3 ms·

Why FiveThirtyEight Gave Trump a Better Chance Than Almost Anyone Else

15 points by mbgaxyz 10y ago

Hydraulix989 10y ago
Should be titled: "How FiveThirtyEight Desperately Tries to Regain its Credibility"
- godzillabrennus 10y ago
  This! I was tracking their polling aggregator in "real time" daily and all it did was give me a false sense of security. Now we have Trump.
  - detaro 10y ago
    If someone gives you a 1/3 to 1/4 chance of something bad happening, relying on that as "secure" seems ... risky. Criticism of specific mistakes they made are good and well, but just saying they predicted the wrong percentage (when we have no good idea what the real one was) seems to me like faulting somebody because they lost a coinflip. The bigger issue IMHO is how much attention is given to polls and predictions, when they try to amplify useful data from tiny differences in key states. Results that indicate "we can't really know" seem likely, but make for really bad headlines. And worst case, some people actually think 70:30 means they can stay at home and not go vote or something...
  - inimino 10y ago
    As I said in another thread[1], I think this is the psychological explanation for a lot of the heat Nate Silver took before and after the election (being "wrong" both ways according to people who don't understand what a 70% chance means). People saw 70% and rounded up to "Hillary". For a psychological counterbalance, imagine your doctor tells you you have a 70% to be alive this time next week. Yet in the political realm people were calling that a "sure thing". [1] https://news.ycombinator.com/item?id=12937736 https://news.ycombinator.com/item?id=12937736
- SamReidHughes 10y ago
  I'm confused. Was their 29% chance "wrong"? Would a 40% chance still be wrong? 45%?
  - Hydraulix989 10y ago
    In general, people don't have such good intuition when it comes to probabilities. Their heuristic is that if the predicted outcome doesn't match the actual outcome, you are wrong.
mshenfield 10y ago
They were pretty clear about the level of uncertainty leading up to the election. They said the data they were seeing and Donald Trump winning could both occur in a random universe slightly more than flipping a fair coin twice and both times getting tails (29%). Probability exists to quantify uncertainty, not to predict the future (though this can help). One thing is clear - we don't live in a random universe, and there are concrete biases present in the way we collect election data. FiveThirtyEight carefully avoids and accounts for many of these sources of uncertainty. For example, herding of polls at the end of an election and the unusually high percentage of undecided voters at the end of the election.
DefaultUserHN 10y ago
This whole election, every time Nate Silver made a prediction about Trump, it turned out to be wrong. That's a 100% failure rate. Now, they're trying to claim that they were the least wrong? LOL.

15 points by mbgaxyz 10y ago

Hydraulix989 10y ago

Should be titled: "How FiveThirtyEight Desperately Tries to Regain its Credibility"

godzillabrennus 10y ago
This! I was tracking their polling aggregator in "real time" daily and all it did was give me a false sense of security. Now we have Trump.
- detaro 10y ago
  If someone gives you a 1/3 to 1/4 chance of something bad happening, relying on that as "secure" seems ... risky. Criticism of specific mistakes they made are good and well, but just saying they predicted the wrong percentage (when we have no good idea what the real one was) seems to me like faulting somebody because they lost a coinflip. The bigger issue IMHO is how much attention is given to polls and predictions, when they try to amplify useful data from tiny differences in key states. Results that indicate "we can't really know" seem likely, but make for really bad headlines. And worst case, some people actually think 70:30 means they can stay at home and not go vote or something...
- inimino 10y ago
  As I said in another thread[1], I think this is the psychological explanation for a lot of the heat Nate Silver took before and after the election (being "wrong" both ways according to people who don't understand what a 70% chance means). People saw 70% and rounded up to "Hillary". For a psychological counterbalance, imagine your doctor tells you you have a 70% to be alive this time next week. Yet in the political realm people were calling that a "sure thing". [1] https://news.ycombinator.com/item?id=12937736 https://news.ycombinator.com/item?id=12937736
SamReidHughes 10y ago
I'm confused. Was their 29% chance "wrong"? Would a 40% chance still be wrong? 45%?
- Hydraulix989 10y ago
  In general, people don't have such good intuition when it comes to probabilities. Their heuristic is that if the predicted outcome doesn't match the actual outcome, you are wrong.

mshenfield 10y ago

They were pretty clear about the level of uncertainty leading up to the election. They said the data they were seeing and Donald Trump winning could both occur in a random universe slightly more than flipping a fair coin twice and both times getting tails (29%). Probability exists to quantify uncertainty, not to predict the future (though this can help). One thing is clear - we don't live in a random universe, and there are concrete biases present in the way we collect election data. FiveThirtyEight carefully avoids and accounts for many of these sources of uncertainty. For example, herding of polls at the end of an election and the unusually high percentage of undecided voters at the end of the election.

DefaultUserHN 10y ago

This whole election, every time Nate Silver made a prediction about Trump, it turned out to be wrong. That's a 100% failure rate. Now, they're trying to claim that they were the least wrong? LOL.