Wednesday, June 8, 2016

Why a tyrant would dance a jig at American exit “polling” or how can citizens ever know that American elections are fair? – the three stories of Joe Lenski

     Consider the absence of exit polls, the wide mischanneling of independent voters (nap) into not casting valid ballots, misregistrations (see below), and the California primary yesterday in the light of my account of New York.  This is written for any person who cares about the fairness of American elections...

                      1. Elections must be fair and be seen by citizens to be fair

      Initial exit polls are taken of those who have just left the voting booth.  These polls are comparatively accurate, within a statistical margin of error, because the poll-takers can ask: whom did you vote for, and by and large, people will say. Not all people respond even to exit polling (some 40-50% do).  So the poll has to capture all relevant types of voters and do some weighting to account for groups that answer less frequently.  Nonetheless, initial exit polls gives a range, within a margin of error of, say, two and a half points to either side, of the election results.  And they are often far more accurate.  In the recent American Republican primary elections,  for example, initial exit polling has gotten Donald Trump’s surprising victories and others’ losses, often within a single point, in 20 of 21 cases…

        In Europe, exit polling provides a model for fair elections.  In Germany, all voting is by paper ballot.  The votes sometimes take 12 hours to count.   But if there is controversy, the ballots can be recounted.  Through the use of an initial exit poll, however, who won or lost can often be announced quickly, if the margin of victory is larger than 5 points.  And if the election is too close to call, that, too, can be announced. Citizens then await reports on the actual counting of the vote.  Representatives of the different candidates can review that counting.  About the basic fairness of the vote – a real count of physical votes – citizens, in ordinary circumstances, have no reason to doubt.
      Now initial exit polling – talking to a random sample of voters, say 1 in 10, about whom they voted for – was first developed in the United States.  The crucial figure was Warren Mitofsky, who used them in American elections starting in 1967 and in an amazing 3,000 elections here and abroad[1] (as of 2014, Edison has done such polling in 11,000 elections in 38 countries).[2] Mitofsky worked with Edison Media and died in 2006.  For many years, he cooperated with Joseph Lenski, today Executive Vice-President at Edison.  And exit polling now has 49 years of experience behind it.

    All that experience suggests that Edison gets it right, in most cases, for most particular groups of voters.  Through assessing relevant characteristics, it can correct for the likelihood that young and college educated voters will fill out an exit poll, but that older, less high income voters for, say, Donald Trump, may be reluctant to do so.  And it can quickly adjust and reweight as the last voters leave the polling stations.  Thus, as Lenski insists to Phillip Bump of the Washington Post as well as Doug Johnson Hatlem, Edison had no difficulty in getting Trump’s victories right, often to a single percentage point, in 20 of 21 cases…
       Further, in the first decade of this century, the American State Department funded initial exit polling to test the basic fairness of elections in at least 14 countries abroad.  Based on large discrepancies between exit polls and supposed recorded votes, they deemed 5 – better than one-third… - to be fraudulent. Thus, as Eric Bjornlund and Glenn Cowan’s Vote Count Verification: a User’s Guide for Funders, Implementers, and Stakeholders, 2011, prepared by Democracy International for the US State Department/AID, underlines (p. 52):
                                                          “Exit Polls
In recent years, domestic and international organizations have increasingly turned to exit polls to verify the officially reported results in the transitional elections of emerging democracies. Outside observers have credited exit polls with playing a key role, for example, in exposing fraud in Serbia and Mexico in 2000, Georgia in 2003, and the Dominican Republic and Ukraine in 2004. U.S.-funded organizations have sponsored exit polls as part of democracy assistance programs in Macedonia (2002), Afghanistan (2004), Ukraine (2004), Azerbaijan (2005), the West Bank and Gaza Strip (2005), Lebanon (2005), Kazakhstan (2005), Kenya (2005, 2007), and Bangladesh (2009), among other places.”

      Bjornlund and Cowan continue:
“Exit polls have long been employed in developed countries to quickly predict the outcome of elections. If conducted in countries with a history of democratic elections and in which citizens have reasonable confidence in their own safety and security, then well-designed exit polls can serve as an effective method for projecting election results.”

       Yet what is good enough for use in other countries, including Germany and England, according to the State Department, the United Nations and the Carter Center, is hidden by the large media corporation, behind a veil of silence inside the United States.  For in 2004, exit polls revealed a large discrepancy – 6.7% in Ohio, a 5% overall shift from Kerry, the apparent winner on election day - to Bush in doctored or so-called exit polls the next day.  As a result of the subsequent controversy, in 2016, initial exit polls are rarely announced on CNN or MSNBC.  Even when they are, their role as a check for fairness both abroad and at home is not mentioned by media pundits. 

