Author Topic: How should we assess differences between polls and voting results?  (Read 9711 times)

Greg Davidson

  • Members
    • View Profile
What if there is a substantial difference between polling results and actual outcomes? There will be a strong temptation to make accusations of cheating or fraud. There may be some ways to rule out cheating as an explanation for divergence between polling and actual votes. First, let's identify the theoretically possible reasons why polling and actual votes may diverge. There are huge differences of opinion over how likely each of these are, and while I believe that the case is strong for some and weak for others, let's take them all as potential explanations that some Americans will believe as a reason for a gap between polling and actual vote totals:

  • Voting officials cheat in some way with ballots cast to give more votes to Trump or Clinton
  • Illegitimate votes cast by non-citizens or multiple votes cast by the same individuals are accepted and counted
  • Reductions in polling sites and other barriers to voting reduce the turn-out for those most affected 
  • The difference in the efforts by the two candidates in terms of investment/staffing for get-out-the-vote efforts delivers a higher share of potential voters to the polls for the candidate who had the biggest ground game
  • External entities somehow hack the voting machines to favor one candidate or the other
  • People hesitant to tell pollsters that they are Trump or Clinton voters

All but the last of these should vary significantly based on State. No matter how likely or unlikely we believe it is that voting officials will cheat in the ballot count, I think we can all agree that it is even less likely that Republicans would cheat to help Clinton or Democrats would cheat to help Trump.

The second potential gap between polling and actual results (of fraudulently cast votes) has been of strong concern to Republicans. Voting laws have changed in part to address this concern, and even though some have been thrown out by the court, there remain significant differences by state in ease of access at the polling place.  If there are gaps between polls and actual results because of fraudulent votes, it is logical to assume that this would occur most in Democratically controlled states with the least stringent voter ID laws and least in Republican-controlled states that have added voter ID laws.

The third potential gap (impediments to voting) has been of strong concern to Democrats, and so if this were to occur, it would be seen by Democrats under-performing relative to polls most in Republican-led states that have reduced access to polling or increased the stringency of voter ID laws.

The fourth potential gap (differential GOTV efforts) would be seen most prominently in states where the difference between Clinton and Trump (number of campaign offices, number of paid staff)  has the biggest gap relative to previous elections.

The fifth potential gap (external hackers) should vary widely over states, because there are so many different voting systems. If there is a gap between polls and actual votes due to hacking, we might see a tendency for states with similar voting technologies would be similarly affected (although we might only see swing states with similar technologies to be hacked).

The final potential gap (people not telling pollsters their true favored candidate) should not be correlated by state. Republicans sometimes assert that support for Trump is not adequately reflected in polls because he is not seen to be politically correct; Democrats sometimes assert that there is a significant population of conservative women who are keeping their support of Clinton to themselves. If there is a nationwide trend of actual vote counts diverging from polling, then it would most support one of these hypotheses for voters not revealing their true intentions to pollsters. 

Assuming that there is a gap between polling and actual results, we could use this analysis to identify which potential causes for that gap are the most plausible.  I have tried to analyze this in a way that is fair to all sides; can anyone think of something major that I have missed?
« Last Edit: November 04, 2016, 01:44:06 PM by Greg Davidson »

Wayward Son

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #1 on: November 04, 2016, 02:11:11 PM »
You forgot the most obvious one: normal polling error.

On top of all those other possible factors, the polls themselves are normally only accurate to (+/-) 2 - 3%, when done correctly.

And we won't even talk about the possible errors in exit polling!  ::)

AI Wessex

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #2 on: November 04, 2016, 02:18:25 PM »
Going beyond Wayward's comment, the two parties appear to recognize and legitimize different polls.  For instance, after each of the debates Trump's campaign relied on on-line polls that showed him having won each by up to a 90-10 split.  They in turn dispute the findings or more traditional phone polls.  In other words, Trump has set up the situation (no surprise) to claim that the system is rigged if he doesn't win because his polls are honest and reliable and the actual voting was dishonest and rigged.  His modus operandi is heads I win, tails you lose.

Romney had the same problem in 2012, where on the day of the election Karl Rove and other conservative media were predicting a Romney landslide, claiming that pollsters like FiveThirtyEight were wrong because the polls they relied on "oversampled Democrats".  What Romney didn't do that I expect Trump will do is to then take the next illogical step and declare the vote rigged.

Greg Davidson

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #3 on: November 04, 2016, 02:32:33 PM »
Statistical error in polling should not be significant, because there are so many different polls, that should be taken care of in the aggregate.

Poll response errors is another significant one; non-English speaking citizens may be under-represented in polling, in which case the gaps should occur primarily in states with those (primarily Hispanic) non-native speakers live. While there might be some overlap between those States and States where Republicans might believe there are substantial numbers of illegitimate votes, we can look to see if there is any difference based on stringency of Voter ID laws.


