statistical analysis of electoral fraud presidential elections in armenia 2013
DESCRIPTION
This is a presentation of analysis done by Policy Forum Armenia (PFA) revealing wide scale election fraud committed during 2013 presidential elections in ArmeniaTRANSCRIPT
Armenian Presidential Elections 2013
Statistical Evidence of Election Fraud
Sassoon Kosian, PFA FellowApril 20, 2013
Overview
PFA is an independent think tank consisting of professionals in many fields
PFA’s mission includes strengthening the economy and democratic norms in Armenia
PFA provides analysis, recommendations and alternative views on major challenges facing Armenia and the Diaspora
PFA unites professionals and researchers in Armenia and in the Diaspora
Learn more about PFA at www.pf-armenia.org
PFA Reports
Armenia’s 2012 Parliamentary Election, Dec 2012;
Armenia: Averting an Economic Catastrophe, Feb 2012;
The State of Armenia's Environment, Dec 2010;
Armenia-Diaspora Relations: 20 Years Since Independence, Feb 2010;
Yerevan's 2009 Mayoral Election: Statistical Analysis, Sept 2009;
Implications of the World Financial Crisis for Armenia’s Economy, Dec 2008;
Armenia’s 2008 Presidential Election: Select Issues and Analysis, July 2008.
The Elections
On February 18, 2013 presidential elections took place
According to official results, incumbent Serge Sargsyan was declared a winner with 58% of the votes while Raffi Hovhannisyan was reported to get the second place with 37% of the votes
Claims of fraud and protests immediately followed
Shortly after the elections PFA conducted a comprehensive analysis of the results
Key highlights of the analysis are presented here, for the full report visit the PFA website
How We Did the Analysis
Official Voting Results Statistical Analysis
Analysis was exclusively based on official data from CEC
Three Types of Analysis
1. Digit Test
2. Voter Turnout Analysis
3. Candidates Share vs. Turnout
# 1: Digit Test
Digit Test checks if reported numbers show an even use of digits (0, 1, 2, 3, …)
In all natural processes all digits occur at approximately the same rate around 10%
A significant deviation from this rule indicates some kind of artificial interference
Digit Test is particularly effective when looking at the last digit in a number
– E.g. too many zeros could mean someone has inflated the numbers by adding 0’s at the end
– … or, someone has fabricated numbers, and numbers with zeros at the end (200, 50, 5000, …) just happen to be easier to fabricate than numbers without zeros
# 1: Digit Test
Too many 0’s and 1’s
Too few 4’s
Last Digit Last Digit
% D
istr
ibu
tio
n
% D
istr
ibu
tio
n
# 1. Digit Test- Digit Test At Work in Other Countries
# 1: Digit Test
Conclusion
There is evidence of manipulating with vote count numbers in regions outside of Yerevan and Gyumri
# 2: Voter Turnout Analysis
Each polling station has a certain number of registered voters
– Voter’s list is maintained by the Police Department
Not all registered voters go to polls – this is true in every society
Voter Turnout = % of registered voters who actually go to the polling station and cast their vote
Voter Turnout Analysis
Analyzes the distribution of Voter Turnout
Investigates if there are any abnormal patterns in Voter Turnout distribution
Key: In free and fair elections Voter Turnout and Share of Votes for each candidate are expected to have a Normal (Gaussian) distribution
# 2: Voter Turnout Analysis- Overall Votes
We expect a Normal (Gaussian) distribution of Voter Turnout
# 2: Voter Turnout Analysis- Overall Votes
In official reported numbers, the distribution of Voter Turnout shows a significant departure from the expected Normal pattern
We observe unusually high turnout (>70%) in many polling stations
Expected Turnout Official Turnout
Inflated voter turnout
# 2: Voter Turnout Analysis- Votes for Each Candidate
Many polling stations registered unusually high number of votes for Serge Sargsyan and unusually low numbers for Raffi Hovhannisian
Expected Outcome
Expected Outcome
Conclusion
Overall Voter Turnout distribution significantly deviates from expected Normal distribution
Analysis shows highly inflated numbers in many polling stations
In many polling stations, Serge Sargsyan received unusually high number of votes while Raffi Hovhannisian received unusually low number of votes
# 2: Voter Turnout Analysis
# 3: Candidates Share vs. Turnout
Up to this point we have seen some really unusual patterns in the voting numbers and candidates’ shares
We may have started to have suspicions
But we have not yet established a clear link between the abnormal patterns of voter numbers and any particular candidate
We would like to find an answer to a question: Who benefited from inflated voter numbers?
To answer the question we need to see the relationship between Voter Turnout and Candidate Shares
# 3: Candidates Share vs. Turnout
This is another common type of analysis for Election Fraud
We need to plot the Candidate Share by Voter Turnout and look for trend
Candidate Share vs. Turnout Analysis
Analyzes the relationship between Voter Turnout and the share of votes for each candidate
Key: In free and fair elections, the % of votes received by any candidate should not be dependent on the number of voters in each polling station.