       Worse yet, within the United States, the National Election Pool corporations – ABC, CBS, FOX, CNN, NBC and the Associated Press - misname dummied up estimates to fit the final results “exit polls.” But these later “polls” involve no additional talking to voters.  Instead, an Edison or election official arbitrarily adapts once reliable exit polls to machine-recorded precinct data until they fit an uncheckable “result.”  In the name of “exit polling,” the NEP perversely buries fairness for rubber-stamping.

     Any dictator or cheat abroad would be grateful to the National Election Pool for providing so convenient a license.  He or she might use Edison’s malpractice in the United States to challenge State Department judgments based on initial exit polls…

                           2.  An Intense Motivation, if Unchecked, for Fraud

        In the United States as elsewhere, competitive election of political leaders has enormous stakes in terms of power and wealth.  Without great constitutional effort and a free, i.e. not simply corporation-dominated, controversy averse press, such elections are not remotely neutral or impartial.  Historically, in America’s elections, those in power try to tilt the scales through property based suffrage, male suffrage, slavery, segregation, Tammany Hall, the Daley machine…Ordinarily – that is, without carefully thought out and public checks to guarantee fairness, people who are powerful or rich have an overpowering motivation to put their finger on the voting balance and keep themselves or their organization in power. 

      In addition, few elections occur without error, and many “errors” are outrageous. Votes are lost; sudden bursts of voting which “determine the outcome” in close cases are reported very late as in Kentucky on May 17.  There is thus good reason for the government to fund an initial exit poll as a publicized test of fairness, say, by an independent panel of academic experts  just on the question of whom did you vote for. Or the government could also fund some minimal, competitive initial exit polling as a check against too many false results.  Once again, if the voters are polled randomly immediately after they vote and if paper ballots exist which can be recounted, as in Germany, it is hard, except in very, very close elections, to falsify results.

        Moreover, in contrast to democratic Europe, half or more of the eligible population does not cast ballots in American elections.  In non-Presidential years such as 2014, sometimes only as few as one-third vote.  One central reason for this:  Europeans vote on the weekend; Americans vote on Tuesdays.  To guarantee the fairness of democracy – one person, one vote -  it would also be good to move the elections to a Saturday or as President Obama has recently suggested, to make the first Tuesday in November a national holiday. That America does not do this again undercuts any appearance of fairness and democracy…

      A Harvard-University of Sydney Election Integrity Project Study for 2015 recently rated American democratic practices below those of any country in Europe as well as those of Argentina, Brazil, Tunisia and Rwanda.[3]  Only silence about other democracies permits the droll assumption of pundits in the National Election Pool that American elections are, in any obvious sense, fair.

       Worse still, the number one public responsibility in America, voting, is no longer a public matter, overseen by the federal government.  Instead, recording the vote has been sold by states to private corporations which make electronic voting machines.  These machines have “proprietary” computer programs which cannot be checked either by the government or by a committee of independent experts.  Like the Volkwagen program which fooled US environmental testers until this year, the computer program itself can itself be falsified.[4]  Volkswagens tested as if they met the standards, but, ordinarily, pollute 40 times as much as is legally allowed.  Perhaps innocently, Walden O’Dell, chief executive of Diebold election machine corporation, pledged victory in Ohio in 2004 to George W. Bush...
       In addition, such programs can be hacked or altered by outsiders.  Except in rural counties which use paper ballots, there is no reliable paper trail to be reviewed.  For instance, “touch screen” machines, used in the 2004 Presidential election, left no paper record.  After protest about that election, some states like California and Colorado, replaced such machines with “optical scan” machines, which do leave a paper trail.  Yet most states today use “touch screen machines” over 10 years old.   Legislatures are unwilling to fund modernization except when there is glaring discrepancy and protest.

       Even “optical scan” machines, however, have “proprietary” programs a la Volkswagon and can be hacked. And the National Election Poll and Edison have “proprietary” control of exit poll data.  These corporations refused to release the raw data from 2004, let alone explain it.[5]  About the fundamentals of our democracy, the word “proprietary” is a scandal.

    In a democracy, elections must both be fair and must appear to be fair.  The contrast to Germany is, once again, glaring.  Without public review of at least initial exit polling by independent experts as well as much sharper checks on and some publically supervised review of the program of voting machines, there is no guarantee that any election in the United States is fair.