Wayward Son

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #4 on: November 04, 2016, 03:30:12 PM »
Statistical error in polling should not be significant, because there are so many different polls, that should be taken care of in the aggregate.

I'm not sure what you mean by this, since the individual polls themselves vary wildly (if you consider a +/- 12% or more from the mean "wild" :)).  The +/- 2% or so is the historically expected error of the aggregate.

TheDrake

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #5 on: November 06, 2016, 06:38:47 PM »
The problems with polling are large and getting larger.

1. Access. Do people have a way to be contacted for a poll?
2. Willingness. Who will take the time to respond?
3. Truthfulness. Voters may not report who they actually support, or may not accurately report their likelihood of voting.
4. Demographic projection. Because polling usually only gets very limited responses, assumptions are made about how representative someone is, and how many will vote like them.
5. Many polls don't include 3rd party, which matters more this year when Johnson isn't mentioned.
6. Acts of "god". Philadelphia public transit striking, dangerous and unpleasant weather, etc.
7. Voter intimidation. If reports are coming in during the early hours of voting, many may stay away later in the day.

I prefer to go to realclearpolitics.com for my polling info. A good example is the recent NBC news/wall street journal survey. They only got 1282 interviews and project nationally. Yet they claim a 3% error. There's a lot of magic in how they select the calls, and a lot of luck in who responds. It is actually remarkable that polls have anywhere near an accurate number.

Convenience and the predicted outcome within a state are particular multipliers. In 1972, 1992, and 2000 a Journal of Politics study concluded that hundreds of thousands of voters stayed home because of inclement weather.

In the sample size, state by state predictions are particularly dicey, and of course that's what matters. Very few publicized polls are conducted within a specific state. You can see the "freshness" of these right here

Before I ascribed any error to election fraud, it would have to be a pretty massive difference. Remembering first that +/-3% is normal, it wouldn't be farfetched to see this a 50-44 lead flipped on election day. Then, remember that the confidence level is 95% - meaning that 1 out of 20 times, we'd expect the margin to be exceeded. That's at least two states right there.

Of course, if fraud were occurring, it is highly likely that it would not be in an area where the margin is that vast. My conclusion - polling variance can't really tell you anything about fraud. That doesn't mean it won't be used that way...

Fenring

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #6 on: November 06, 2016, 09:49:00 PM »
I like the kind of breakdown Greg has made here, even though I'm not convinced it's as simple as he lays out to deduce things from the data. I do appreciate, at any rate, the desire to mine the data. I'd add one item to the list, though:

7. The numbers presented as the result of a vote are literally made up and aren't a result of a correct count skewed by particular interference (throwing out votes, vote flipping, etc.)

Maybe there's a way to prove in some states this cannot have happened, but I thought I'd throw it out there as a 'possible' means of cheating the result. Maybe this would only be plausible in states that use electronic voting.

That said, I would suggest inspecting the scope of the question to not only examine why vote results differ from polls in general, but to specify why a vote result would significantly differ from exit polls. When conducted with a fairly large sample size exit polls are supposedly accurate to within a few percentage points, and some experts whose statements I've read have said that any result several standard deviations away from that cannot reasonably be expected to be legitimate. I haven't done the math personally, but if this is true then exit polls should be a far better indicator of true results than any other type. Online polls are basically BS, phone polls suffer from some of the issues that others have specified (dishonesty, lack of representation depending on dominant language), and the same goes for any other type. Exit polls will suffer from the issue of willingness to admit having supported someone questionable, but avoids many of the other polling pitfalls. Additionally, it accounts fully for things like last minute changes in decision, bad sampling, biased questions, and even probably mostly eliminates trolling.

I would be quite satisfied if extensive exit polling was done, in addition to getting better control of the electronic voting security issue. The other issues Greg brings up are important, but I'm not sure I have an informed opinion on them.

Greg Davidson

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #7 on: November 06, 2016, 11:53:32 PM »
Exit polls don't cover early voters, and you still have the issue that not all people will be equally willing to participate in an exit poll.  For example, one hypothesis might be that some women in some pro-Trump communities may be culturally adverse (or literally afraid) to describe to a pollster that they have voted for Hillary Clinton in the privacy of a voting booth.

TheDrake

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #8 on: November 07, 2016, 09:08:18 AM »
About exit polls

Quote
The imperfections of the exit polls are not hard to show. Here are two quick examples, based on official voter turnout statistics:

The exit polls showed that voters over age 65 were 18 percent of the electorate in Iowa in 2008, but 26 percent in 2012. The official state turnout statistics instead show that the share increased to 23.6 percent, from 21.9 percent, over the same span.

In North Carolina, the exit polls showed that the black share of the electorate dropped to 23 percent in 2008, from 26 percent in 2004, and held steady at 23 percent in 2012. The state turnout statistics say the share rose from 18.6 percent in 2004 to 22.3 percent in 2008, and then to 23.1 percent in 2012.