In other words, we should not see a trend.
# 3: Candidates Share vs. Turnout- Research on other countries national polls
There is a clear difference in pattern between democratic and non-democratic countries
# 3: Candidates Share vs. Turnout
Analysis reveals a strong relationship between the voter turnout and % of votes received by Serge Sargsyan
Higher turnout meant more votes for Serge Sargsyan
Voter Turnout
% V
otes
for
Ser
ge S
args
yan
# 3: Candidates Share vs. Turnout
We see the opposite picture with the votes received for Raffi Hovhannisian
Higher turnout meant fewer votes for Raffi Hovhannisian
Voter Turnout
% V
otes
for
Raf
fi H
ovha
nnis
ian
Conclusion
We have established a strong link between turnout and the main candidates’ share of votes
Serge Sargsyan significantly benefits from inflated voter turnout while Raffi Hovhannisian suffers from it
# 3: Candidates Share vs. Turnout
Overall Results Of the Analysis
Elections were significantly rigged in favor of incumbent Serge Sargsyan
Analysis included multiple types of statistical tests all pointing to fraud
Fraud was more prevalent in remote rural areas
The extent of fraud was large enough to change the outcome of the elections!
– Fair polls would have either resulted in outright victory for Raffi Hovhannisian or at least given a second round of voting
How Fraud Takes Place
Many forms of fraudulent actions lead to rigged elections
– Ballot stuffing
– Vote bribing
– Intimidation by criminal oligarchy on the street
– Administrative pressure
– Multiple voting (evaporating ink)
– Manipulating the numbers in the books
Voters list is the single most important reason enabling other forms of fraud
Voters’ list is believed to be significantly inflated as it has never taken into account the massive emigration since Armenia became independent
The Role of Criminal Oligarchy
Oligarchy is the primary support base of the ruling regime
Oligarchy provides machinery, muscle and money to the regime
It’s a shared business with common interests
The role of oligarchy is in its highest demand during elections
The “operatives” of election fraud consist of Republican Party grassroots functionaries and street thugs employed by oligarchs
– Often you cannot tell one from the other, they overlap to a large extent
Shared interests are protected by both legal and criminal methods
Voters List Needs To Be Publicized
Current election law does not allow to publish the list of registered voters who actually participated in the vote
– We only need to publicize who voted, not how they voted
This has lead to missing voters’ ballots being fraudulently used by the ruling party
Vicious circle – fraudulently elected National Assembly adopts laws preventing free and fair elections
Election code, in particular on voters list, is a major reason of election fraud
Constitutional Court
Election outcomes were challenged in Armenia’s highest court, the Constitutional Court, by Raffi Hovhannisian’s team and another presidential candidate, Andrias Ghukasyan
The plaintiff demanded to annul the results of Feb 18 elections
– Legal case included specific facts of fraud
– … and PFA’s 10 page long, detailed statistical analysis
The Court didn’t find sufficient grounds to challenge the outcome!
The Court completely ignored PFA’s analysis!
Request of vote recount was met with the most cynical excuse: the Constitutional Court does not have time to count the votes!
Western Observers
Western Observers’ assessments serve as a major factor for the government in creating a legitimate appearance, and they have largely succeeded
As always, the assessment has been inadequate and superficial
Fraud is a multi-billion dollar industry internationally
Fraudsters are smart, they always learn new ways to defraud
Anti-fraud is also a big industry employing powerful analytical methods to learn and fight against fraud
There has to be a shift from ‘observing’ to a deeper and comprehensive analysis
Anything short of that is inadequate with very serious implications for Armenia
Reactions From Society
Strong reaction from the small but vibrant civil society
We have seen a major shift from previous elections – higher sense of responsibility
Everyone knows the elections were rigged but nobody can prove it!
Raffi Hovhannisian and his team have been staging numerous protests in Yerevan and in regions
So far that has not lead to any tangible results
Reactions In Diaspora
Diaspora has traditionally supported the government, whoever might be in charge of it
While the sense of loyalty is much appreciated, we have to realize what it has gotten us to and where it’s leading to
By and large, Diaspora organizations have not participated in democracy building in Armenia
There must be a shift
– Always support the people and democratic norms in Armenia
– Support the government only when it’s democratic
– Pressure the government when it’s not democratic – remember, your voice counts!
Talk to Diaspora organizations and political parties
THANK YOU!
Non-PFA Independent Analysis
The Protesters Are Right: Evidence Suggests More Election Fraud in Last Week’s Elections in Armenia ( Fredrik M Sjoberg, a Postdoctoral Scholar at Columbia University – The Harriman Institute)
Non-PFA Independent Analysis
Independent Russian analysts
Non-PFA Independent Analysis
Independent Russian analysts