Exit Polls versus Reported Vote Counts. 2016 Presidential Primaries

Please share this article:

by Theodore de Macedo Soares.

In the United States, computerized election vote counts are essentially unverified [1].  Computer counts are non-transparent and non-observable by ordinary citizens.  For these reasons, and in order to prevent hard-to-detect computer vote fraud, the Federal Constitutional Court of Germany (their version of the US Supreme Court) in 2009, effectively banned the use of computers to count Germany’s ballots [2]. In order to be able to verify the results of their elections, Germany reverted to the hand counting of all ballots in front of citizen observers [3]. Other technologically advanced countries such as Canada, France, Ireland, Italy, Denmark, Finland and 53 other countries protect the integrity of their elections with hand-counted paper ballots [4].

In the US, citizens attempting to independently verify the computerized vote counts predominantly rely on exit polls as the means to ascertain the correctness of the unverified computer vote count. USAID in their 2015 booklet “Assessing and Verifying Election Results” stated “[e]xit polls are powerful analytical tools … [a] discrepancy between the votes reported by voters and official results may suggest that results have been manipulated, but it does not prove this to be the case.” [5]

Although the focus of this report is on the discrepancies between the exit polls and the computerized vote counts, it must be mentioned that other researchers have analyzed the 2016 Democratic Party primaries from related perspectives and their results are not reassuring.  Axel Geijsel (Tilburg University, Netherlands) with Rodolfo Cortes Barragan (Stanford University), for example, recently released their working paper on 22 states with votes backed by a paper trail versus 14 states that did not.  They found that States with a paper trail yielded higher support for Sanders (51%) than Clinton (49%).  States without a paper trail yielded higher support for Clinton (64%) than Sanders (35%) [6].

As in other US elections [7], the large discrepancies between exit polls and the unverified computer vote counts mainly in the Democratic Party primaries of many states in the 2016 presidential primaries, although not undisputable proof of computer based vote fraud, certainly raises concerns as to their veracity.  The lack of significant discrepancies in some states, on the other hand, does not mean they are free of concern.  Wisconsin, for example: the first CNN published exit poll comparison with the computerized vote count yielded a small discrepancy of 2% in Sanders’ favor.  A few hours earlier; however, NBC News at 4:24 PM broadcast the percentages of white voters voting for Clinton (32%) and Sanders (68%) in early exit poll results [8].  Using these percentages to estimate the results of this earlier exit poll result [9] yielded a large discrepancy of 13.4% in favor of Clinton when compared with the unverified computer vote count (see Democratic Party table).

The Wisconsin exit poll first published by CNN had very different white vote percentages for Clinton and Sanders at 40% and 58% respectively. Other documented news broadcasts of early exit poll releases of demographic characteristics of voters in other primary states closely matched CNN’s first published exit polls.  It seems unlikely that there would have been a large surge of white, mainly Clinton voters in the last hour or two of the exit poll data collection.  The more likely explanation is that, as Edison Research admits, assuming the vote count is accurate, they alter their actual exit polls to match the incoming vote counts as quickly as they can as soon as the polls close [10].  Here (and perhaps in other states), they may have altered their actual exit poll data to match incoming vote totals at or just prior to the closing of polls in the state and prior to the first exit poll publication by CNN.  Only the release of Edison’s actual raw exit poll data annotated with their detailed adjustments will be able to clarify and reveal the true discrepancies between their unaltered exit polls and the final unverified computer vote counts in this and other states.

Another important part of the picture in these primaries (and in other US elections) are the complaints of voter suppression and allegations of fraud rampant in the 2016 Democratic Party primary contests in many states that generally benefited Clinton’s campaign.  These accounts have been widely and loudly circulated in the Internet (see this aggregator of news reports).  These accounts only serve to make acute the concerns of possible computerized vote fraud.  As demonstrated by Princeton University researchers [11], and others, computerized vote fraud is much easier to accomplish, hidden from sight and leaving no trace, does not cause the ruckus of the more visible voter suppression efforts.