Quote
How can the exit polls be off by so much? The biggest thing to remember is that they’re just polls! They’re usually based on a sample of a few dozen precincts or so in a state, sometimes not even including many more than 1,000 respondents. Like every other type of survey, they’re subject to a margin of error because of sampling and additional error resulting from various forms of response bias.

D.W.

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #9 on: November 07, 2016, 09:28:46 AM »
I for one would never answer an exit poll. 

Wayward Son

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #10 on: November 07, 2016, 10:57:16 AM »
Remember, Ferning, according to exit polls, Al Gore won Florida by something like 6 points, or about 6 million votes. :)

Exit polls are not very reliable.

Fenring

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #11 on: November 07, 2016, 11:33:00 AM »
Of course exit polls aren't totally reliable. But at least they sample people who've actually voted. TheDrake's quote is quite accurate, and indeed the problem with most exit polling is the small sample size. If less than 1,000 people are sampled then it's not worth very much IMO. It has to be done seriously or not at all. Polls taken prior to a vote have many purposes, which can include doing a litmus test for campaign policy, making predictions, in the case of biased polling to create an apparent result, and to create news. A poll done after a vote, however, doesn't really help with any of these other than with creating news releases faster than the results come in. What it can do, however, is act as a sort of incomplete verification of how the vote went, which has more of a systemic use than early polling does.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #12 on: November 07, 2016, 04:42:12 PM »
What if there is a substantial difference between polling results and actual outcomes? There will be a strong temptation to make accusations of cheating or fraud. There may be some ways to rule out cheating as an explanation for divergence between polling and actual votes. First, let's identify the theoretically possible reasons why polling and actual votes may diverge.

How can you make an exhausitive list without including the largest reasons for divergence?  Poorly constructed polls, lack of rigor in selection criteria/methodology and/or inability to randomly sample.  Even legitimate divergences in whether you attempt to poll "likely" voters or all potential voters can be decisive.

Quote
Quote
Voting officials cheat in some way with ballots cast to give more votes to Trump or Clinton
All but the last of these should vary significantly based on State. No matter how likely or unlikely we believe it is that voting officials will cheat in the ballot count, I think we can all agree that it is even less likely that Republicans would cheat to help Clinton or Democrats would cheat to help Trump.

Of course the highest risk for this kind of cheating is in districts in swing or battleground states, or states with a large urban vs non-urban party disparity where one party is so dominant that the other can't even bring in local observers but has to bring them in from outside the community.  It's no accident that Chicago and Philadelphia have had voter fraud issues and are the target for such complaints.

Quote
Illegitimate votes cast by non-citizens or multiple votes cast by the same individuals are accepted and counted

Not sure why you'd think illegal alien voting would cause a disparity between polls and voting?  Unless you think illegal aliens would identify themselves routinely to pollsters as illegal?  The concern with illegal voting is that it influences an election inappropriately not that it doesn't match the polling.

Multiple votes by a single individual, would on the other hand, if accomplished at scale and in a single direction potentially have an impact.

Quote
The second potential gap between polling and actual results (of fraudulently cast votes) has been of strong concern to Republicans. Voting laws have changed in part to address this concern, and even though some have been thrown out by the court, there remain significant differences by state in ease of access at the polling place.  If there are gaps between polls and actual results because of fraudulent votes, it is logical to assume that this would occur most in Democratically controlled states with the least stringent voter ID laws and least in Republican-controlled states that have added voter ID laws.

Voting laws were not changed to confront the point of differences between polling and voting results, nor would such differences necessarily or even likely identify a problem with illegal aliens voting.  This claim doesn't really have anything to do with your point, so I won't argue it further.

Quote
Quote
Reductions in polling sites and other barriers to voting reduce the turn-out for those most affected 
The third potential gap (impediments to voting) has been of strong concern to Democrats, and so if this were to occur, it would be seen by Democrats under-performing relative to polls most in Republican-led states that have reduced access to polling or increased the stringency of voter ID laws.

This is the Democratic fake issue they use to argue against Republican concerns  about voter fraud.  In all honesty, no matter how many polling places have been reduced the coverage map is still completely adequate to allow everyone to vote on election day, and when you add in the ever increasing emphasis on early voting and mail in voting, the real access to voting has done nothing but increase.

Quote
Quote
The difference in the efforts by the two candidates in terms of investment/staffing for get-out-the-vote efforts delivers a higher share of potential voters to the polls for the candidate who had the biggest ground game
The fourth potential gap (differential GOTV efforts) would be seen most prominently in states where the difference between Clinton and Trump (number of campaign offices, number of paid staff)  has the biggest gap relative to previous elections.

This is legit and can definitely make a difference where it converts a voter who wasn't likely to vote into one that does. 

I'd add this is one area that the Dems have had a large advantage, they have a much easier time identifying clumps of likely voters that they can drive to the polls than Republicans do.

