What 99.7% Turnout Actually Means
OFFICIAL CEC SPREADSHEETS VIA WAYBACK MACHINE
Before we show you the data, let's understand what 99.7% turnout means in practice.
Imagine your neighborhood has 1,000 registered voters. The government says 997 of them came to vote on election day. Only 3 people stayed home. Think about that for a moment. In your neighborhood of 1,000 people, was there really nobody sick? Nobody traveling? Nobody elderly or disabled who couldn't physically get to the polling station? Nobody who simply didn't care about politics? Nobody working a shift they couldn't leave? Nobody who forgot, or overslept, or just decided to stay home?
Only 3 people out of 1,000 did any of those things. And this wasn't one exceptional precinct -- it was the average across nearly 2,000 precincts in an entire country.
No democracy on Earth produces these numbers. Not even countries that force citizens to vote by law.
The Data: Five Elections, One Pattern
PRECINCT-LEVEL ANALYSIS OF OFFICIAL CEC FILES
OWL recovered official election result files from the Wayback Machine's cache of res.elections.am. We analyzed precinct-level voter counts for every major election from 2005 to 2017. The pattern is consistent and unmistakable.
| ELECTION | YEAR | PRECINCTS | AVG TURNOUT | PRECINCTS >95% | % OF ALL PRECINCTS >95% |
|---|---|---|---|---|---|
| Constitutional Referendum | 2005 | 42 districts | 97.8% | 40 | 95% |
| Parliamentary | 2007 | 1,906 | 99.7% | 1,886 | 99% |
| Presidential | 2013 | 2,016 | 98.2% | 1,998 | 99% |
| Constitutional Referendum | 2015 | 1,985 | 97.2% | 1,983 | 99.9% |
| Parliamentary (aggregate sheet) | 2017 | 2,010 | 99.7% | 1,950 | 97% |
| Parliamentary (per-constituency sheets) | 2017 | ~110 per district | 49-54% | few | <5% |
Look at the last two rows. They are from the same election and the same official file. One says 99.7%. The other says 49-54%. We will return to this.
2007: The Year 99% of Precincts Exceeded 95%
The 2007 parliamentary election produced the most extreme numbers in our dataset. Across 1,906 precincts, the average turnout was 99.7%. Not the maximum. The average.
This means 99% of all precincts reported more than 95% of registered voters showing up. Multiple precincts reported exactly 100% turnout -- every single registered voter cast a ballot.
The following precincts reported that every single registered voter came to vote. Not 99.9%. Exactly 100%:
| PRECINCT | REGISTERED VOTERS | BALLOTS CAST | TURNOUT |
|---|---|---|---|
| Precinct A | 1,970 | 1,970 | 100.0% |
| Precinct B | 1,803 | 1,803 | 100.0% |
| Precinct C | 1,690 | 1,690 | 100.0% |
| Precinct D | 1,840 | 1,840 | 100.0% |
| Precinct E | 1,804 | 1,804 | 100.0% |
| Precinct F | 1,962 | 1,962 | 100.0% |
| Precinct G | 1,867 | 1,867 | 100.0% |
These are not small village precincts with 20 people. These are precincts with 1,690 to 1,970 registered voters -- and the official data says every single one of them voted. Not a single person was sick, traveling, working, or simply uninterested. In seven separate precincts. On the same day.
2013: The Presidential Election
Serzh Sargsyan ran for re-election and won with 58.6% of the vote. Across 2,016 precincts, the average turnout was 98.2%. That means 1,998 precincts -- 99% of all precincts -- reported turnout above 95%. Out of every 1,000 registered voters, on average 982 showed up. Only 18 stayed home. Per precinct. In the entire country.
2015: The Constitutional Referendum
This is the most striking single number in the dataset: of 1,985 precincts, 1,983 reported turnout above 95%. That is 99.9%. Only 2 precincts in the entire country had turnout below 95%. This referendum changed Armenia's constitution from a presidential to a parliamentary system -- a move widely seen as designed to keep Sargsyan in power after his presidential term limit. And according to official data, virtually every registered voter in every precinct in the country came to vote on it.
The Smoking Gun: One File, Two Realities
FILE: resulteng02.04.17.xlsx -- OFFICIAL CEC SPREADSHEET
This is the most important finding in our analysis.
The file resulteng02.04.17.xlsx is the official English-language election result spreadsheet for the April 2, 2017 parliamentary election, published by the Central Election Commission on its own server at res.elections.am. OWL recovered it from the Wayback Machine.
This single Excel file contains multiple sheets. Two types of sheets tell very different stories:
The aggregate summary sheet -- labeled YYH-NAX -- lists 2,010 precincts with an average turnout of 99.7%. According to this sheet, 97% of all precincts (1,950 out of 2,010) had turnout above 95%.
This is the sheet that produces the official "national turnout" figure. It is the number the CEC would cite if asked about election participation.