     Still, outside of very close elections, initial exit polling can show that the results are, within a range of 4 or 5 points and often much closer.  But initial exit polls are the only test for fair elections in the United States. Recounts in America, even where possible though some combination of paper ballots and optical scan machine tapes and when not postponed or obstructed by partisan Secretaries of State… - are usually not, by themselves, such a test.

             3.  The startling Clinton-Sanders primary exit poll/final results discrepancies

Charnin Chart
       Given how often exit polling gets it right – for instance in 153 out of 154 cases in the 2004 elections, these discrepancies, all in the direction of Hillary Clinton, suggest a pattern.  Now, some are in states with Republican governors and secretaries of State such as Alabama and Texas. (h/t Joe Conason)  In those states, even if there are city officials who are pro-Clinton, it is likely that some extra-official manipulation, say hacking of registration lists or manipulation of old machines by outsiders, might be part of a reasonable explanation.  But note: many of these huge discrepancies in the South, over and above Clinton’s serious victories there, contribute to the false corporate media statement that Clinton has received 2.5 million more actual votes than Sanders.  That figure omits all the caucus states won by Sanders (say, with 66% of better as in Colorado and Washington).  In elected delegates, the race is currently 55% Clinton – 45% Sanders.  But this statement also projects large numbers of phantom votes for Clinton – beyond initial exit polls – which are unlikely.  Given the absence of reliable paper trails, however, the actual difference beyond initial exit polling would be difficult to check.

    Some of these discrepancies  checked Sanders’ momentum.  For instance, the initial exit poll showed Sanders the winner in Massachusetts 52.3-45.7.  But with an 8 point shift, Clinton was the recorded winner, seeming validation of her support outside the South.  In Illinois, one of Clinton’s home states, Sanders wins 50.7-48.4 Clinton had a 4.1% shift which gave her the victory.  This is far less than in New York, but the consequence for Clinton’s campaign was enormous.
     Having just lost in Michigan, Ohio, where independent voters were excluded, was a good election for Clinton.  She is shown as winning there on the exit poll 51.4-47.6   but the recorded results were a far wider deviation 56.5-42.7, a large 5.2% beyond the margin of error.  And a loss in New York would have been devastating to her campaign.  Even a small Clinton victory, say 52- 47.6% on the initial exit poll would have boosted the Sanders campaign.   But instead, the recorded result was 57.95 to 42.05, 15.9 points and a very large, 6.2% beyond the exit poll’s margin of error.  It is this discordant result on which Joseph Lenski gave 3 interviews…

                4.  Edison’s monopoly of exit polling in the United States and its hidden doctoring of exit “polls”

        Now exit polling is expensive.  Abroad, the State Department funds it.  Inside the United States, a National Election Pool of 6 large corporations – ABC, CBS, CNN, NBC, FOX and the Associated Press – pays for it.  But these corporations are all run for profit, not to serve a public good. When there is controversy, they claim the exit polls are “proprietary” and refuse to release the data.[6]  In addition, the Pool hires but a single company, Edison Media, to do the exit polls.  And the Edison monopoly has but a single spokesperson, its Vice President, Joe Lenski, who rarely talks to the press.
       Like Warren Mitofsky, Lenski is a genuine expert on exit polling.  And also like Mitofsky, Lenski is an executive of a corporation who has overriding, profit-oriented motivations.  Thus, legally and certainly journalistically speaking, Lenski has a massive conflict of interest.  As a supposed guarantor of the fairness of elections, the corporate media practice of going to Lenski alone is hopeless.

          Moreover, Lenski himself does not write about exit polling.  At most, he gives cautious interviews, only when there is major controversy.   In such interviews, he slowly clarifies technical features of what Edison does in polling, but refuses to offer any specific analysis of the questionable results.  Yet commenting on the exit poll/recorded vote discrepancy in New York, three reporters take this corporate executive to be “the authority” on how exit polling is done.  Thus, for the New York primaries, Lenski is almost a Wizard of Oz. 
      And yet Lenski is someone who needs to pretend for the sake of Edison’s and NEP profits that American elections are – without question - fair. In New York, the initial exit poll showed Clinton 52% Sanders 47.8.  The recoded vote was, once again, Clinton 57.95 to Sanders 42.05, a 15.9% difference, 6.2% outside the margin of error.  Nothing Lenski says acknowledges the magnitude of this disparity.