Quote
Quote
External entities somehow hack the voting machines to favor one candidate or the other
The fifth potential gap (external hackers) should vary widely over states, because there are so many different voting systems. If there is a gap between polls and actual votes due to hacking, we might see a tendency for states with similar voting technologies would be similarly affected (although we might only see swing states with similar technologies to be hacked).

You are completely wrong in how you'd trace the impact, you'd be highly unlikely to see similar impacts based on voting machines in different areas (unless you posit absolutely incompetent hacking).  All hacking based manipulation can be expected to be concentrated in key areas where there is a good potential to hide the impact.  If you found the same manipulations in districts where hacking would be completely useless that would in fact be shocking.

Quote
Quote
People hesitant to tell pollsters that they are Trump or Clinton voters
The final potential gap (people not telling pollsters their true favored candidate) should not be correlated by state. Republicans sometimes assert that support for Trump is not adequately reflected in polls because he is not seen to be politically correct; Democrats sometimes assert that there is a significant population of conservative women who are keeping their support of Clinton to themselves. If there is a nationwide trend of actual vote counts diverging from polling, then it would most support one of these hypotheses for voters not revealing their true intentions to pollsters.

Except this one is in fact an observed phenomena, for whatever reason, some people do misrepresent their position to pollsters.  I don't think its hesitancy so much as it wanted to present themselves publically as on the "correct" team even if privately they disagree and will vote otherwise.  I certainly know some women who absolutely will not say they are voting for Trump publically, because they have other friends who'd consider them a gender traitor.  Have no idea what they'd tell a pollster.

Quote
Assuming that there is a gap between polling and actual results, we could use this analysis to identify which potential causes for that gap are the most plausible.  I have tried to analyze this in a way that is fair to all sides; can anyone think of something major that I have missed?

If there is a gap, which there almost has to be given some of the margins for error (heck some polls are not within each others' margin of error), then you'll most likely be able to glean why only by looking at the individual poll mechanics.  The most likely cause will be sampling error.

Voter fraud and manipulation may, but is unlikely to, show up in such an analysis.  End of day, fraud can be effective and still be within the margin of error for a poll.

Greg Davidson

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #13 on: November 08, 2016, 01:32:18 AM »
Quote
In all honesty, no matter how many polling places have been reduced the coverage map is still completely adequate to allow everyone to vote on election day, and when you add in the ever increasing emphasis on early voting and mail in voting, the real access to voting has done nothing but increase.

Seriati, tell us all tomorrow how long you had to wait in order to vote. Then let's look at the longest lines in the country. Here's a prediction of mine: the longest lines will not be randomly distributed, they will not be in some state like Oklahoma or Massachusetts. Instead, they will be in areas with minority voters where Republicans are in charge of deciding how many polling places to set up. Would you be willing to bet otherwise?

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #14 on: November 08, 2016, 02:57:30 AM »
Quote
In all honesty, no matter how many polling places have been reduced the coverage map is still completely adequate to allow everyone to vote on election day, and when you add in the ever increasing emphasis on early voting and mail in voting, the real access to voting has done nothing but increase.

Seriati, tell us all tomorrow how long you had to wait in order to vote. Then let's look at the longest lines in the country. Here's a prediction of mine: the longest lines will not be randomly distributed, they will not be in some state like Oklahoma or Massachusetts. Instead, they will be in areas with minority voters where Republicans are in charge of deciding how many polling places to set up. Would you be willing to bet otherwise?

Yes, I would.  Based on past history, they will be in Urban centers that have been largely dominated by Democrats on a local level.  There are literally tens of thousands of polling places, there is no chance that zero will have problems with lines, some of which will be tied to break downs in voting machines, or volunteers not turning up for shifts, and perhaps a few with dramatically higher voter turnout than experienced in past elections.  Given that it would literally be impossible to avoid any such situation, and they realistically can only occur in high density polling locations, what is a reasonable rate in your view?  I strongly suspect that you will literally not be able to positively identify any as the result of deliberate malice.

I don't generally have to wait, but my state prohibits early voting and I believe even mail in voting without cause (and is heavily controlled by Democrats), so they have little choice but to maintain a large number of polling places.

AI Wessex

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #15 on: November 08, 2016, 07:01:53 AM »
You have to close your eyes to pretend not to see that the reduction in voting places is partisan for the most part.  Here's one article with examples:
Quote

Consider North Carolina, the epicenter of GOP voter suppression. Just days after the Supreme Court gutted the Voting Rights Act in Shelby County v. Holder—thereby allowing the state to restrict voting without federal oversight—the overwhelmingly Republican Legislature requested data of voting preferences by race. The legislators then promptly passed an omnibus bill that, in the words of the U.S. Court of Appeals for the 4th Circuit, “target[ed] African Americans with almost surgical precision.” A key provision of the new law slashed early voting, including several days of Sunday voting, which black voters favored. As the state explained in court with startling candor, “counties with Sunday voting” were “disproportionately black” and “disproportionately Democratic”—a fact that persuaded the Legislature to severely cut back early voting.

