Predictions for the 2016 Election

[Nous]

It is obvious by the fact that in practically every US state and the District of Columbia (all but 4 states), the majority (in most cases the vast majority) live in urban areas.

What does "it" in your sentence refer to? What is "obvious"? What is the problem that the electoral system of electing the President supposedly solves?

Point is that because 80.7% of the US population live in urban areas, and most live in major urban areas on the East Coast, small sections of the West Coast, and Florida, it is easy to see large parts of the country that would be ignored during campaigns

When every citizen is allowed to one vote that (a) will not be wasted (as is the case in the electoral system) and (b) will not be unequal in weight (as is the case in the electoral system), then no one is ignore.

It is in the current state-winner-take-all system that the vast majority of the country is ignored in elections. See: http://www.npr.org/sections/itsallpolitics/2012/11/01/163632378/a-campaign-map-morphed-by-money

and:

http://www.fairvote.org/2012chart

All the candidates' time, attention and money goes to a handful of "battleground" states.

and in national politics.

Straw man. The method of electing the President does not determine what will happen in Congress (or what the President will do) regarding "national politics". According to the Census data you provided, the majority of people in every state live in "urban areas" (areas with a population of 2,500 or more--which is not very "urban" to my mind), thus under the state-winner-take-all system currently used in 48 states, the winner of the Presidential election in each state is determined by people who live in "urban areas".

Every vote would count the same, but that is not at issue.

Vote wasting, voter apathy (due to knowing that one's vote will not go toward electing the President), and unequal vote weight are highly important democratic issues, and are problems directly caused by the anti-democratic electoral system of state-winner-take-all method of electing the President.

Los Angeles does not control the outcome of statewide elections in California and therefore is hardly in a position to dominate a nationwide election. The fact that Los Angeles does not control the outcome of statewide elections in its own state is evidenced by the fact that Republicans such as Ronald Reagan, George Deukmejian, Pete Wilson, and Arnold Schwarzenegger were elected Governor in recent years without ever winning Los Angeles.

This is a silly comparison, as there are several major cities in California.

The paragraph I quoted does not make a "comparison". It uses Los Angeles as an example of the fact that big cities (even the biggest of the big) do not necessarily determine the winner of an election.

Nevertheless, if candidates had to campaign in ("only") the 50 largest cities, it would nationalize their campaigns and require them to appeal to many more people and to much more diverse populations than they do in the current electoral system. In the current election, the candidates spend almost all their time, money and attention in FL, NC, NH, NV, PA, OH.

My point was to the concentration of population in certain parts of the country.

Again, what problem does the state-winner-take-all of awarding electoral votes supposedly solve?

As is made obvious by the map I provided, the US population is overwhelmingly concentrated in the Eastern Seaboard, Florida, relatively small sections of the West Coast, and several areas in the South East. This would make a popular vote election unfair to many states in the Union.

How is each person having one non-wasted voted that is equal in weight to every other vote "unfair"?

Again, according to the Census data you provided, the state-winner-take-all method of awarding electoral votes only ensures that all the electoral votes of the state go to the candidate as determined by people living in "urban areas".

 

[Nous]

There was a 14% spread iirc

Where did you get that idea?

That percentage doesn't take into account future events so doesn't relate to 'chance of winning'. Just look at the history of 'chance of winning' to confirm that.

What the hell does that mean? Show me "the history of 'chance of winning,'" and I'll look at it.

it is fair to assume that Clinton is favourite.

Why do you believe that "it is fair to assume that Clinton is the favorite," if your claims were true about all the polls erroneously calculating their margins of error, and the results actually fall within the margins?

I know it is true as I am familiar with the evidence without having to search for it, you can choose not to educate yourself if you like though. If you wish to think a poll with a 3.5% MOE at 95% confidence is actually correct 95% of the time then fine by me. You'll be wrong, but not my job to fix your credulity if you are too lazy to find it yourself.

