The first step is to admit you have a problem. Seems to me you are in denial. My relatives in Ohio said all along Trump was going to win that state handily and you kept saying No. Maybe you need to get out of your echo chamber in NY and drive around the country. The signs (literally) of a Trump victory were everywhere in these swing states.
16
The vast majority of people in this country are lucky if they can do the arithmetic to make change at the grocery store. They don't have the tools to understand or properly interpret anything in the realm of probability and statistics.
If they did, they could figure out for themselves that state sponsored lotteries and casino gambling are nothing more than a cynical and cruel con designed to to have 100% odds of draining money from the pockets of people who can least afford to lose it and for which the poor suckers are overpaying to play the game.
No one who tells you they won $1,000 in the lottery tells you they lost $10,000 winning it and it doesn't seem people do a very good job of realizing that.
Plus, I doubt any of the pollsters have very good models for dealing with the possibilities of respondents intentionally lying to screw things up.
There never seemed to be much mentioned about the number of undecideds. Don't recall seeing anything at Nate Silver's site. They were probably buried in there in the averages and confidence intervals and probabilities. But there's nothing like a mystery candidate named Undecided taking up space on the graphs who's polling a heck of a lot more than the real 3rd party candidates.
If they did, they could figure out for themselves that state sponsored lotteries and casino gambling are nothing more than a cynical and cruel con designed to to have 100% odds of draining money from the pockets of people who can least afford to lose it and for which the poor suckers are overpaying to play the game.
No one who tells you they won $1,000 in the lottery tells you they lost $10,000 winning it and it doesn't seem people do a very good job of realizing that.
Plus, I doubt any of the pollsters have very good models for dealing with the possibilities of respondents intentionally lying to screw things up.
There never seemed to be much mentioned about the number of undecideds. Don't recall seeing anything at Nate Silver's site. They were probably buried in there in the averages and confidence intervals and probabilities. But there's nothing like a mystery candidate named Undecided taking up space on the graphs who's polling a heck of a lot more than the real 3rd party candidates.
5
You shouldn't have shown the "probability of winning" as the most prominent statistic. I design data for a living, under much less important circumstances, and I never in 1 million years would have mislead with this data. It bothered me throughout the entire election.
Probability of winning belongs halfway down the page of other statistics on 538.com. It's for nerds. It's not actual information. It does not belong on the homepage of the newspaper of record's app every single day.
Who on earth cares about the probability outside of statisticians? People care about the actual numbers of likely voters. It should have been showing likely votes in the likely turnout, or something similar, which would have been closer to 50-50 the whole time.
Probability of winning belongs halfway down the page of other statistics on 538.com. It's for nerds. It's not actual information. It does not belong on the homepage of the newspaper of record's app every single day.
Who on earth cares about the probability outside of statisticians? People care about the actual numbers of likely voters. It should have been showing likely votes in the likely turnout, or something similar, which would have been closer to 50-50 the whole time.
21
Trump's win was a low probability event any way you look at it. The prediction explanation should have made clear that low probability events like missed field goals happen all the time but magical thinking reduced his chance of winning to zero in the minds of readers.
10
Another relevant quote is Lloyd Christmas: "So, you're telling me there's a chance."
9
It is one thing to prognosticate within the bounds of an article. It is another to run a banner across your website with your prediction. The placement of the banner added to the perception that the statistic represented the certainty rather than the odds that Hillary Rodham Clinton would be elected president.
12
Analyze away. She lost because of her aloofness, entitlement, unlikable demeanor and more. The biggest grandiose neglect was toward a frustrated working class that in the face and yes insult of being left out was further driven to Trump, like him or not, by the Hollywood executives, movie and recording stars. Oh yeah "we got this" they really said. Then it all crumbled with help from Comey and the fact that she never had a message. You deserve your misery along with your blind feminist sisters. Blinded by the light of a woman Presidency.
13
Comey performed as "the Jeb Bush" of 2016. He gave the election to Trump and the 'Trumpets", the same as Jeb Bush gave that election to his brother "What's His Name"...........
5
The polls played a huge roll in this election. Who wasn't seeing the percent chance that Hillary was going to win right there on the front page of the online NYT? How did that affect voter turnout or third party votes? The news agencies need to rethink the way they report this information.