In July, the 4th Circuit blocked this legislation from taking effect this election cycle, holding that it violated both the Voting Rights Act and the Equal Protection Clause as a race-based voting restriction. But Republican-controlled county election boards implemented the early voting cuts anyway. (These boards retain power over county-level voting practices and claimed they had decided independently to roll back early voting, not implement the now-blocked law that did the exact same thing.) The boards colluded to cut early voting—and Sunday voting especially—in Democrat-heavy areas, as GOP election board chairmen urged each other to follow the “party line.” When one Republican chairman agreed to open a Sunday voting site where black voters could cast ballots after church, his fellow GOP chairmen called him a “traitor.” The transparency of these suppression efforts is staggering: On Sunday, the North Carolina GOP boasted that black early voting is down this year, when that fact is almost certainly attributable to Republicans’ own attack on black voting.

Ultimately, North Carolina counties cut an astonishing 27 voting sites altogether this year and dramatically reduced early voting hours in many of the remaining sites. The result was entirely predictable: monstrously long lines that force voters—many of them elderly—to stand outside for hours upon hours just to cast a ballot. As the Nation’s voting rights expert Ari Berman has reported, the story is similar in other states previously covered by the Voting Rights Act, including Texas (403 poll closures since Shelby), Louisiana (103), and Alabama (66). The problem is especially acute in Arizona, which, as Berman reports, “reduced the number of polling places by 70 percent from 2012 to 2016, from 200 to just 60—one polling place per 21,000 registered voters.” And in Ohio, GOP-instituted cuts to early voting in Cincinnati created a half-mile line of 4,000 people that snaked under an interstate and through a public park.

TheDrake

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #16 on: November 08, 2016, 07:32:37 AM »
I think people should have a right to a speedy vote like the right to a speedy trial. However, we're not doing so well on the speedy trial thing either in heavily populated areas, despite the 1974 speedy trial act. Here in my part of Texas (predominantly white suburb) I have three polling stations within a mile radius. Hmmm.

TheDeamon

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #17 on: November 08, 2016, 09:03:06 AM »
Quote
In all honesty, no matter how many polling places have been reduced the coverage map is still completely adequate to allow everyone to vote on election day, and when you add in the ever increasing emphasis on early voting and mail in voting, the real access to voting has done nothing but increase.

Seriati, tell us all tomorrow how long you had to wait in order to vote. Then let's look at the longest lines in the country. Here's a prediction of mine: the longest lines will not be randomly distributed, they will not be in some state like Oklahoma or Massachusetts. Instead, they will be in areas with minority voters where Republicans are in charge of deciding how many polling places to set up. Would you be willing to bet otherwise?

I would conjecture that the favored polling locations with long lines that see air time will be in racial minority areas, as the media will be looking for it, and it'll get eyeballs. "Look at all these poor minorities waiting for their turn to vote."

While there will likewise be a number of mostly white polling locations that likewise see long waiting times, particularly as we get into the evening hours.

The other thing to watch out for (un)intentional "voter bombing" by GOTV efforts. On the unintended side, we have the well meaning volunteers would may not begin their efforts until getting home from work at the end of a regular workday...

On the intentional side, a GOTV effort that is using multiple shuttles could opt to "accidentally" arrive in quick succession to each other at a polling place. There aren't many polling places that I've seen that could easily absorb 3 dozen plus voters turning up within minutes of each other. But hey, it'll make a nice photo op about "long lines at polls."

Now considering that Democrats have used charter and/or school buses for GOTV in the past, and most charter buses can hold about 40 people each.... Have three charter buses visit the same precinct within, say, 5 minutes of each other starting at 6PM, and see what that does to the line.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #18 on: November 08, 2016, 09:39:53 AM »
You have to close your eyes to pretend not to see that the reduction in voting places is partisan for the most part.

So, NC closed 27 voting sites this year, and that's evidence of something nefarious?  NC has over 100 counties, looked at 10 of them randomly, they averaged between 17 and 29 counties, with 2 outliers over 40.  You should be embarrassed by some of these arguments.

Do you have evidence that less early voting has disenfranchised any voters that want to vote?  Cause, frankly claiming that people need dozens and dozens of days to get their vote in, or else they are being disenfranchised is kind of nonsensical to me, when there are other states (that you don't care about) where you have to vote on the single election day.

Also nice to see that you admit and recognize that the Voting Rights Act is still in force and provides protection, even if you can't comprehend how ridiculous it is to apply a different standard to citizens who weren't even borne when their great great great grandparents were in charge and acted with discrimination.  If you really think that's an appropriate standard then you ought to hold the Democrats in contempt since they were the ones that applied those policies.

