that the rejected ballot count of 86,501. Conspiracy theories say the separatist government tried to reject whatever it could to change the outcome. However:
There’s a problem with the first question in the OP: For many elections, and certainly for very important elections, there’s no single eve-of-the-election poll. There’s usually several such polls and they won’t agree with each other exactly. Furthermore, not all polls are equally good. For example, fivethirtyeight weights the polls for president and other offices based on their perceived quality and also adjusts their results for bias. What they give is a running average of a number of different polls after these weights and adjustments are applied.
Not sure why you think parliamentary election aren’t direct?
1998 in the Queensland Federal seat of Blair.
The sitting candidate was the LIB Cameron who was being challenged by the LAB candidate Clarke and the populist xenophobic leader of One Nation Pauline Hansen.
Pre-poll results showed a collapse in the LIB vote and widespread cross party support for Hansen who was clearly the front runner by over 10% from LAB.
Hansen polled 36.8%, LAB 25.1% and Lib 21.6%, a swing against them of 24.5%.
However all parties preferences against Hansen and LIB snuck ahead of LAB, allowing LIB to win 52.9% two party preferred.
Polling was cruder back then, but everything pointed to Dewey defeating Truman in 1948.
And when the polled people after the election, asking who they voted for, Dewey won, too.
Evidently a lot of people voted for Truman and wouldn’t admit it to pollsters.
One analysis I read said this was one of the first elections that significantly included scientific polling. However, it suffered from the bias that a lot of the large-sample polling was done by telephone, so it biased toward urban better off voters who had phones and were more inclined to vote for Dewey. Truman appealed to a lot of rural and poorer voters who were overlooked in the polling.
it also explains the famous headline, because returns from the bigger urban centers got reported first, and it appeared Dewey was winning. As rural votes trickled in, the tide turned.
It’s not something we haven’t seen recently in a number of presidential elections, where early returns indicate a state is going one way and sometimes is even called that way by the media, only to have the late returns swing it the other.
This list of close election results doesn’t make it easy to determine the largest initial vote margin overturned on a recount, but I didn’t notice anyone trailing by more than a few hundred votes and then winning on a recount/recanvass.
In other words, unless his allegations of massive voter fraud somehow pan out, Matt Bevin has a snowball’s chance in hell of continuing as governor of Kentucky (the results on election night had him losing by over 5,000 votes to Beshear).
In 1936, the well-respected* Literary Digest * polled an unprecedented 2.4 million people and predicted Alf Landon would beat Franklin Roosevelt with a landslide 57% of the popular vote. Their polls had successfully picked the winner of the five previous Presidential elections.
Roosevelt won a staggering 62% of the vote. It destroyed the magazine’s credibility and the Literary Digest folded in 1938. That wasn’t a question of overcoming a deficit, though. LD conducted a spectacularly bad poll - essentially talking only to people who were doing pretty well in the middle of the depression. It’s now used as a case history in statistics courses.
My grandmother was a lifelong Republican but admitted she had voted for Truman. She said she had entered the booth with every intention of voting for Dewey but then couldn’t stand the thought of “looking at that damned mustache for four years.”
If you include referendums, the majority of opinion polls right up to the day before the Brexit vote showed Remain (stay in the EU) to be well ahead of Leave. Obe poll showed a 55%-45% margin on the day before the referendum: Opinion polling for the United Kingdom European Union membership referendum - Wikipedia
A short but informative article about the flaws of polling in 1948. Plus note the final Dewey lead was a measly five points. It was just the pollsters had been telling the public so long the election was over they were afraid of calling attention to any doubts they had.