You may Google it if you like, or you may look in some scholarly databases if you want to be more rigorous. Seek and ye shall find

To the best of my knowledge, your claims are made out of ignorance. For all you have been able to demonstrate, you don't know how the various pollsters calculated their margins of error. If your claims were true about all the pollsters erroneously calculating their margins or error, then I'm certain you would be happy to substantiate them.

I'll ask again: So you're saying that the polls used in the RCP and 538 averages are not anonymous?

See previous answer.

So you can't figure out what you think is an "anonymous" poll and what is not?

I don't know why you spew such stuff about polls being wrong because they're not "anonymous" when your idea just has no connect to reality whatsoever. Obviously you cannot link to a single "anonymous" poll that is supposedly more accurate than the polls used by RCP and 538.

 

[Augustus]

To the best of my knowledge, your claims are made out of ignorance. For all you have been able to demonstrate, you don't know how the various pollsters calculated their margins of error. If your claims were true about all the pollsters erroneously calculating their margins or error, then I'm certain you would be happy to substantiate them.

Why is it ignorance to not accept polls as being highly accurate, but it isn't ignorance to purposely avoid looking for information that challenges your own initial assumptions despite being told such information exists and is readily available?


Disentangling Bias and Variance in Election Polls

Here we empirically analyze 4,221 polls for 608 state-level presidential, senatorial, and gubernatorial elections between 1998 and 2014, all of which were conducted during the final three weeks of the campaigns. Comparing to the actual election outcomes, we find that average survey error as measured by root mean square error (RMSE) is approximately 3.5%, corresponding to a 95% confidence interval of ±7%—twice the width of most reported intervals


"The history of polling data with established institutions provides evidence that the spread in candidate preferences and the reported margin of error represents less than a 95% confidence interval."

Should More Polls Be Interpreted as Too Close to Call?
Hill, Walter W. PS, Political Science & Politics46.2 (Apr 2013): 329-332.

What the hell does that mean? Show me "the history of 'chance of winning,'" and I'll look at it.

Its on his website, on the very prominent, big colourful graph.


From another polling organisation:

[Polling organisation] reminds the reader that the results of this survey should be interpreted as significant information on the current relative performances of candidates for the forthcoming presidential elections. On no account could it be considered as constituting a predictive element of the results of the election day.

Make sense?

So you can't figure out what you think is an "anonymous" poll and what is not?

Seems like you can't read. Forget it.

I don't know why you spew such stuff about polls being wrong because they're not "anonymous" when your idea just has no connect to reality whatsoever. Obviously you cannot link to a single "anonymous" poll that is supposedly more accurate than the polls used by RCP and 538

Again seems like you can't read

You're missing the point.

It isn't that there are "magical accurate polls hidden somewhere", it is that there is overconfidence in the accuracy of polls in general.

Simple enough?

 

[Nous]

Disentangling Bias and Variance in Election Polls

Here we empirically analyze 4,221 polls for 608 state-level presidential, senatorial, and gubernatorial elections between 1998 and 2014, all of which were conducted during the final three weeks of the campaigns. Comparing to the actual election outcomes, we find that average survey error as measured by root mean square error (RMSE) is approximately 3.5%, corresponding to a 95% confidence interval of ±7%—twice the width of most reported intervals

Thank you. I wonder why providing that link was so difficult for you.

The authors of this (for some reason, unpublished) study used state-level data for gubernatorial, senatorial and presidential elections from 1998-2014, and, by comparing poll results to the actual election outcomes, found an average root mean square error (RMSE) of approximately 3.5%, corresponding to a 95% confidence interval of +/-7%, significantly higher than than the typical simple random sample error of 1.9%.

But the authors also note that--for reasons they were apparently unable to determine--the results of Presidential polls are much more accurate. For more reasons than one, including according to the findings of this study, it would obviously be erroneous to generally apply an error rate of +/-7% to the polls in the race for President.

The authors state that “reported margins of error typically only capture sampling variability, and in particular, generally ignore errors in defining the target population (e.g., errors due to uncertainty in who will vote).” Nevertheless, the authors contrarily acknowledge that polls attempt to screen for “likely voters,” which is indeed reported for many of the polls for President.

