And had those rules been different the behavior of both the candidates and the electorate would have been different. Speculate all you want, it's irrelevant.golfcart wrote:Like them or not, those are the rules of the game...
And had those rules been different the behavior of both the candidates and the electorate would have been different. Speculate all you want, it's irrelevant.golfcart wrote:Like them or not, those are the rules of the game...
No offense but polls poll likely votersLTLFTcomposite wrote:So WetEV you're telling us pollsters were unaware of how presidential elections work? You're not making a very convincing argument for the "science" here.
But I did pick the source I was following before the election, you have a point. FiveThirtyEight was not the center estimate.golfcart wrote:I wouldn't want to accuse you of cherrypickingWetEV wrote: 28.2% chance of Trump winning, as of Nov 8th. (Polls-plus forecast)
https://projects.fivethirtyeight.com/20 ... id=rrpromo
So flip two coins. Got two heads? Now is that really unlikely?
10.5% is not so crazy of an outcome. So I'll round up 10.5% up to 12.5% and repeat my question. Is flipping three coins and having all heads so amazing?golfcart wrote:but your own link puts the odds of Trump winning the electoral college and losing the popular vote (which is what happened) at 10.5% in the "crazy and not so crazy scenarios" section at the bottom.
The popular vote matters when discussing national opinion polls, as that is what the polls are measuring. And yes, the people running the polls know the rules.golfcart wrote:And enough about the popular vote already, it didn't matter one bit. Like them or not, those are the rules of the game...
Everyone got it wrong, badly, is my point. I don't really think you are cherrypicking, I am just giving you a hard time about your comments regarding my trustworthiness after rounding up 1 percentage point.WetEV wrote:
10.5% is not so crazy of an outcome. So I'll round up 10.5% up to 12.5% and repeat my question. Is flipping three coins and having all heads so amazing?
And round down to 6.5%, repeat with four coins.
You can't judge a probability based prediction as flawed on one outcome. If the weather forecast said "10% chance of rain", and it rained tomorrow, would that be a "failed forecast"? Even if it didn't rain again, with the same forecast, for next 9 days?
I think that misses the point, those predictions were the likelyhood of Trump winning the election not Trump winning the popular vote. They poll people at the state level as well, they don't just take a national poll and then assume that each state votes in the exact proportions that the country votes. I'm not mad at you, I see where you are coming from... I just disagree and have made my case for why I disagree.WetEV wrote:The popular vote matters when discussing national opinion polls, as that is what the polls are measuring. And yes, the people running the polls know the rules.golfcart wrote:And enough about the popular vote already, it didn't matter one bit. Like them or not, those are the rules of the game...
Sorry, but no measurement is without error, other than perhaps a total count if you are sure you counted each and every item.golfcart wrote:If the polling were representative of reality there should be no error,
I'll admit that me saying no error was poorly worded, they shouldn't have all been wrong in the same direction it should have roughly balanced out is what I should have said. Randomly distributed error is not what we saw, a bias (in the statistical sense) is what we saw. I am sure you know what that is since you seem to have some knowledge of statistics.WetEV wrote:Sorry, but no measurement is without error, other than perhaps a total count if you are sure you counted each and every item.golfcart wrote:If the polling were representative of reality there should be no error,
Sampling error is well founded mathematics. Ignore it at your peril. As did some of the talking heads.
Question: Suppose "X" Trump voters changed their minds at the last second, and voted for Clinton.
What is the smallest number of "X" that would have changed the result?
What percentage of the total vote was that?
Polling does have its issues with sampling. During the last election cycle I was sampled by pollsters because I was reachable by landline. This time around I only have a cell phone so pollsters can't reach me; like many with cell phones I don't answer calls from numbers that I don't know. How do the pollsters get a representative sample of likely voters if there is a large subset they flat can't reach? And that sample of cell-phone-only folks is skewed younger — even though I am decidedly not young — since they are more likely to not have landlines. So, polling has sampling issues and I was surprised how close they came to national poll numbers despite the difficulties.golfcart wrote:...Weather forecasts are a bad analogy anyways... they are limited by our ability to properly sample the conditions of the atmosphere, our need to make simplifications in how we model the atmosphere, and some level of chaos in the system. Plus, a 10% chance of rain just means that there is a 10% chance that it will rain somwhere in the forecast area, it tells you little else. Most people don't understand what probability of precipitation forecasts really are. If it were a 10% chance of rain and it drizzled for 20 minutes I'd call that fine. If it were a 10% chance of rain and it poured for 12 hours I would question the accuracy of those forecasters and try to figure out what was so unique about this situation that allowed them to get the forecast so wrong...
Final Result Trump 306 Clinton 232WetEV wrote:Question: Suppose "X" Trump voters changed their minds at the last second, and voted for Clinton.
What is the smallest number of "X" that would have changed the result?
What percentage of the total vote was that?
Moving the goalposts and responding to your own questions rather than responding to my posts is not the sign of someone making a strong point. What use is the minimum X? The question to ask is what is the "X" in each state he won but she was favored between predicted and actual and what are the odds that every single X was biased in the same direction? Are you familiar with the concept of a straw man, you have provided a nice example?WetEV wrote:Final Result Trump 306 Clinton 232WetEV wrote:Question: Suppose "X" Trump voters changed their minds at the last second, and voted for Clinton.
What is the smallest number of "X" that would have changed the result?
What percentage of the total vote was that?
Pennsylvania. 20 Electoral votes. Trump 2,970,733, Clinton 2,926,441 Difference 44,292, number of Trump switching to Clinton = 22,147
Flip state total Trump 286 Clinton 252
Michigan. 16 Electoral votes Trump 2,279,543 Clinton 2,268,839 Difference 10,704, number of Trump voters to Clinton = 5,353
Flip 2 states total Trump 270 Clinton 268
Wisconsin. 10 Electoral votes. Trump 1,405,284 Clinton 1,382,536 Difference 22,748, number of Trump voters to Clinton = 11,475
Flip 3 states total Trump 260 Clinton 278
Total votes to flip result = 38,974 out of 128,838,341 = 0.03% (ignoring third party votes)
http://www.cnn.com/election/results/president
An alternate definition of an ass kicking. 0.03% win.