In using exit polls as an indicator of possible computerized vote fraud, it is important to take into account the following possible sources of error, common to all surveys:  Coverage error (not able to sample selection of the population), nonresponse error (not able to poll all persons in the chosen sample), sampling error (when the survey sample is different than the population being measured), and measurement error (from inaccurate responses) [12]. Exit polls are often criticized because the reported margin of error for the poll, only addresses the statistical sampling error as a function of its size: the smaller the sample, the larger the margin of error.

A popular and frequent criticism of exit polls centers on nonresponse error as the cause for the discrepancies between the exit polls and the final vote count—the voters of a particular candidate may be overrepresented because they are more willing to fill out the exit poll’s anonymous questionnaires than the voters for another candidate. This criticism; however, is generally put forth without any evidence and has been refuted as the explanation for the discrepancies in previous elections [5].  Additionally, Edison Research, the polling firm contracted by the main news networks to perform exit polling, works very hard to reduce such errors by taking note of the characteristics of the nonresponders and adjust their exit polls accordingly [10].

Important to note: The approach taken in the following tables takes into account all the possible sources of error inherent in exit polls.  It does this by comparing the exit poll results for the main candidates in the Republican Party primaries with their actual vote totals. In this way the actual total survey error (TSE) can be measured.  With the exception of the primaries in two states, West Virginia and Texas, with very large discrepancies (18% and 10.6% respectively) suggestive of possible computer vote fraud disfavoring Trump, the exit polls for Republican Party primaries have been fairly accurate.  By increasing the standard statistical margin of error (MOE) by 32%, and excluding the outliers in West Virginia and Texas, the percentage of polls exceeding this increased MOE goes down to 5.15% which is very close to the expected 5% in an exit poll with a confidence interval of 95%. The tables below take into account the total survey error by increasing the MOE by 32%.

The two tables that follow allow for a comparison of the discrepancies between the exit polls (EP) and the final vote count (VC) in the Democratic Party and Republican Party primaries of 2016.  The Democratic Party table features the contests between Clinton and Sanders and the Republican Party table features the contests between Trump and the other candidates combined.

In the Democratic Party primaries, Sanders vote totals were lower in the final count than projected in the vast majority of the exit polls.  In 21 of 25 primaries the discrepancies between the EP and VC favored Clinton by an average of 6.9%.  The top ten averaged a large discrepancy of 10.6% (median 10.2%).  The average of the four discrepancies favoring Sanders was 3.2%.

Democratic Party Table. 2016 Primaries

https://tdmsresearch.com/2016/06/20/democratic-party-table-2016-primaries/ (Table and notes opens in new tab)

In sharp contrast, the discrepancies in the Republican Party primaries were as likely to favor Trump as the other candidates (13 and 10 respectively).  With the exception of the primaries in two states, West Virginia and Texas, that saw very large discrepancies (18% and 10.6% respectively) disfavoring Trump, the Republican Party races exhibited what should be expected with impartial vote counts and exit polls; sometimes the discrepancies favored Trump, and sometimes went against him.  The average of the discrepancies favoring Trump was 2.7%.  The average of the discrepancies going against Trump, excluding West Virginia and Texas, was almost identical at 2.8%.  These averages discrepancies, less than half of the average of their TSE’s (7% and 8% respectively), demonstrate that indeed, exit polls can be very accurate.

Republican Party Table. 2016 Primaries

https://tdmsresearch.com/2016/06/20/republican-party-table-2016-primaries/ (Table and notes will open in new Tab)

Illustrative of the difference between the discrepancies found in the 2016 Republican and Democratic Party primaries occurred in Massachusetts. The exit polls projected a Sanders win by 6.6% and yet he lost by 1.4%, a total discrepancy of 8%.  The exit polls for the Republican Party race; however, occurring at the same locations, at the same time, and using the same methodology closely agreed with reported vote totals with the largest discrepancy among the five candidates at less than 1% [13].