It isn't that there are "magical accurate polls hidden somewhere", it is that there is overconfidence in the accuracy of polls in general.

So your claim that the unnamed “anonymous polls” that “have tended to report higher numbers for Trump” has nothing to do with the polls being “anonymous”?

 

[Underhill]

I suspect Trump will go down like Mondale in 1984 and lose by 20 million votes.

 

[Augustus]

Thank you. I wonder why providing that link was so difficult for you.

Because it is easy enough to find yourself, rather than expecting others to do the legwork for things they already know but don't have to hand.

The authors of this (for some reason, unpublished) study used state-level data for gubernatorial, senatorial and presidential elections from 1998-2014, and, by comparing poll results to the actual election outcomes, found an average root mean square error (RMSE) of approximately 3.5%, corresponding to a 95% confidence interval of +/-7%, significantly higher than than the typical simple random sample error of 1.9%.

But the authors also note that--for reasons they were apparently unable to determine--the results of Presidential polls are much more accurate. For more reasons than one, including according to the findings of this study, it would obviously be erroneous to generally apply an error rate of +/-7% to the polls in the race for President.

The authors state that “reported margins of error typically only capture sampling variability, and in particular, generally ignore errors in defining the target population (e.g., errors due to uncertainty in who will vote).” Nevertheless, the authors contrarily acknowledge that polls attempt to screen for “likely voters,” which is indeed reported for many of the polls for President.

2nd article (for some reason, published) was about presidential polls.

So your claim that the unnamed “anonymous polls” that “have tended to report higher numbers for Trump” has nothing to do with the polls being “anonymous”?

Again, it was another reason why polls may be less accurate than you think, not a reason why Trump is going to win.

Again, there is potential evidence that people may be less willing to identify themselves with a 'controversial' candidate in person to person polling. This might help you get the point:

"pre-election polls consistently underestimate support for female candidates when compared to white male candidates. Furthermore, our results indicate that this phenomenon—which we dub the Richards Effect, after Ann Richards of Texas—is more common in states which exhibit traits associated with culturally conservative views of gender issues."

I'm Not Voting for Her: Polling Discrepancies and Female Candidates
Stout, Christopher, Kline, Reuben. Political Behavior33.3 (Sep 2011): 479-503.

 

[Lighthouse]

So you believe that the sample size of the polls is too small? How do you account for their consistency?

I simply believe that assuming a winner from polls is not very logical.

There are inconsistencies and unpredictable s as well, so I don't count for their consistency.

The IBD/TIPP many say has been the most consistent the past few elections, and I still wouldn't put too much stock in its accuracy that predicts a winner.

 

[Nous]

2nd article (for some reason, published) was about presidential polls.

None of the recent polls at RCP show Clinton ahead by double the the spread of their margins or error: http://www.realclearpolitics.com/ep...s/general_election_trump_vs_clinton-5491.html

So what is the basis for your claim that "Clinton is the favorite"? The OP asks for a prediction--is your prediction that she will win?

Again, there is potential evidence that people may be less willing to identify themselves with a 'controversial' candidate in person to person polling.

I am unaware that people are asked to identify themselves in the phone polls such as commonly conducted.

You still haven't cited any "anonymous" poll in this election that is more accurate than the commonplace phone polls.

 

[Nous]

So you believe that the sample size of the polls is too small? How do you account for their consistency?

I simply believe that assuming a winner from polls is not very logical.

There are inconsistencies and unpredictable s as well, so I don't count for their consistency.

The IBD/TIPP many say has been the most consistent the past few elections, and I still wouldn't put too much stock in its accuracy that predicts a winner.

By "consistency," I meant the fact that the great majority of polls in the current election show Clinton ahead.

 

[Augustus]

None of the recent polls at RCP show Clinton ahead by double the the spread of their margins or error: http://www.realclearpolitics.com/ep...s/general_election_trump_vs_clinton-5491.html

So what is the basis for your claim that "Clinton is the favorite"? The OP asks for a prediction--is your prediction that she will win?