        Sadly, there is neither the appearance of fairness in a monopoly system, characterized by “proprietary” secrecy, nor the reality…

       In addition, the corporate press has deep-sixed one of the most startling news stories of this election.  When Bernie Sanders launched his campaign, he polled at 4.5% to Hillary Clinton’s 60%.  The nomination seemed Clinton’s in a cakewalk.  Yet Sanders has raised more money in small donations from ordinary people than any major candidate in the last 50 years has received mainly from the wealthy and corporations (a single hedge fund manager put $58 million on to Ted Cruz, his initial stash…). Sanders is met everywhere with large crowds; Clinton is not.  Sanders has brought many young people, hurting through student debt and lack of jobs, into politics.   He has won 45% of elected Democratic delegates.  He captures most independents and wins easily in polling against Trump.  Clinton does not.
       Yet, in the corporate media, Clinton is uniformly styled the inevitable nominee, Sanders’ victories tagged with the clause, “but he cannot win enough delegates.”  Sanders’ positions go unreported or at most debunked.  There is, naturally great discontent and distaste from below at the corporate media which has almost an official line, as one used to say of authoritarian regimes, which no reporter dares challenge.

     In addition, media commentators all pretend that American voting is, beyond question, fair.  And any outsider who raises a question, even about blatant discrepancies, as demonstrated in the chart in section two, is irrationally labeled a “conspiracy theorist.”  For by a sleight of hand, with that dismissal, neither pundits nor Joe Lenski need provide a specific count of  the discrepancies.  And no corporate reporter may suggest in print or on the air that there is a problem of large-scale errors in American elections, let alone, that a “count” may not be fair. 

      As Phillip Marlowe says in The Maltese Falcon, look at the number of the discrepancies in the table. Consider how rarely initial exit polls get it wrong.
      And unfortunately, this dismissal of “conspiracy theorists” must then include the US State Department’s, the United Nations’ and the Carter Center’s protest of elections abroad which deviate too much beyond the margin of error of exit polls.  This “tut-tutting” must also include the German government and even Joe Lenski who rightly boasts about how often Edison’s initial exit polls get it right, for instance, for Trump.  And finally, they must dismiss the very well argued 100 page John Conyers’ report about the deviations in 2004.  Though written for Democrats, the authors get the accuracy of exit polling and detail many kinds of fraud in Ohio.

        Better, for pundits, to dismiss any criticism out of hand.  For about 2004, neither Mitofsky, though he was requested, nor the National Election Pool, though CNN promised, has ever issued the whisper of an adequate response.[7]

    5.  How Joe Lenski contradicts himself in his first two interviews about exit polling in New York

       On April 19, starting at 7 o’clock, Wolf Blitzer on CNN kept announcing the initial exit poll as 52 for Clinton-47.6 for Sanders, and saying the recorded margin would close during the early evening. That was a reasonable thought, given the accuracy of such polls.  But the “results” did not close.  Clinton was recorded on the machines with a victory of 57.95 to 42.05.  

        This exit poll deviation coincided with a pre-voting, massive de-registering of Democrats.  At polling stations on Long Island and the Bronx, 16-25% of Democratic voters found their registrations changed to independent or Republican or not dated.   Nothing of such magnitude had occurred in previous elections. They had to file “affidavit ballots.” In Brooklyn, 123,000 were disenfranchised and two election officials fired the next week.  These arbitrary pre-voting removals are all activities overseen by Democratic Party officials – often partisans of former New York Senator Clinton.  For since younger voters are for Sanders, this disenfranchisement seems to have affected mainly Sanders voters. 

        Some of these changes in New York could perhaps have been done by outsiders  hacking registration lists.[8]  And this practice echoed one in Arizona according to  a careful Anonymous report.[9]  In New York, some wrongly disqualified voters even sued to be allowed to vote and were allowed to vote by a Judge.  Yet election officials did not routinely tell voters they disqualified that this was a possibility. Most were given “affidavit ballots” which remain, to this moment, uncounted.

        Let us now consider the context of this election.  In this most hotly contested primary in the Big Apple, the most international, shining city in the United States, with Bernie Sanders drawing amazing crowds of 27,000 in Washington Square Park with a line outside of another 20,000, 20,000 in the South Bronx, and 20,000 in Brooklyn, with an intense need for former New York Senator Hillary Clinton to stop Sanders’ momentum, with Mayor Bill Diblasio campaigning all out for her and the Democratic apparatus trying, at all costs, to protect her candidacy, the turnout out for the New York primary was a dismal 19.7%, the second lowest of any primary this year except Louisiana.[10]

            If you believe that this was possible without massive error or abuse, there may be a bridge in Brooklyn you would like to buy. The Democratic Party seems to have had, at minimum plainly with Diane Haslett-Rudiano, the chief Board of Elections clerk in Brooklyn, its finger on the scale.[11]

       Given the controversy, Lenski gave two interviews to reporters uncritical of the Primary and, after “weeks” of requests, wrote brief answers to 10 questions from a more skeptical one.  In all three interviews, Lenski damps down any mention of fairness in initial exit polls.   But there is a glaring, self-serving contradiction in  Lenski’s first two accounts:

     To Joshua Holland of Raw Story, Lenski says that Edison exit polls in America are not designed to see who, very likely, within a certain range, won the election.  Only simple questionnaires for people in “transitional democracies” do that.  The Media Corporations of the National Election Pool want only the “complexities” of demographics, Lenski insists, but no results. And Edison “complies.” This stance accords with the NEP simply not reporting initial exit polls since the glaring deviation in the Bush election in 2004.