24
I for one was misled into false optimism by The Upshot 's numbers. Nonetheless I voted for Hillary. How many were given false optimism and hence failed to vote?
11
So many theories about why Trump won -- Social media, racist rural whites, low Black turnout, Trump's a con man, Hillary is a bad candidate, the FBI, the Russians, the Koch brothers, Anthony Wiener, the Millennials, the male patriarchy resisting powerful women, the failure of the Obama coalition, third party candidates, Bernie voters staying home, the Electoral College, the Media, bad polling, gerrymandering. Maybe people just preferred Trump and the Republicans.
12
"Maybe people just preferred Trump and the Republicans."
...except that HRC won the popular vote.
...except that HRC won the popular vote.
13
Bad, primitive models and poor mastery of statistics sounds more like a true explanation of your pathetic grand fail.
25
All I can say, with all due respect, is that I will not be looking to The Upshot for any prognostications in the future.
38
Did you ever consider that maybe the people who were willing to talk to pollsters were skewed toward Clinton? The vote shaming in this election was/is off the charts. I have a friend in a traditionally blue state who voted for Trump but would not put a bumper sticker on her car or a sign in her yard because she was afraid of vandalism. Do you think she would have been truthful if a stranger called her and asked who she supported?
26
What if instead of "modeling", you've had done some old fashioned fieldwork?
22
The polls weren't wrong
Americans need a Woodward/Bernstein Journalist team to dig up the truth. All of the polls could not be this wrong. If you are an investigative reporter, please please begin investigating.
Tricia Tennesen
South Carolina
Americans need a Woodward/Bernstein Journalist team to dig up the truth. All of the polls could not be this wrong. If you are an investigative reporter, please please begin investigating.
Tricia Tennesen
South Carolina
8
I love polls and polling analysis but I think we need fewer polls. They have too much influence on behavior. Americans don't understand math - not compound interest, not interest in general, and certainly not probability theory! Maybe polls that focus less on who than why or what - policies or position. So much depends on how the question is phrased, and I suspect from personal experience that is not well done (none of the above is my usual response - nothing matches). But straight voting preference as asked this cycle is not a good predictor.
BTW I am one of those people who do not answer the phone for an unknown number, AND I won't do an internet poll. Lots of us are hard to reach.
BTW I am one of those people who do not answer the phone for an unknown number, AND I won't do an internet poll. Lots of us are hard to reach.
13
Questions??
In the aftermath of this election, main street media has embarked on a flurry of self-recrimination. I think disregards counter-intuitive reality in favor of wishful thinking and cliché has proved to be the root cause.
I just wondering are there any other established beliefs/conventional wisdom are wrong yet most of us still unaware the wrong (since there is no clear cut scores, such as Electoral College Votes, to prove otherwise.)
In the aftermath of this election, main street media has embarked on a flurry of self-recrimination. I think disregards counter-intuitive reality in favor of wishful thinking and cliché has proved to be the root cause.
I just wondering are there any other established beliefs/conventional wisdom are wrong yet most of us still unaware the wrong (since there is no clear cut scores, such as Electoral College Votes, to prove otherwise.)
2
How about a sign test on the errors in the state polls' predictions? If it indicates a one-sided error in favor of Mr. Trump, then how about some research on sandbagging?
Maybe it's not the odds themselves but how the you display the probabilities. To say a candidate has a 93 percent chance of winning makes it sound like a done deal, when the raw numbers show a difference of less than 2 percentage points. The graphic needs to temper the "cast in stone" aura of inevitability that a high percentage chance of winning suggests.
For example, when the Cubs were down 3-1 and Nate Silver wrote that both the Cubs and Trump had a 25 percentage chance of winning, the probability of Clinton winning seemed a lot less solid than when just Trump number was reported. It seemed boggier still after the Cubs won the Series.
For example, when the Cubs were down 3-1 and Nate Silver wrote that both the Cubs and Trump had a 25 percentage chance of winning, the probability of Clinton winning seemed a lot less solid than when just Trump number was reported. It seemed boggier still after the Cubs won the Series.
5
Polling is an imperfect science. I can live with the majority of polls getting it wrong, since my work in the social sciences over the years has taught me why traditional polling methods are becoming increasingly ineffective each year. That I understand. What's hard to comprehend is why the staff at the NYTimes felt it was more important to use its influential platform to help elect HRClinton instead of adhering to its original vision of keeping all sides accountable.