A trend across multiple polls, even within the margin of error, could make one the favourite.

She'll probably win, but not much point in making predictions until closer to the election imo. Knowing polling figures 3 weeks ago was useless to predict polling figures today, so why are today's ones any different?

I am unaware that people are asked to identify themselves in the phone polls such as commonly conducted.

You still haven't cited any "anonymous" poll in this election that is more accurate than the commonplace phone polls.

Do you actually read my replies?

As I previously clarified, I meant anonymous = not person to person.

And for about the 4th time, I have never claimed, implied or insinuated that there are "more accurate" polls, just that there might be a possibility that some people don't want to admit supporting Trump.

 

[Nous]

A trend across multiple polls, even within the margin of error, could make one the favourite.

That's what I would say. When poll after poll, conducted by diverse organizations, week after week, show with rare exception the same candidate ahead, even when the lead is within the margins of error, it would seem to become increasingly unlikely that these organizations are committing the same error or different errors with the same effect. The Shirani-Mehr study does not address this issue of a consistent trend.

She'll probably win, but not much point in making predictions until closer to the election imo.

There is little value in making accurate predictions for which there is a near certainty; it shows little insight. I predict the sun will not explode tomorrow. I wonder how much someone would pay me for that prediction.

I (and other people) were predicting a win for Clinton months ago, before either of them even got their party nominations.

Do you actually read my replies?

As I previously clarified, I meant anonymous = not person to person.

And for about the 4th time, I have never claimed, implied or insinuated that there are "more accurate" polls, just that there might be a possibility that some people don't want to admit supporting Trump.

Anonymous polls have tended to report higher numbers for Trump than those carried out personally which might suggest there is something to it.

So, by this statement, you were not trying to indicate that these (so far, unidentified) "anonymous" polls reflect a truer level of support for Trump than the polls used by RCP and 538?

 

[Augustus]

So, by this statement, you were not trying to indicate that these (so far, unidentified) "anonymous" polls reflect a truer level of support for Trump than the polls used by RCP and 538?

5th time lucky...

Different, not necessarily more accurate.

I don't know if they are more accurate or not. Different polls using 'correct' methodolgy will arrive at different results even from the same data set, never mind using a different sample.

It could simply be randomness, these polls might be based on 'correct' methodologies that favour Trump marginally, it could be to do with sample construction, or something else.

My original point was that it might be evidence that more people would vote for Trump than were willing to admit it, as you asked me what 'rational' reasons there were for saying there was a possibility that voters don't always reveal their true preferences in polls. This was one reason that such an effect might exist, not evidence that it does exist.

Understand?

 

[Augustus]

That's what I would say. When poll after poll, conducted by diverse organizations, week after week, show with rare exception the same candidate ahead, even when the lead is within the margins of error, it would seem to become increasingly unlikely that these organizations are committing the same error or different errors with the same effect. The Shirani-Mehr study does not address this issue of a consistent trend.

There are non-overlapping margins of error though which means at least some of these polls are doing something significantly wrong (or the margin of error should be significantly larger as I said before).

I agree that the overall trend makes Hillary the favourite, my initial point though was that it is far less of a certainty than many people are claiming.

 

[Atman]

I imagine Trump will likely lose, but the "alt-right" movement that propelled him to the position of Republican nominee will likely survive, and begin to define itself in more concrete terms

 

[Nous]

There are non-overlapping margins of error though which means at least some of these polls are doing something significantly wrong (or the margin of error should be significantly larger as I said before).

I agree that the overall trend makes Hillary the favourite, my initial point though was that it is far less of a certainty than many people are claiming.

I'm not sure how much “certainty” in the polls anyone has relied on or claimed as the reason behind their predictions here.

There are several reasons to doubt that the individual poll results you've cited here are informative about who will win the election. For instance, they all calculate a single national result. If the phone numbers called are random, then more people in “solidly red” "solidly blue"* states are called, and perhaps (as far as I know), the answers from more people (“likely voters”) in these states make up the results. That doesn't help us.