Quote
The problem is especially acute in Arizona, which, as Berman reports, “reduced the number of polling places by 70 percent from 2012 to 2016, from 200 to just 60—one polling place per 21,000 registered voters.” And in Ohio, GOP-instituted cuts to early voting in Cincinnati created a half-mile line of 4,000 people that snaked under an interstate and through a public park.

You do recall I specifically checked Arizona for you, and found multiple counties with over 100 poling places each.  So you clearly have a broken quote here, or are just repeating nonsense.  Not to mention, you have complete disregard for whether polling places actual had heavy attendance in western states where voting by mail is far more common.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #19 on: November 08, 2016, 09:41:54 AM »
I would conjecture that the favored polling locations with long lines that see air time will be in racial minority areas, as the media will be looking for it, and it'll get eyeballs. "Look at all these poor minorities waiting for their turn to vote."

While there will likewise be a number of mostly white polling locations that likewise see long waiting times, particularly as we get into the evening hours.

Already have anecdotals from friends with half hour or more waits in heavily white districts, see it on the news yet?  And have first person accounts (now second person since I'm telling you) of votes being switched from Trump to Hillary at polling places in Pittsburgh.

Greg Davidson

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #20 on: November 08, 2016, 10:17:23 AM »
Quote
While there will likewise be a number of mostly white polling locations that likewise see long waiting times, particularly as we get into the evening hours.

Quote
already have anecdotals from friends with half hour or more waits in heavily white districts, see it on the news yet?

Wow, half an hour!

I was talking about two or more hours, in some recent Presidential elections it has gone to 2-3 hours. I have never had to wait in line to vote that long.

And with this thing called the internet, I am pretty sure if there are 2-3 hour waits in mostly white polling locations, I am pretty sure that it will get mentioned.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #21 on: November 08, 2016, 11:28:11 AM »
Now I have a friend who waited over an hour and saw no signs of getting to vote so she left.  I think we have a very high turnout, which means even reasonable people who made adjustments to polls based on historical trends are going to be surprised with lines.  If you add on top maliciousness, I think you'll get your wish of being able to point to some very long lines.

Wayward Son

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #22 on: November 08, 2016, 11:32:54 AM »
Quote
And have first person accounts (now second person since I'm telling you) of votes being switched from Trump to Hillary at polling places in Pittsburgh.

Have you friend contact the Feds and report it.  No one wants anyone monkeying around with the vote, one way or another.

Maybe the Feds will do something.  Or maybe they won't.  But you got to try to fix problems like this, not just complain about them.

And let us know what happened.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #23 on: November 08, 2016, 11:53:07 AM »
Haven't heard the results but they did complain at the polling cite and to the authorities that they identified.  If I get an update will let you know.

AI Wessex

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #24 on: November 08, 2016, 11:54:10 AM »
Now I have a friend who waited over an hour and saw no signs of getting to vote so she left.  I think we have a very high turnout, which means even reasonable people who made adjustments to polls based on historical trends are going to be surprised with lines.  If you add on top maliciousness, I think you'll get your wish of being able to point to some very long lines.
You say this as if it doesn't bother her or you.  I'm thinking that your argument is that an anecdotal case where a friend of yours (a "good guy") didn't vote because of inadequate voting facilities cancels out everything that anyone else has said about intentional disenfranchisement and lines where people did wait possibly for hours to vote.  I guess reports of problems therefore don't amount to anything to worry or think about.

AI Wessex

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #25 on: November 08, 2016, 01:14:23 PM »
Quote
So, NC closed 27 voting sites this year, and that's evidence of something nefarious?
Actually, yes.  Would you like me to look it up for you?  You'd be amazed how much reliable information you can get on even the first page of a google search.  I'll even give you the search request: "north carolina voting site shut down".  Obviously, a more extensive search will give you even more good information about this problem.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #26 on: November 08, 2016, 03:24:44 PM »
AI, you are a one trick pony on these posts.  That's barely one location for every 4 counties, with each county having somewhere between 15 and 25 remaining polling places (or more).  If you can demonstrate a material impact, without just posting to some hack that didn't do any research, go for it.

AI Wessex

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #27 on: November 08, 2016, 05:19:54 PM »
I'm trying to be consistent in the face of your constant denials.  You could do your own checking, but if you don't want to you also don't have to respond if you are tired of it. 

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #28 on: November 08, 2016, 06:42:06 PM »
You mean like how I've literally checked multiple states and dozens of counties in response to what I can only describe as your baseless assertions?  Do some research yourself and quit just accepting everything that you get spoon fed.