The same file contains separate sheets for each electoral district, with precinct-level results broken down by constituency. These sheets tell a completely different story:
| ELECTORAL DISTRICT | PRECINCTS | AVG TURNOUT |
|---|---|---|
| District 1 | 108 | 53.9% |
| District 2 | 120 | 52.4% |
| District 3 | 108 | 50.3% |
| District 4 | 114 | 48.2% |
| District 5 | 136 | 49.0% |
| District 6 | 144 | 49.0% |
Average turnout: 49-54%. Precincts above 95%: fewer than 5%.
These two datasets are in the same official Excel file, published by the same Central Election Commission, about the same election on the same day. One says 99.7% turnout. The other says 49-54%. Both cannot be true.
The per-constituency numbers -- 49-54% -- are realistic. They are consistent with what you would expect in a real election. They match normal patterns seen in democracies worldwide.
The aggregate number -- 99.7% -- is not realistic. It is not consistent with any known democratic election anywhere in the world. It matches the pattern of fabricated data.
The aggregate sheet was not computed from the per-constituency data. It was created separately. And its numbers do not add up.
How Armenia Compares to the World
INTERNATIONAL ELECTION DATA COMPARISON
To understand how extreme Armenia's reported turnout was, compare it to real elections in established democracies -- including countries where voting is required by law.
| COUNTRY | ELECTION | YEAR | TURNOUT | MANDATORY VOTING? |
|---|---|---|---|---|
| United States | Presidential | 2020 | 66.8% | No |
| France | Presidential (1st round) | 2022 | 73.7% | No |
| Germany | Federal (Bundestag) | 2021 | 76.6% | No |
| Belgium | Federal | 2019 | ~87% | Yes -- fines for not voting |
| Australia | Federal | 2022 | ~91% | Yes -- fines for not voting |
| Armenia (Sargsyan era) | Multiple | 2007-2017 | 97-99.7% | No |
| North Korea | Supreme People's Assembly | 2019 | 99.99% | Effectively mandatory |
Australia has mandatory voting. Citizens who fail to vote receive a fine. The government sends reminder letters before the election and penalty notices afterward. Despite all of this, Australia reaches roughly 91% turnout. Not 99%. Not 97%. Ninety-one percent.
Belgium also has mandatory voting with fines. Its turnout is around 87%.
Armenia had no mandatory voting law. No fines. No penalty for staying home. And yet, according to official CEC data, Armenian elections consistently produced turnout numbers 6-13 percentage points higher than countries that legally require voting and punish non-voters.
The only country that reports comparable numbers is North Korea.
The Pattern Across a Decade
This is not one anomalous election. It is a consistent pattern across five elections and ten years:
- 2005 referendum: 97.8% average, 95% of districts above 95%
- 2007 parliamentary: 99.7% average, 99% of precincts above 95%
- 2013 presidential: 98.2% average, 99% of precincts above 95%
- 2015 referendum: 97.2% average, 99.9% of precincts above 95%
- 2017 parliamentary (aggregate): 99.7% average, 97% of precincts above 95%
Every single election during the Sargsyan era produced turnout numbers between 97% and 99.7%. The numbers never drop below 97%. They never show the natural variation you see in real elections -- where some precincts have high turnout and others have low turnout based on local conditions, weather, demographics, and enthusiasm.
In real elections, you see a bell curve of turnout across precincts. Some precincts at 40%, some at 60%, some at 80%, averaging out to a national figure. In Armenia's Sargsyan-era data, the curve is compressed into a narrow band above 95%. Nearly every precinct, in every region, in every election, reports nearly identical extreme turnout. This is not what organic voter behavior looks like. This is what happens when numbers are generated rather than counted.
The Numbers Speak for Themselves
The data analyzed in this article comes entirely from official Central Election Commission spreadsheets, published on the CEC's own server at res.elections.am, and recovered from the Wayback Machine's public archive. These are the Armenian government's own numbers, from their own files, on their own website. OWL did not create this data. We only counted what the government published.
The 2017 parliamentary election file -- resulteng02.04.17.xlsx -- contains two datasets about the same election that cannot both be true. The aggregate sheet reports 99.7% national turnout. The per-constituency sheets report 49-54%. The per-constituency numbers are consistent with real elections. The aggregate number is consistent with North Korea.
We are not making accusations about specific individuals. We are not speculating about methods. We are presenting numbers -- the government's numbers -- and comparing them to what is mathematically possible in a real election. The conclusion is in the data itself.
Related Investigations
Sources: Official CEC election result spreadsheets recovered from Wayback Machine cache of res.elections.am. Files analyzed include resulteng02.04.17.xlsx (2017 parliamentary), result files for 2007, 2013, and 2015 elections, and 2005 referendum district-level data. International turnout data from IDEA International voter turnout database. All analysis performed on publicly archived data. No live servers were probed or scanned.