With the exception of the primaries in three states (SC, MO, and WI), all of the exit polls for the Republican Party and the Democratic Party primaries occurred on the same day, at the same locations, with the same interviewers, and used the same methodologies. Furthermore, to dispel once again the notion that exit polls may be wrong because those more enthusiastic for a particular candidate would participate in the exit poll in greater numbers, with Sanders and Trump, each party featured very popular anti-establishment candidates.  The Republican Party race, widely acknowledged as more divisive and polarized than the Democratic Party race, would have been more likely to exhibit the mix of enthusiastic and reluctant exit poll respondents usually offered, without a shred of evidence, why exit polls go wrong.  The usual attempts to explain the discrepancies between the exit polls and the unverified computer vote counts, baseless in the past, clearly do not work for these primaries.

Various media pundits will continue to undermine exit poll results as a possible indicator of election fraud, going so far as to characterize those concerned with election integrity as “conspiracy theorists.” The common weakness in their commentaries is an unspoken blind faith on unverified computer vote counts.  Such unquestioning faith apparently prevents them from even acknowledging the possibility that exit polls, scientifically conducted and honed by decades of practice are accurate, and that computerized vote counts are not. In contrast, Germany’s high court squarely acknowledged the possibility of hard-to-detect fraudulent computer vote counts in its decision to abolish computer counts in their elections.

Blind faith in computer vote counting is rapidly eroding. Until the United States joins the many other countries that safeguard and promote the integrity of their elections with publicly observed hand counting of paper ballots, US elections will remain questionable and suspect.

Acknowledgement: Thanks to Frances Lee Ellis for proofreading this article and her contributions that promoted clarity.


  1. Audits of any election, if conducted at all, are hidden from view—only four states specify that observers can verify the marking on the ballots (only one state for primary elections). Primary election results are only audited by 13 states.  Only 5 of these states require that the results of the audit and the data be made public.  None of these states  perform audits according to the best practices of a risk-limiting audit.  Indeed, only one state is even experimenting with conducting proper audits of any election.
  2. Federal Constitutional Court of Germany. Use of voting computers in 2005 Bundestag election unconstitutional. Press Release No. 19/2009 of 03 March 2009. (The Court explicitly stated, as one of the main reasons for their decision, “deliberate electoral fraud committed by manipulating the software of electronic voting machines can be recognized only with difficulty.”) Available at: http://www.bundesverfassungsgericht.de/SharedDocs/Pressemitteilungen/EN/2009/bvg09-019.html;jsessionid=E6F650A3BEB6A549ED0A3BDE4252BAD7.2_cid393 . Last Accessed on June 12, 2016.
  3. Deutsche Welle (DW). No concerns over election fraud in Germany. September 21, 2013. Available at: http://www.dw.com/en/no-concerns-over-election-fraud-in-germany/a-17102003. Last accessed on June 12, 2016.
  4. International Institute for Democracy and Electoral Assistance (International IDEA). Vote Counting methods. Available at: http://www.idea.int/vt/vote_counting_methods.cfm. Last accessed on June 12, 2016.
  5. Assessing and Verifying Election Results. April 2015. Available at: https://www.usaid.gov/sites/default/files/documents/1866/DI%20Assessing%20and%20Verifying%20Election%20Results%20Summary%20Document%20-%20FINAL%20PDF%20(without%20bleeds)%20(5-19-15).pdf.  Last Accessed on June 12, 2016.
  6. Geijsel, A., Barragan, R.C. Have we witnessed a dishonest election? A between state comparison based on the used voting procedures of the 2016 Democratic Party Primary for the Presidency of the United States of America. Tilburg University, Netherlands. Stanford University, Stanford, California. June 8, 2016. Available at: https://drive.google.com/file/d/0B6mLpCEIGEYGSlRsV0IxV1ByXzQ/view.  Last Accessed on June 20, 2016.
  7. Robert F. Kennedy Jr. Was the 2004 Election Stolen? June 01, 2006. Available at: http://www.commondreams.org/views06/0601-34.htm. Last accessed on June 12, 2016.
  8. Source: https://richardcharnin.wordpress.com/2016/04/06/a-preliminary-probability-analysis-of-the-wisconsin-primary/. Last accessed on June 15, 2016.
  9. In this estimation, the proportion of the white (84%) and non-white (16%) electorate and the proportion of the non-white voters voting for Clinton (60%) and Sanders (39%) were kept the same as in the CNN published exit poll.
  10. Bump P. How exit polls work, explained. The Washington Post. April 22, 2016. Available at: https://www.washingtonpost.com/news/the-fix/wp/2016/04/22/how-exit-polls-work-explained/. Last accessed on June 13, 2016.
  11. Ariel J. Feldman, J. Alex Halderman, and Edward W. Felten. Security analysis of the diebold accuvote-ts voting machine. Center for Information Technology Policy. Princeton University, 2006.  Available at: https://citp.princeton.edu/research/voting/. Last accessed on June 14, 2016.
  12. Groves, R. (1989). Survey errors and survey costs. New York. Wiley.
  13. Theodore de Macedo Soares. The Suspect Massachusetts 2016 Primary. com, March 10, 2016. Available at:  https://tdmsresearch.com/2016/06/14/the-suspect-massachusetts-2016-primary/.
Please share this article:
0 0 votes
Article Rating
Notify of
Inline Feedbacks
View all comments
8 years ago

