How Polling Firms Condone the Many Deficiencies of our Election System
Last edited Sat Apr 16, 2016, 03:40 PM - Edit history (1)
For those of you who don’t believe that the United States has serious problems with our election system, a 2015 report by the Electoral Integrity Project, which gathered assessments from over 2,000 experts on the subject, found the United States to rank last among Western democracies. There are many reasons for this, but suffice it to say that all the problems that contribute to this ranking have very serious consequences to our democracy and make it considerably less democratic.
Yet most Americans are blissfully unaware of most of these problems, mainly because our national news media refuses to cover them in any serious depth. As an example, the 2004 Presidential Election was characterized by massive fraud, especially in Ohio, which gave the election to George W. Bush. Only one national news person, Keith Olbermann, deigned to discuss this on national TV, and he eventually lost his job because of that.
How polling firms condone our failed election system
I will use Nate Silver as an example here, not because I believe that he is worse than any of the others, but because I have read more extensively about his methods – in his own words – than that of any other pollster or polling firm. There are basically three ways in which pollster methods and/or silence about various issues condone, and therefore contribute to, some of our many election system failures.
Poll adjusting
One is through the “adjustment” of poll data. The polls we see are generally adjusted to give more weight to “likely voters”. In turn, who is a likely voter is determined based on past experience with exit polls from previous elections. But what actually determines who is a likely voter?
It is of course true that some demographic groups are actually more likely to attempt to vote than others. But what else in our country contributes to who is a likely voter? Many nefarious activities, legal and illegal, have been found to contribute to this. One of the “legal” mechanisms is the many voter ID laws that have been passed in many states in recent years, with the obvious purpose of disenfranchising poor and minority voters.
Other mechanisms are the reduction of polling places and the selective purging of voters, both with the purpose of helping one’s preferred candidate. We saw this recently in the Arizona Democratic primary, where in Maricopa County, which contains about half of Arizona’s population, polling places were reduced from the previous election from over 200 to 60, and untold numbers of prospective Democratic voters were erroneously told that they couldn’t vote because they were not registered as Democrats. As the result of the massive reduction in polling places, where voting lines stretched for as long as a half mile, and prospective voters had to wait for several hours to vote, Election Day voters constituted less than 15% of total Maricopa County voters – the remainder being early voters. Both of these processes contributed greatly to Bernie Sanders’ large loss in Arizona. The polling place reduction because Sanders fared far better among Election Day voters (61% for Sanders) than among early voters (34% for Sanders) because many early voters voted at a time before Sanders became well known. The purging of voters also hurt Sanders greatly, as an investigation was able to identify 113 prospective Sanders voters and only 2 Clinton voters who were purged.
My main point here is that the “adjustment” of polls by pollsters, which give greater weight to “likely voters”, are often if not usually based on the fact that whether or not a voter is a “likely voter” is in large part dependent upon the preventing of certain demographics of voters from voting by nefarious and/or illegal means in previous elections.
But I have never seen any pollster acknowledge this.
Additional “polls plus” adjusting
Nate Silver talks about further adjustment that he uses in his predictions, which he refers to as “polls plus”. This constitutes his final predictions. He discusses some of the considerations that go into this “polls plus” adjustment. The discussion is not very specific, but one of the major considerations is who has control of the state. For example, if the election is between a Republican and a Democrat, a Republican Governor and major election officials would go a long way towards adjusting the “polls plus” number in favor of the Republican.
But what is it about control of the election apparatus that would cause an adjustment in favor of the party in charge? Again, we get to the issue of rigging the election in various ways in favor of the preferred candidate. How much does a governor of one of the candidates’ parties contribute to the end result? I have heard that this makes a difference of 3% in the average election. How much of this is due to illegal means? We don’t know exactly, but we certainly know that it often contributes something to it. In discussing his “polls plus” adjustments, Nate Silver doesn’t even mention this as a possibility.
A few words about exit polls
Those of you who spent much time on DU during the 2004 Presidential election and the months and years that followed will remember the great exit poll discrepancy of 2004, in which, according to national exit polls John Kerry won the national vote, whereas George W. Bush won the national vote according to the official vote count. The difference between the exit polls and the official vote count was about 4% nationally. The difference was particularly great in the important swing states, where slight differences in the vote count might make a difference between winning and losing. But there was only one state where it did make a difference, and that was Ohio, where the exit poll discrepancy was over 6%. Ohio would have given the election to John Kerry.
Further investigations found numerous anomalies in Ohio, and eventually a mass of evidence was accumulated that made it highly probable, if not certain, to make a long story short, that the 2004 Presidential election in Ohio was electronically manipulated to give the election to Bush (and there was massive evidence of voter suppression as well). I wrote many DU posts on these issues at the time, but I think that perhaps the best summary I wrote of the evidence for electronic manipulation in Ohio occurred many years later, when Bush was no longer president, in this post.
Exit polls tend to be far more accurate than pre-election polls, for several reasons: 1) They assess whom the voter actually voted for, rather than whom he or she intends to vote for at some later date; 2) Pre-election polls use models that estimate which poll respondents are likely to vote in an election, based on data from previous elections. These models may or may not accurately apply to the current election. To the extent that they don’t apply, the results can be substantially biased; 3) The accuracy of pre-election polls depends on obtaining a representative sample of voters for the poll. That is no simple matter. One very large potential source of error is that voters who use cell phones rather than land lines may be grossly under-represented in pre-election polls. These problems are almost totally eradicated by exit polls. Exit polls are taken of voters as they leave their polling places. It doesn’t depend on telephones or other overly complicated sampling methods.
In the recent Ohio Democratic primary, the exit poll discrepancy was more than 10%, with Clinton doing more than 10% better in the official results than in the exit polls. She did better in the official vote count than in the exit polls in every state for which I could find data, but Ohio was the worst. This 10% difference is huge. Compare that to the 6-7% exit poll discrepancy in the Ohio 2004 Presidential election, which determined the Presidency in favor of George W. Bush. DU rightly made a huge big deal out of this, with thousands, or maybe even millions of posts on the subject. Shouldn’t the even bigger exit poll discrepancy in the Ohio (and other) Democratic primary warrant more discussion than we’ve seen? Also, this was one of the very few states for which Nate Silver over-estimated Sanders’ performance in his pre-election polls. Yet, not a word from any of the pollsters about this very surprising anomaly – as we saw in 2004.
How pollsters contribute to our election problems by totally ignoring these issues
As I said, there are many reasons why the U.S. ranks last among Western democracies in its election integrity. Yet many or most Americans continue to believe that we have the most democratic democracy in the world. As I hope I made clear, many of our election failures are reflected in the various poll adjusting models created by most if not all professional pollsters. Yet, they never mention the fact that their models actually incorporate these failures.
If pollsters would me honest about their models by acknowledging even the possibility that those models, based on exit polls from previous elections, often or usually reflect illegal and/or nefarious means of purging voters, and if they would note when the official results of elections show large discrepancies from exit polls, well outside the “margin of error”, that would greatly help to call attention to the American people about our many election failures. That might generate the political pressure to do something about those failures. Conversely, when pollsters choose not to mention these issues, which is almost always the case, the “accuracy” of their pre-election polls tends to add to the authenticity of the official results. This in effect condones our many election problems, which hurt our democracy tremendously.
= new reply since forum marked as read