Grant

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #29 on: November 09, 2016, 07:52:32 AM »
It seems to me that NC actually had 78 more early polling sites with a total of 5,868 more total early voting hours offered, they just opened these locations later then they usually did while keeping them open longer. 

https://www.ncsbe.gov/press-releases?udt_2226_param_detail=53
https://s3.amazonaws.com/dl.ncsbe.gov/One-Stop_Early_Voting/2016/2016_Early_Voting_Hours.pdf

This seems to me to be an odd way to suppress the early vote, offering more hours and more sites, but maybe someone else can glean some info from the data presented.  If voter statistics are correct, more blacks voted yesterday in NC than in 2012. 

https://enr.ncsbe.gov/voter_stats/

But, I mean, Slate says it so it's got to be true, right?

Grant

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #30 on: November 09, 2016, 07:57:31 AM »
I think people should have a right to a speedy vote like the right to a speedy trial. However, we're not doing so well on the speedy trial thing either in heavily populated areas, despite the 1974 speedy trial act. Here in my part of Texas (predominantly white suburb) I have three polling stations within a mile radius. Hmmm.

I think that's a nice sentiment, but it's certainly not in anyone's Constitution and certainly doesn't seem to be a natural human right (if you actually believe in such things).  There seems to be a difference between having a right, and having it be easy. 

scifibum

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #31 on: November 09, 2016, 02:52:49 PM »
After yesterday it seems like #6 was a strong factor.

Ronald Lambert

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #32 on: November 09, 2016, 05:13:17 PM »
Another cause for discrepancy between exit polls and actual votes was what was seen in Florida in 2000, when the infamous "butterfly" ballot was confusing in the way it was aligned, and led thousands to vote for Patrick Buchanan when they thought they were voting for Al Gore. In that case the exit polls accurately revealed how the people THOUGHT they had voted. That was enough to flip Florida for Bush, which was enough to give Bush the election nationwide.

AI Wessex

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #33 on: November 11, 2016, 12:18:23 AM »
Another cause for discrepancy between exit polls and actual votes was what was seen in Florida in 2000, when the infamous "butterfly" ballot was confusing in the way it was aligned, and led thousands to vote for Patrick Buchanan when they thought they were voting for Al Gore. In that case the exit polls accurately revealed how the people THOUGHT they had voted. That was enough to flip Florida for Bush, which was enough to give Bush the election nationwide.
That's a good point.  We could blame about half a dozen different things that if taken in isolation could explain why Gore lost.  My thinking is that Gore lost because Bush got more votes (where they counted the most), just like Clinton lost because Trump got more votes (where they counted the most).

Ronald Lambert

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #34 on: December 13, 2016, 05:03:36 PM »
It has been reported that the final result of the recount in Wisconsin is that Donald Trump gained an additional 162 votes. Link: http://ijr.com/2016/12/754565-final-results-are-in-from-wisconsin-recount-jill-stein-was-right-trumps-vote-count-was-off/?utm_source=email&utm_campaign=morning-newsletter&utm_medium=owned

Thank you, Jill Stein, for spending so much money to accomplish this result.

And in Michigan--

It is not just one precinct in Detroit that had a discrepancy between the number of ballots claimed and the number actually recorded by poll workers:

Quote
Green Party candidate Jill Stein’s Michigan recount unintentionally exposed major voter fraud in Detroit.

    Election officials in Michigan found that 37% of precincts in Detroit tabulated more ballots than the number of voters tallied by workers in the poll books.

    Hillary Clinton won Wayne County over Donald Trump 67% to 30%.

    State officials are planning to examine about 20 Detroit precincts where ballot discrepancies occurred.

    Detroit News reported:

    Voting machines in more than one-third of all Detroit precincts registered more votes than they should have during last month’s presidential election, according to Wayne County records prepared at the request of The Detroit News.

    Detailed reports from the office of Wayne County Clerk Cathy Garrett show optical scanners at 248 of the city’s 662 precincts, or 37 percent, tabulated more ballots than the number of voters tallied by workers in the poll books. Voting irregularities in Detroit have spurred plans for an audit by Michigan Secretary of State Ruth Johnson’s office, Elections Director Chris Thomas said Monday.
Link: http://dcwatchdog.org/what-voter-fraud-michigan-recount-uncovers-too-many-votes-in-37-of-detroit-precincts/

 The report said that state officials are going to investigate this. In other words, this is a matter of criminal investigation. They will be questioning every worker in those 20 precincts, asking them how they did not notice this, and how they could sign off on sealing boxes containing ballots that are labelled as having many more ballots than they actually contain. In at least one precinct, a box sealed and officially signed off by poll watchers, labelled as having 300 ballots, only contained 50 ballots. There is no way that could be a mistake.

D.W.

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #35 on: December 13, 2016, 05:15:21 PM »
An aunt who's a poll worker just commented about this.  Apparently the machines can get stuck and read a ballot multiple times.  And the workers need to sort this out...  Her location was much smaller scale than Detroit.  What a mess.  :(

The thing I don't get, living in MI, is how a recount got stopped in part because of this discrepancy.  Wouldn't this just confirm that a hand count is indeed needed?

Not that I expect the outcome to change, but exposing inconsistencies is a pretty huge deal and we should be thanking JS for uncovering it. 