I've been a subscriber of the Times for years. I try to stay informed by accessing multiple media sources. Growing up I saw top traditional newspapers as essential for acquiring an objective perspective. However, the biased way you covered this election, with such blatant support for HRClinton at every turn shocked me. I felt embarrassed for you...and in the process I lost a once highly valued and trusted news source.
History verifies that one of the reasons our US democracy has lasted longer than any other is due in great part to the critical role that a free press has played since our inception. I know that 100% objectivity is an unrealistic expectation (even though it ought to be the goal of every news outlet). But by allowing yourselves to become a pure propaganda voice for ANY side, party or view you forfeit your higher calling as democracy's watchdogs, placing our country in enormous jeopardy. Forget this election. Restoring YOUR integrity is much more important than who wins a single election.
I've been a subscriber of the Times for years. I try to stay informed by accessing multiple media sources. Growing up I saw top traditional newspapers as essential for acquiring an objective perspective. However, the biased way you covered this election, with such blatant support for HRClinton at every turn shocked me. I felt embarrassed for you...and in the process I lost a once highly valued and trusted news source.
History verifies that one of the reasons our US democracy has lasted longer than any other is due in great part to the critical role that a free press has played since our inception. I know that 100% objectivity is an unrealistic expectation (even though it ought to be the goal of every news outlet). But by allowing yourselves to become a pure propaganda voice for ANY side, party or view you forfeit your higher calling as democracy's watchdogs, placing our country in enormous jeopardy. Forget this election. Restoring YOUR integrity is much more important than who wins a single election.
1
I read the Times cover to cover everyday. I thought their coverage of Trump was more than fair. His rise was partly due to the media's lack of scrutiny early on. The Clinton email scandal was the creation of right wing media and then amplified throughout their chain of propaganda site to inflate its importance. She exercised poor judgement, but that's where the story should have ended. If she had sold secrets to the Russians the story couldn't have been reported with any more vitriol and outrage than the email story was reported. The idea that she could go to jail for that sloppy behavior is ludicrous. The Times picked up that story and ran with it and never let it go which I think was irresponsible. If anything Trump was given too much of a pass early on. Outside of the opinion pages I detected no bias toward Clinton whatsoever.
3
That's odd. I saw bias against HRC in almost every article. Does "both candidates are flawed" sound honest to you?
I am also looking for a reliable news source. I read politically opposite news sources and try to average them together to get the truth. Probably not valid math but there are few other choices.
I am also looking for a reliable news source. I read politically opposite news sources and try to average them together to get the truth. Probably not valid math but there are few other choices.
3
The upshot is that The Upshot and everybody else got it wrong.
Political polls, like weather forecasts, typically are presented with an unjustified air of confidence and specificity -- and then, when the predictions turn out to be wrong, the mistaken predictors have the nerve to take exception to their exasperated critics.
If polls were presented as educated guesses instead of precise, scientific measurements, there would be less backlash when they go awry.
The great irony of political campaign polling is that when polls are of the greatest interest -- when an election is close, or shifting -- they're the *least* reliable.
So why bother?
Political polls, like weather forecasts, typically are presented with an unjustified air of confidence and specificity -- and then, when the predictions turn out to be wrong, the mistaken predictors have the nerve to take exception to their exasperated critics.
If polls were presented as educated guesses instead of precise, scientific measurements, there would be less backlash when they go awry.
The great irony of political campaign polling is that when polls are of the greatest interest -- when an election is close, or shifting -- they're the *least* reliable.
So why bother?
9
I believe that many Clinton supporters failed to vote simply because all of the media models predicted she was an almost certain win. People who were having a busy day, bad weather, or feeling sick just didn't put in the extra effort to get to the polls, because she had an 85% chance of winning. The roll of the media in this race cannot not be understated. This is a sort of false confidence bias that results from polling itself. Maybe in two years the New York Times can skip the prognosticating, or say it is too close to call. I have noticed that when a race for president is very close, turn out is much higher.
17
I agree. This is especially amplified by some news outlets reporting a ridiculous win probability for Clinton.
For instance, the Huffington Post, a website that is especially targeted at liberals, had a Clinton win probability of 99 % leading up to the election!