       To Philip Bump at the Washington Post, however, Lenski boasts that initial exit polling by Edison got it right about Trump in many primaries and, surprisingly, about Sanders in Michigan despite the pre-election polls.  That claim is more plausible.  Everyone, even Big Media Corporations, wants to know who won or lost. But Lenski contradicts his previous interview point-blank.  It is dangerous for Lenski’s business model for Edison’s initial exit polling to get it right too much, because then the issue of rare discrepancies becomes glaring.

       Thus, in  both interviews, Lenski shies away from the very high rate of success on initial exit polls in establishing a margin of error and guaranteeing fairness of results.  Instead, he sticks to technicalities.  He says as little as he can get away with.  He does not mention the initial exit poll/recorded vote discrepancy in New York. Such a strategy was also pursued by Warren Mitofsky about the 2004 election.   Even Lenski’s answer seems to toy with the interviewer and reader; it is, as we will see, not a thoughtful response about the huge discrepancy.

      Now Edison possesses the relevant data, and manipulates it, with dazzling  speed, on election day.  Lenski could easily get and give real answers which the public could then think about and debate. And yet the Edison Corporation/National Election Pool refuse – stonewalling on “proprietary grounds – even to release the figures.

       Listen to Lenski’s two interviews in turn. 

       Joshua Holland interviews Lenski in a flamboyant dismissal of questioning as nonsense: “On Tim Robbins, Election Fraud and How Nonsense Spreads Around the Internet,” Raw Story, April 27, 2016.  To Holland, Lenski acknowledges the core democratic/moral use of exit polling abroad:

      “As for using his [Lenski’s] results to suss out fraud, he says that American exit polls are ‘just not designed for that type of precision. They’re surveys, and like any other survey, they have a margin of error [so do exit polls used abroad - AG]. The precision that a lot of these people are talking about just doesn’t exist with our polls.”

          From his comment about “sussing out fraud,” Holland is  aware of this fundamental issue, but mistakenly, turns away from it.     

         “’In emerging democracies,’ says Lensky [sic], ‘the exit polls are designed specifically to catch any manipulations of the vote count, and also to bring some transparency so voters can trust the vote count.”

      This is Lenski’s sole reference to the central purpose of initial exit polling: to check the fairness and appearance of fairness of elections. Holland drops the ball.  For any of us might say: Stop! Do we Americans not want transparency, too?

        Lenski then adds a superficial, technical point:

       “’They have a lot more locations in the sample, they do a lot more interviews and they use  a much, much smaller questionnaire. In some  cases, they just ask, ‘who did you vote for?’” A brief questionnaire, he explains, increases the response rate. `The more interviews you do, the more locations you cover and the shorter the questionnaire, the higher response rate you’ll get, and that all leads to a much smaller margin of error.’”

       But then, Lenski admits in a roundabout way that the National Election Pool suppresses initial exit polls in America:

      `The news organizations that sponsor our exit polls are just looking for a sense of who voted, and what motivated them to vote. They use longer questionnaires – typically with about 20 questions - and it takes a little longer for voters to fill them out.’ Lensky [sic] says that while the methodology hasn’t changed much since exit polling was first introduced in 1967, the fact that news organizations post preliminary data in real-time leads to ‘a lot of commentary on social media.’”

         Now, there is a simple remedy to ensure democracy here.  If the issue were merely that Edison’s polls ask too many questions, and that polls, funded by US AID abroad, are simple, then Edison could  simplify the polls to ensure transparency in the US, too.  That might be bad for Edison’s business model – they might have to charge less or hire more poll takers – but it would be a serious advance for ensuring the trust of voters in the fairness of our democracy…

      Alternately, the federal government could fund a committee of experts or even another corporation to do an exit poll, at different voting stations from Edison’s, on this one question. And this, and this alone – by law – would be announced by pundits in the corporate media in every case as an exit poll.  Nothing else should be misnamed an exit poll.  That would guarantee fairness and the appearance of fairness.  And Edison could still make money from its demographic, trustworthiness and leadership capability questions about the candidates.