TheDeamon

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #36 on: December 13, 2016, 06:18:57 PM »
An aunt who's a poll worker just commented about this.  Apparently the machines can get stuck and read a ballot multiple times.  And the workers need to sort this out...  Her location was much smaller scale than Detroit.  What a mess.  :(

Or that two ballots get "stuck" together and go through the reader at the same time, resulting in "skipped" ballots. (odd how Trump came out on the short end in WI though)

So some degree of machine error is reasonable. But yeah, what seems to have been going on in Detroit, Democratic stronghold that it is, points to something rotten going on there. Doesn't mean the rest of the state stinks too. ;)

Quote
The thing I don't get, living in MI, is how a recount got stopped in part because of this discrepancy.  Wouldn't this just confirm that a hand count is indeed needed?

Not that I expect the outcome to change, but exposing inconsistencies is a pretty huge deal and we should be thanking JS for uncovering it.

My rudimentary understanding is that Stein has been determined to lack "standing" for triggering the recount, and as neither Trump nor Hillary have officially asked for one, there probably won't be a full scale recount. Although is I was certifying election results, it certainly would cause me to start (randomly) auditing other polling locations as well to see how wide spread it is, or is not. However, all indications are that only Detroit threw up all kinds of red flags.

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #37 on: December 13, 2016, 06:41:51 PM »
    Voting machines in more than one-third of all Detroit precincts registered more votes than they should have during last month’s presidential election, according to Wayne County records prepared at the request of The Detroit News.

Do you have any more info?  I looked at this report a while back, the vast majority of the polling places involved had a discrepancy of only one vote. It seemed like a case of over reporting/ misleading reporting to me.

TheDeamon

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #38 on: December 13, 2016, 07:18:17 PM »
    Voting machines in more than one-third of all Detroit precincts registered more votes than they should have during last month’s presidential election, according to Wayne County records prepared at the request of The Detroit News.

Do you have any more info?  I looked at this report a while back, the vast majority of the polling places involved had a discrepancy of only one vote. It seemed like a case of over reporting/ misleading reporting to me.

Going with primary sourcing on this, and Google:

http://www.detroitnews.com/story/news/politics/2016/12/12/records-many-votes-detroits-precincts/95363314/

Quote
Detroit’s mismatched votes

Here is a breakdown of the irregularities in Detroit’s 662 precincts:

■236 precincts in balance — equal numbers of voters counted by workers and machines

■248 precincts with too many votes and no explanation (77 were 1 over; 62 were 2 over, 37 were 3 over, 20 were 4 over, 52 were 5 or more over).

■144 precincts with too few votes and no explanation (81 were 1 under, 29 were 2 under; 19 were 3 under; 7 were 4 under; 8 were 5 or more under)

■34 precincts out of balance but with an explanation

Although as you read the article, it's pretty clear that the experienced officials are spouting lines that indicate the machines should have been biased towards over-reporting the number of ballots, not under reporting. Due to the multiple page ballot, and a few other factors at play in Detroit. So they have an anomaly here that is very odd.

Also:

Quote
Republican state senators last week called for an investigation in Wayne County, including one precinct where a Detroit ballot box contained only 50 of the 306 ballots listed in a poll book, according to an observer for Trump.

City officials have told state officials that ballots in that precinct were never taken out of a locked bin below the voting machine tabulator on Election Day, said Secretary of State spokesman Fred Woodhams.

“That’s what we’ve been told, and we’ll be wanting to verify it,” Woodhams said. “At any rate, this should not have happened.”

The state is not calling the audit an investigation, “but based on what we find, it could lead to more,” he said.

Wayward Son

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #39 on: December 14, 2016, 10:51:37 AM »
One thing I don't understand is why only Hillary and Trump have standing to ask for a recount.

As a voter, I have an interest to know if my vote was counted, or even overcounted.  Why can't I pay for an audit to make sure my vote counted (if I have the cash).  Don't I have a standing in that?  ???

Seriati

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #40 on: December 14, 2016, 10:56:27 AM »
It's a well established part of our law that you have to have standing to sue.  Effectively you have to be harmed.  Otherwise, anyone could force anyone else to incur legal expenses and waste their time to test theories.

In this case though, you might be able to argue that you did have standing, given your status as a voter.  But that has to weighed against the risk of not finalizing the vote count before the federally required deadline, in which case you'd risk disenfranchising the voters of your entire state to run your test.  Given we can't even agree to "disenfranchise" the dead consistently, how is it going to fly to disenfranchise entire states?

D.W.

  • Members
    • View Profile
Re: How should we assess differences between polls and voting results?
« Reply #41 on: December 14, 2016, 11:05:15 AM »
And the expense.  It's a lot of work to do a recount.
I think however the ability to put to bed rumors and allegations of voter fraud would be worth the expense in this case.  What money value do we place on confidence in our system?