Somebody should send these results to Sanders – he might not be so quick to endorse Clinton.
Big picture: the US needs to go back to paper ballots.

3 years ago
Reply to  MC

It’s not just that. It’s the fact that this occurs in both parties, and likely most major elections. And there is a documentary about that explains that we do not even have the right legally to examine the counting machine software. If you ask me, that is intentional, with the intent of stealing our country, at the base of it.

7 years ago

I have been browsing online more than 4 hours today, yet I never found any interesting article like yours.
It’s pretty worth enough for me. Personally, if all web owners and
bloggers made good content as you did, the web will be a lot more useful
than ever before. http://www.yahoo.net

7 years ago

Attractive component of content. I simply stumbled upon your website and
in accession capital to claim that I get actually enjoyed
account your blog posts. Anyway I’ll be subscribing on your feeds or even I success you
get right of entry to consistently rapidly.

7 years ago

This website definitely has all of the info I
needed concerning this subject and didn’t know who to ask.

7 years ago

I feel this is certainly one of the this type of lot important information for me personally.
And i’m happy studying your article. But would like to remark on few common things, The site
style is wonderful, the articles is truly excellent :
D. Just right process, cheers

Larry A Summers
Larry A Summers
7 years ago

PLEASE INVESTIGATE EXIT POLL DIFFERENCES BY COUNTY IN WISCONSIN, NORTH CAROLINA AND PENNSYLVANIA. Then correlate this with the voting machine type in each county. If there is a correlation, the difference may be due to Russian hacking

5 years ago

Why would it be evidence of “Russian hacking” as opposed to good old American election fraud? Americans screwing over Americans like has happened many times before. When the simplest answer is there in front of you, why look for straw men?


[…] margin of error is indication of fraud. I have hope that this gains traction before the EC vote. Exit Polls versus Reported Vote Counts. 2016 Presidential Primaries – TDMS|RESEARCH "The difference between genius and stupidity is that genius has its limits." […]

7 years ago

So I’m curious, now that we’ve seen the results of the general election: do you have any thoughts about rigging in that?

Chris Bystroff
7 years ago

Thanks, TDMS for your expert poll-watching. What is the next step? It is easy to state the quixotic challenge (let’s go back to paper ballots), but much harder to find the next hand-hold in the long climb.

6 years ago

Great content. I have subscribed to the RSS feed.

4 years ago

Thanks for your hard work. I tried following a couple of links in your footnotes but they’ve gone dead. Maybe you’ll get a chance to update them?