Timeline for US Election results 2016: What went wrong with prediction models?
Current License: CC BY-SA 3.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 16, 2016 at 20:52 | comment | added | Dmitry Rubanovich | The shortcoming of the polls were not necessarily in the counting. They could have been in the failure to consider certain scenarios and not asking the questions which would estimate the said scenarios' probabilities. I am saying that this is a fairly likely scenario. In fact, polling to find out how Libertarians switched their votes after the fact would show whether I am right or wrong. | |
| Nov 16, 2016 at 20:50 | comment | added | Tim | Then expected from polls conducted before elections, while all this thread deals about their shortcomings... | |
| Nov 16, 2016 at 20:48 | comment | added | Dmitry Rubanovich | @Tim, but you do know who. Trump got more votes than expected. So in the absence of other evidence, you have to assume that they went to him. In the absence of contrary evidence, you have to assume the most likely scenario when making estimates. | |
| Nov 16, 2016 at 20:46 | comment | added | Dmitry Rubanovich | @Tim, (continued) There is also the matter of the fact no polls specifically asked Libertarians which issues would be most likely to sway their votes. And the fact that strict constitutionalism is generally a significant part of their agenda. | |
| Nov 16, 2016 at 20:46 | comment | added | Tim | But (a) they voted "someone else" -- you don't know whom; (b) others also switched their preferences (Clinton supporters, Trump supporters etc.), so in the end we do not know what was the impact of all those switches. | |
| Nov 16, 2016 at 20:44 | comment | added | Dmitry Rubanovich | @Tim, the poll taken the day before the election gave Johnson 5% of the popular vote. He ended up getting 3% of the popular vote. It may seem like this is within the margin of error of 3%, but 4%-5% Johnson support was consistent for a few weeks before the election across all polls. So a drop to 3% on the election day is fairly significant. That's 20%-40% of would-be Johnson voters who voted for someone else. Given that all the states which were close were won by Clinton if Johnson got more than 3% and were won by Trump if Johnson got less than 3%, I think these numbers are significant... | |
| Nov 16, 2016 at 6:24 | comment | added | Tim | Afaik, polls ask about possible vote-switching and forecasts take it into consideration. Do you have any information that suggests that before the election there was any such information that was not taken into consideration by any forecast or is this a pure speculation? | |
| Nov 11, 2016 at 22:44 | comment | added | Dmitry Rubanovich | @John, but it's not just NH. Minnesota: Johnson 4%, Trump lost by 2%; NV (a harder argument to make, but still the trend holds): Johnson 3.5%, Trump lost by 2%; Maine: Johnson 5%, Trump lost by 3%; Colorado Johnson 5%, Trump lost by 3%. | |
| Nov 11, 2016 at 22:27 | comment | added | John | I understand that. I was trying to explain that the libertarian party is strong in NH. | |
| Nov 11, 2016 at 22:01 | comment | added | Dmitry Rubanovich | @John, NH Libertarians stuck with Johnson (4%). Trump lost the state by 1%. | |
| Nov 11, 2016 at 21:58 | comment | added | John | NH has all the people from the free state project living there. Thriving libertarian party and active supporters. | |
| Nov 11, 2016 at 21:54 | history | edited | Dmitry Rubanovich | CC BY-SA 3.0 |
added 21 characters in body
|
| Nov 11, 2016 at 21:49 | history | answered | Dmitry Rubanovich | CC BY-SA 3.0 |