For instance, the Huffington Post, a website that is especially targeted at liberals, had a Clinton win probability of 99 % leading up to the election!
3
When the media labels Trump an unconventional candidate, it's using a euphemism for what he really is. I suspect the biggest contribution to the polling error was the fact that a segment of the voters in the rust belt states were not truthful with the pollsters because they were ashamed of their own motivations to vote for Trump. Trump was a deplorable candidate when he blamed others for their problems. Yes, I am going to say it: He gained popularity using methods similar to Hitler's. And, Trump made ridiculous abstract promises that any one would be a fool to believe. At some level, these voters knew that Trump was making racist, xenophobic declarations, and they were ashamed to reveal their support.
1
My question is what likelihood the model assigned to a surge in third party candidate support. As things turned out in Wisconsin, for example, Donald Trump got the same percentage of votes that Mitt Romney did in 2012. The difference was that 4% of the non-Republican vote in Wisconsin went to non-Democratic candidates. Similar scenarios played out in Pennsylvania and Florida and, indeed, in the national vote totals. Had this contingency been given a higher weighting, perhaps the non-votes for Trump would not have automatically interpreted as votes in favor of Clinton.
1
Trump cracked Hillary's aces. It sucks but it happens.
3
Updating the now unfunny line, "liars, damned liars, and pollsters..."
8
Humans are too complex to build accurate models what they will do
But I have an inkling based on Cokie Roberts observation that voting for the American President is the one exception to the rule: ' Americans vote for Presidents not on what they say, but whom they are"
Voters could perfectly read Mr. Trump. They just could not read Ms. Clinton,
'The most opaque person you'll ever meet in your life,”
Charles E Schumer, D, Senior Senator NYS, with whom Ms. Clnton served while being the Junior Senator from NYS.
But I have an inkling based on Cokie Roberts observation that voting for the American President is the one exception to the rule: ' Americans vote for Presidents not on what they say, but whom they are"
Voters could perfectly read Mr. Trump. They just could not read Ms. Clinton,
'The most opaque person you'll ever meet in your life,”
Charles E Schumer, D, Senior Senator NYS, with whom Ms. Clnton served while being the Junior Senator from NYS.
4
Everyday I looked at the NYT Presidential Forecast and took heart from it and obviously too much comfort. Maybe it lulled others into a false sense of security. Still, I thank NYT reporters and the Editorial Board for all their work during this election to try to wake up the populace to the problems of electing an inexperienced and unqualified reality tv start to the highest office.
6
To those who confuse 85% with 100%: would you play a round of Russian roulette? Your odds of surviving unharmed are 83%, after all (assuming an ideal six-shooter).
Events with a probability of 1/6 happen routinely, as anyone who has ever played dice can tell.
Events with a probability of 1/6 happen routinely, as anyone who has ever played dice can tell.
21
Human beings are notoriously non-monotonic.
2
Why should there be an ability to reliably forecast an election result? If there was, the electorate would be programmed robots.
3
I suspect less "Box and Jenkins" times series analysis and more E.T. Jaynes "Principle of Maximum Entropy" needs to be injected into The Upshot's processes.
1
What has to be determined, by viewing the data of all Boards of Election and the census, is the proportion of registered voters to those who actually voted, among various ethnicities, age groups and genders, in all presidential elections from 2000 thru 2016.
This data should also compare registered voters to eligible voters. Correlating data should also determine how many citizens are unable to vote because they are former felons or current felons, including ethnic and regional patterns.
One thing we know for certain, is that neither Donald nor Hillary were themselves popular among most Americans. In the general election, many people voted for a candidate in protest of the other candidate.
In the Primaries, none of the many Republican candidates were popular, although Trump ascended by the other candidates splitting a majority of the votes. And the media gave Trump far to much free exposure.
But there were only two meaningful candidates competing in the Democratic Primaries. Many Democrats, not only young adults, but Baby Boomers too, greatly preferred Bernie Sanders, the main exception to those age groups being African Americans in particular in the South. Almost all Bernie supporters voted for Hillary in the General Election, with a few strays voting for Jill Stein.
Of course, many Democrats of all ages voted for Hillary, more out of respect for her. But, she was old news to young adults, and to everyone.