On the other hand, if the poll results are made up of similar numbers of people in each state (i.e., if each state is represented equally in a poll's final tally), this still skews the results.

The fact is that regardless of how the random calling and surveying is conducted or weighted, national polls are inherently not very informative of the possible electoral tallies. Attempting to identify which way the battleground states are going is more informative. And this election is so odd, the battleground states have changed since the nominations, and even in recent weeks.

In any case, my earlier predictions in this election--i.e., prior to either Clinton or Trump becoming the nominees--obviously had nothing to do with such polls. Except in those rare cases where there is foreknowledge of a landslide win, the election of the President is determined by a handful of non-partisan swing voters in a few states. (I am ever so slightly a swing voter, at least outside of Presidential elections, and I definitely would have voted for Romney over Clinton for President, or almost any other Democrat than Obama. Romney is just one of the excellent Republican Governors Massachusetts has had in recent decades.) Anyway, I believe non-partisans engage in a lot of serious consideration of the issues and the candidates--they don't make their decisions on frivolous grounds. And in this election, there is just no comparison between the candidates. One is a former First Lady married to a surprisingly popular former President and who was highly involved in her husband's administration; an attorney; an elected representative who was Secretary of State. The other is a real estate/resort/casino developer (with a dubious history of success) and “reality” TV star who has never expressed any interest in public service or being a representative of the people in any capacity, except now in the most prestigious and powerful office of the nation; he doesn't seem to have the attention span necessary to stay on script; according to the ghost writer of his first book, he does not read books. One can go on and on with such contrasts--Trump doesn't come out more favorably than Clinton in any comparison. Show me 5 people who are non-partisan swing voters who are unequivocally for Trump.

Obviously the trend or consistency of the current polls influence my prediction in this race at this point, but I would nevertheless be highly skeptical if the consistency or trends-within-MoE showed Trump in the lead.

5th time lucky...

Different, not necessarily more accurate.

What does "it" refer to in this sentence:

"Anonymous polls have tended to report higher numbers for Trump than those carried out personally which might suggest there is something to it."

?

Obviously it doesn't make sense that "it" refers to something such as "the idea that there is a difference".



* I commonly get mixed up on which party is "red" and which is "blue".

 

[Nous]

Just to briefly update: http://www.270towin.com/ has a fun interactive electoral map. Note that Trump has to win most every current close state in order to get 270 electoral votes. For instance, if he doesn't win AZ, he must win every other battleground state--though he won't need NE's one district.

Similarly, if he doesn't win WI, then he must win all other battleground states, though he wouldn't need NE's one district.

Unless something changes, these seem to be very unlikely scenarios.

 

[Monk Of Reason]

Its pretty nail in the coffin that HIllary will win. This isn't just some bias as I'm not voting for her or Trump. But she has almost all of the swing states with a clenched fist. Arizona and Florida are the only two really battle worthy states but Florida has leaned Clinton the whole time nearly and there is no doubt that wont' change.

 

[Revoltingest]

Articles like this'n make it impossible for me to make any predictions....
https://www.washingtonpost.com/news...pect-her-to-win-post-abc-tracking-poll-finds/

 

[idav]

Articles like this'n make it impossible for me to make any predictions....
https://www.washingtonpost.com/news...pect-her-to-win-post-abc-tracking-poll-finds/

Is that a tough one when 6 in 10 think Hillary will win and only 3 in 10 expect trump to win. I take it the other 2 in 10 think that someone else will win, those people are very optimistic along with the 1 in 10 who think trump will win easily.

 

[Nous]

Articles like this'n make it impossible for me to make any predictions....
https://www.washingtonpost.com/news...pect-her-to-win-post-abc-tracking-poll-finds/

What is it about this poll that causes you to think Trump has an equal chance of winning? It shows Clinton favored by a margin larger than the margin of error, among people 63% of whom did not identify as Democrats. How does one get a Trump win out of that?

In any case, as noted above, national polls have very little relevance at this point. Is it plausible that Trump is going to win all of these states: NC, GA, FL, OH, WI, IA, NV, AZ, UT, NH?