      On April 22, three days after the New York primary, Phillip Bump interviewed Lenski for his article “How exit polls work, explained” for The Washington Post.[12]    

     In contrast to Holland, Bump shows no understanding of what an initial exit poll is nor its central role as an indicator of fairness.  Instead, he asserts mistakenly:

     “Exit polls provide remarkably quick data on elections that comes with a smaller and smaller margin of error as an election night passes, telling us who came out to vote for which candidate and why. “

       In all interviews, Lenski alerts the interviewer to the possibility that young voters are oversampled in initial exit polls.  But listen to Bump's and Lenski's  conversation about this ostensible error.  For Lenski immediately reports that Edison adjusts for it.  And he then abruptly shifts to an entirely different issue, diluting initial exit polls with “recorded results”:

      “BUMP: So the New York issue, I can't help but notice that you said that younger people are more likely to fill out the surveys and then also that Sanders was over-represented in the initial estimate. Do you think there's a link there?

LENSKI: Oh, yes, there's definitely a link there.  We're adjusting for that throughout the day. As I said, we know the response rate in our 35 precincts. We know that younger voters are more likely to choose to fill out the questionnaire than older voters, and that's typically the case so we're already making those adjustments.

      Obviously in this case, that was even more than normal. As soon as we started getting sample precinct returns, we made that adjustment even more so that we'd match the actual results.”

      Note that only for the United States, Lenski, echoed by the Washington Post and without a single question from Bump, inverts the purpose of exit "polling":

      “As soon as we started getting sample precinct returns, we made that adjustment even more so that we'd match the actual results.”

       But then Lenski rightly crows:

       “In Michigan, we actually had exit polling all day showing Bernie Sanders up by two points though every pre-election poll had Hillary Clinton up by 10 points or more -- so we're sitting out there on a limb the other way. In that case, the exit poll was right and the pre-election polls were wrong. It happens both ways.”

       Sorry, Joe, the State Department and Edison fund only initial exit polls, not pre-election polls abroad…[13]

      In any case, Lenski contradicts, point blank, his claim to Joshua Holland in Raw Story. 

      “’While everyone is talking about the Democratic side, we went out at 9 o'clock saying that Donald Trump was going to get 58 percent of the vote. He got just about 60 percent of the vote. Everything we did on the Republican side hit the mark. I understand when the data moves as much as the Democratic data moved between 9 o'clock and 9:45, that causes a lot of consternation out there. But there are plenty of other states where we've been right on.”

      Once again, contra his interview with Holland, Lenski says to Bump that the purpose of exit polls  “is to project a winner.”   He tries to soften this with a comment about  seeking a demography of voters and a falsehood that "historically" only the latter - and not fairness nor even naming a winner - is what such polling "mainly" seeks.

       “’There are two important uses of the exit poll. One is to project a winner. But the main use of the exit poll that night and historically is to have the most accurate representation of the demographics of voters.’”

        But while that is true for Edison’s business model for the NEP inside the United States, it is false about what the State Department does abroad, what Europeans do, and what Edison did in Iraq in 2014.[14]  What Lenski hides is the profound issue of fairness for American elections of which initial exit polling is the sole guarantee.

      As if it were uncontroversial, Lenski then shifts to Edison’s inversion and misnaming of “exit polling” inside the United States:

       ’In a state like New York, we had 35 exit poll precincts and 50 sample precincts in the state. The sample precincts include the exit poll precinct. We have a reporter in each polling location in the sample, and as soon as the results are posted at that polling place, they will call those results in.’

       ‘Shortly after poll closing, we can quickly compare what the exit poll of that precinct said the votes were and what the actual votes were. So that's when you'll see a fairly quick adjustment to the exit poll estimates after poll closing.’

      That would be reasonable – still an exit poll – done around 7 o’clock, the closing time.  But following  machine counts to “adjust the poll” would not. Yet the latter is all Lenski and Bump talk about:

      “Like in New York, we were showing a four-point margin in the exit poll at 9 o'clock, but by 9:45 we were showing a 12-point margin. That's because we can quickly compare precinct-by-precinct what the exit poll results were and what the full results for that precinct were. So we're seeing precinct-by-precinct that the actual results were that Hillary Clinton was doing four points better than she did in the exit poll in that precinct, we will adjust the results [of the exit poll] accordingly.”

       Lenski just slips in and legitimizes this ambiguity about  exit polling in, without even a name, say, rationalizing machine "results."  This is not "Exit Polling Explained," as in Bump’s unctious title, but exit polling misexplained....

