Watching the elections on MSNBC, I’ve noticed they’ve already projected Hillary winning, and Gore taking Florida…even though Lazio (in NY) and Bush are winning (in FL), according to their percentages. Why are they projecting Gore to take FL and Hillary to take NY when they are behind in the votes?
It’s like predicting anything, they use old trends, exit polls, and then guess.
If they guess wrong, it’s forgotten in two hours.
Generally, this kind of prediction is made based on profiles of the precincts.
E.G., in the great state of Urbrus, all city dwellers are Democrats and all farmers are Republicans. The cities have a horrendous mess trying to get everyone in and out of the polls, while the small towns generally get their votes tallied as soon as the second milking is complete. If an hour after the polls close, 90% of the farmers’ votes have been tallied, (100% for Bush) and only 22% of the city votes have been tallied (100% for Gore), Bush may lead 200,000 to 170,000. However, when the remainder of the cities’ votes have been tallied, Gore will be expected to pick up at least another 603,000 votes while Bush will be expected to pick up only 222,222. Therefore, they call Gore the winner 773,000 to 422,222. (I did the math in my head; I’m sure there’s an error in there.)
It is not nearly that cut-and-dried, of course, but that’s the basis of their predictions.
And of course, they could be wrong. They base their projections on earlier polls of the electorate, and on historical data. If there is a paradigm shift, those assumptions may be wrong.
It looks like that has happened to some degree in Florida. It was called for Gore, but as more votes have come in that didn’t trend the way the predictions indicated, Florida has moved back in the ‘too close to call’ category.
The major networks contract out to a private firm that does the election calling. That firm may be eating some crow tomorrow.
CNN did the flip-flop on Florida, as well.
Interestingly, they declared California a Gore victory the instant (literally) that the polls closed out here. Checking the state results chart a few minutes after 8:00 Pacific time revealed that Gore had “won” with 0% of the vote (0% of precincts reporting).
Right as I post this (updated 8:43 p.m., Pacific time), 23% of the precincts have reported, and Bush is ahead 52%-44%. Gore is still listed as having “won” the state.
I know, by most accounts, Gore is a lock to win California. Still…
With 19% of the precincts reporting, Bush is still holding a 2% lead in California.
I’m assuming that they called California for Gore based on Exit Polling. There’s probably a few very liberal, large districts that haven’t reported yet (like San Francisco).
Another interesting fact: Florida has now had 82% of all precincts reporting, Bush still leads by 2% (over 120,000 votes), yet now the networks won’t call it for Bush.
By my calculation, there are only about 320,000 votes left to count. Gore would have to get about 70% of them to win. So why won’t they call it? In fact, they aren’t even mentioning Bush’s lead.
Anyway, Bush is going to win. He’s only 24 votes away, wiht 61 votes left. He’ll probably win even if he loses Florida. Gore will only win if he manages to carry Florida, Washington and Wisconsin.
Right now abcnews.com has both major candidates at 242 electoral votes. Gore needs:
-
Florida and any other state or;
-
Every state but Florida.
Bush needs the same.
CNN political people are saying that most of the votes not counted yet in FLA are predominantly democratic, which is, they say, why they haven’t given FLA to bush yet.
California’s votes always trend Republican in early returns because of absentee voting. At 10:50 pm PST, Gore had gone ahead and will likely cruise to an easy win. Los Angeles County takes a long time to report its votes and it is already breaking 60-40 for Gore.
Statistics 101-- When you do a poll of any kind you can never get truly accurate results. The best you can do is say, " the results are X% (your confidence interval) accurate +/- Y%(Your margin of error)." The smaller the X and Y, the larger the sample you need.
Say your exit pools showed Gore leading Florida by 5% of the vote and your margin of error was =/- 4%. You’d be reasonably sure that Gore was going to win if your confidence interval was high enough (say 99%). But every once and a while you’d be wrong-99% isn’t 100%.
Also, according to ABC news Voter News Service (a news consortium that includes ABC, CBS, CNN, Fox, NBC and The Associated Press) said the “early readings of their exit polls” showed Gore won. Now the results of thier polls didn’t change. However it is possible that the data from earlier in the day showed Gore had a bigger margin, but the later voters voted for Bush. However, in setting up an accurate poll, you have to choose your sampling accurately. You must make sure your respondents come from the complete spectrum of the demographics and you must make sure that the percentages accurately reflect who is voting.
If the Gore voters got out early and the Bush voters late and you didn’t wait for all your exit poll date to come in, your results would be invalid. If that’s what they meant by “early reading of their exit polls” then there’s the problem right there.
Two more problems, first (as someone here mentioned) exit polls don’t take into account absentee ballots. Second, no network want to be last with the news. If one says Gore wins Florida, there will be tremendous pressure for the others to say so too. In 1976 pretty much everyone except CBS said that Gerald Ford won Ohio. Then everyone backed off their predictions–it was declared to close to call. About 3AM, Ohio officially went to Jimmy Carter and the rest is history. The guy at CBS (I don’t recall his name) that stuck by his guns and said the race was too close to call admitted that if the other networks hadn’t backed off the prediction and if Ford had indeed won, he (the CBS guy) would have been out of a job.
In an extremely tight race, picking a winner is difficult, even with the best of exit poll results. Twice in seven election this has happened. And here’s my prediction–Unless congress forces the networks to be more responsible, it will happen again because the networks will **never **do it on their own.