             6.  The ostensible skews suggested by interviewers are quickly corrected for by weighting     

       In their questioning, Bump and Holland are both at pains to suggest that young voters, say 19-30 and 70%-30% for Sanders, are unusually eager to fill out questionnaires and must be overrepresented in the initial exit poll.  They say this again without acknowledging the magnitude of the discrepancy.  Though serious enough, there are several additional problems with this thought.  First, as the vote is concluded, exit pollers can weight the final results.  That is, if too few older Clinton voters did not respond – this is a long recognized problem in the 2016 primaries – the poll can be immediately adjusted.[15]  

        Put differently, as of 2014 ,Mitofsy and Lenski have overseen 11,000 exit polls in 38 countries.  What they do is an unusual kind of iterated random sampling which gives, normally, accurate results within a margin of error.  For instance, in 2004, the Edison Mitofsky Company produced accurate initial exit polls in 120 elections in 50 states as well as in 23 primaries.[16]  That is a pretty good record. For the Presidential election alone, the initial exit polling showed Kerry won nationally by 3% and in Ohio by 4%, and had to be recast 5 and 6.7% the next day to accord with a Bush “victory.”  That major “error” occurred in only 1 of 144 cases, a .0069% chance on an “error” in initial exit polls inside the U.S. in that year.  Yet that sole, initial exit poll “misestimate” in the Presidential election is the corrupt origin of what Bump, following Lenski, misdescribes as “exit polls.”

      To withdraw Edison’s initial exit poll in 2004 and without providing data, Mitofsky hypothesized a large “within precinct error” (WPE) for Kerry voters.  Claiming “proprietary” authority, Edison and the NEP released only limited data in January 2005.  At that time, analysis by Stephen Freeman and others showed that the skew in that data was the other way.  In the overwhelmingly Bush districts (80% or better for Bush) which Edison released, Bush voters tended to over-report…[17]
       Further, Mitofsky’s and Lenski’s repeated “explanations” of error in the exit polling are implausible.  Without remarking the size of the discrepancy or showing from the data they have collected how precisely the initial exit poll data went awry, they simply assert, as “authorities,” Wizard of Oz-style, that the exit poll, in this rare case, must have been mistaken. No other explanation is allowed
    Once again, the putative error they invoke, even were it substantial, could not, by itself, explain the discrepancy.  But this malpractice is very good for Edison’s business model.   The corporate media rely on their “authority.” No questions are asked or answered. 
      Talk about the Emperor’s gauzy clothes…
      Second, a segment of lower income white voters who are Republicans vote for Trump.  Perhaps they, too, might be less anxious to speak to poll-takers than college educated Republicans who supported “more respectable” others.  But exit pollers weight their votes, given observation about the 50% who do not answer.  And Edison has gotten Trump’s victories right, often within a point in 20 of 21 cases.  What’s the special problem, again, for doing the same with Clinton voters? Lenski suggests none…

      For, third, Lenski himself points out that Edison weights samples and makes adjustments when the polls close (when better than 90% of the vote is counted).  And they seem to have done this weighting of young and old accurately on the Democratic side in other instances throughout the primaries and even in the New York election.  The putative error seems to be in but a part of the sample.  For  Sanders won in all the rural counties in New York as well as some smaller cities. Initial exit polls seems to have had no problem, there.   In some other primaries that Sanders has won, for instance, Michigan, Edison exit polling did not overestimate the young. So the claim that Sanders’ standing in the New York City initial exit poll was over-recorded due to the exuberance of youth and the reticence of age is very likely false. 

        As the critical interviewer, Doug Johnson Hatlem, suggests, this makes the discrepancies between the initial exit polls and Clinton’s margins – she exceeds the margin of error in 9 primaries, he suggests – even more glaring.  After repeated attempts “over weeks” to reach Joe Lenski, the latter eventually wrote out answers to 10 questions.[18]  (Note Lenski spoke conversationally with interviewers who wanted to use his expertise against critics)

       “[Hatlem:] It seems like a tougher field on the GOP side with so many candidates, including not one but two anti-establishment candidacies. In Georgia, for instance, where the first wave missed on the Democratic side by 12.2%, it nailed the GOP race with deadly accuracy: 40% Trump (versus a 38.8% finish), 24% Cruz (versus a 24.4% finish) and 23% Rubio (versus a 23.6% finish). It looks like you’ve missed the margin of error just once for Republicans. In Texas you had an ~10.6% error on the gap between Cruz and Trump. This makes sense. We often hear the margin of error is +/-x.x 19 times out of twenty. In this case, Edison has gotten it right on the GOP side within the margin of error 20 times out of 21 (for the figures I can find). On the Dem side, you’ve gotten it right within the margin of error just 13 times out of 22. Is this information correct and if so, why has Edison polling been so much more accurate on the Republican side this cycle? [this question tries to show Lenski that Hatlem has done his homework and knows something about statistics]

[Lenski:] As I mentioned above the calculation of total error for an exit poll survey differs from the 
standard sampling margin of error calculation that I assume that you are using so I wouldn’t agree 
with your statement about how many of the exit poll surveys were within the margin of error.” 

       Note again the oddness of this response. Lenski does not tell us how many exit  poll/recorded 
vote discrepancies exceeded an ordinary margin of error but not the one Edison’s exit poll surveys 
use.  If for instance, that explained half the discrepancies, that might be reassuring, but still leave the 
question about the others in tact.  So he just doesn’t say.   Earlier he explains to Hatlem that the latter
involves both a margin of error in the selection of random precincts and polling stations to be studied 
combined with a margin of error from only sampling a limited number of voters within that polling 

        Lenski then reverts to his standard non-answer:

         “However, if a differential non-response among younger voters is a cause for exit poll errors it 
would make sense that the errors would be larger on the Democratic side because the differences in 
vote between younger and older voters on the Democratic side in this primary season are much larger 
than on the Republican side.  Bernie Sanders has been typically receiving 70+% of the vote among 
17-29 year olds in the 2016 primaries while Hillary Clinton has been receiving 70+% of the vote 
among voters 65+. [But in earlier interviews, Lenski has shown that this isn’t so big a deal to 
reweight for…]

        On the Republican side the Trump percentage among younger and older voters tends to only 
differ by ten points or less.  It would then make sense that if the exit poll were overstating the number 
of younger voters it would have much more effect on the Democratic side.”

       Hatlem asks for one small detail from Lenski’s vast data. Lenski, the expert, could easily have given it to him.  Instead, Lenski, the Edison executive, delphically sends him to the Roper Center where Hatlem has yet to be admitted.  One wonders whether the complete data, even from 2004, let any serious analysis even of that, is there... 

       Writing for Counterpunch, Hatlem is so far the only serious analyst of these discrepancies.  And Counterpunch is putting effort into researching issues and contacting a variety of experts. Hatlem is skeptical of many explanations which imply fraud. And he looks for explanations, for instance, the mis-registration of a large number of voters who then cast only uncounted affidavit ballots, which might account for part of the discrepancy. One person cannot take on all of this material, but he has made an important effort.  What he says so far is more intelligent and fair minded than Lenski.

     Now Hatlem traces the use of 10 year old voting machines in some of the discrepancies, thinking they might be more easily hacked.  And Lenski tells him to follow up – perhaps because this finding will not apply to New York.  Despite the long delay in getting Lenski’s response, Hatlem is so blown away that the expert answers his questions that he forgets - or perhaps does not yet understand - Edison’s unjustifiable monopoly role in exit polling and worse, Edison’s distortion of what exit “polling” is inside the United States.

        For without these insights, Hatlem’s case about serious error or fraud is much weaker than it should be.  Prima facie, the unchallenged assumption that American elections are fair, in the corporate media, by Edison and by bipartisan agreement, needs a serious defense.  It has so far received none.  And it is doubtful that reasonable questions can be answered without reform making clear, as in Germany, the special role of initial exit polling in upholding fairness as well as initiating some minimal competition among companies/poll-takers to get a rigorous answer to the question for each of us: whom did you vote for?

[1] House Judiciary Committee Democratic Staff/John Conyers Report, Preserving Democracy: What Went Wrong in Ohio, 2005, p. 75. 
[4] Barbara Simons, “What if Volkswagen made voting machines?,” September 25, 2015, Verified Voting,
[5] Conyers Report, p. 75
[6] Idem.
[7] Idem.
[8] See Doug Johnson Hatlem, “Purged, Hacked and Switched: on election fraud allegations in Hillary Clinton vs. Bernie Sanders,” Counterpunch, May 12, 2016.
[13] Lenski fuzzes up another  important point here. Pre-election polls, even sorted by Nate Silver, are notoriously changeable and often prove far off in the election. They are not close to more accurate or useful than initial exit polls.
[14] See Edison’s report on the exit poll in Iraq at
[15] See the Roper Center explanation of weighting: 
[16] Fritz J. Scheuer and Wendy Alvey, Elections and Exit Polling, pp. 66-67.
[17] US Count Votes, National Election Data Archive Project, Analysis of the 2004 Presidential Election Exit Poll Discrepancies, pp. 9-10.

No comments:

Post a Comment