richard verifying election results (ndi)
TRANSCRIPT
Verifying Election ResultsVerifying Election Results: : Employing Statistics, Databases and Mobile Employing Statistics, Databases and Mobile
Phones to Enhance MonitoringPhones to Enhance Monitoring
Richard L. Klein
Senior Advisor, Elections
The ProblemThe Problem
Elections are an accountability mechanismElections need to reflect aggregate citizen
preferences to be effectiveHowever, citizens may have little
confidence that official results reflect their collective preferences
Notably, results may be changed after they leave polling stations
The ResponseThe Response
Parallel Vote Tabulation (PVT) Citizens independently collect official
results from individual polling stationsTransmit figures to a central locationAdd numbers togetherAnalyzed the dataShare findings
Need for SpeedNeed for Speed
StatisticsDatabasesSMS (Texting)
PVTs needs be completed quickly - before official results are announced
Electoral authorities will always have more resources than citizen initiatives
Citizens, however, can still be faster by employing:
StatisticsStatistics
Reduces overall burden on citizen groups
Greater attention to difficult to reach polling station
Focus on Quality over Quantity
Use statistics to deploy accredited, well-trained observers to a representative random sample of polling stations
DatabasesDatabases
Ensures monitors are deployed to all sampled polling stations
Enhances data securityFacilitates collection of data and data
quality controlSpeeds and simplifies data analysis
Employ databases to manage information about monitors as well as observation data
SMS (Texting)SMS (Texting)
Speeds and simplifies movement of information
Allows greater effort to be made collecting data from remote polling stations
Permits two-way communication increasing response rates and improving data quality
Develop SMS protocols to enable monitors to transmit observation data by text message
PVT RequirementsPVT Requirements
Comprehensive monitoring of the entire electoral process
Continuous monitoring at sampled polling stations from opening until the end of counting
Coherent top-down network of citizensTraining of citizens on monitoring and
reporting
PVT BenefitsPVT Benefits
More information available more quickly
Systematic information on positive and negative aspects of election
Ability to verify official results or suggest “true” result of an election
Increased voice citizen on conduct of elections
Ghana 2008/09Ghana 2008/09
CODEO/CDD-GhanaNational Network with experience
observing four national electionsPVT for presidential electionSample size of 1,070 polling stationsUnified database of monitor
information and observer dataMoved 90% observer data via SMS
SampleSampleStratified geographically to ensure
representativenessTable 1. Distribution of Sampled Polling Stations by Region
Region
Official Number of
Polling Stations
% Share of Total Polling
Stations
Number of Sampled Polling Stations
% Share of Sampled Polling Stations
Ashanti 3,667 17.5% 185 17.3% Brong Ahafo 2,382 11.3% 120 11.2% Central 1,763 8.4% 93 8.7% Eastern 2,609 12.4% 135 12.6% Greater Accra 2,457 11.7% 125 11.7% Northern 2,043 9.7% 103 9.6% Upper East 1,028 4.9% 50 4.7% Upper West 854 4.1% 44 4.1% Volta 1,965 9.4% 100 9.3% Western 2,240 10.7% 115 10.7% Total 21,008 100.0% 1,070 100.0% Note: Number of polling stations provided to CODEO by the Electoral Commission.
Tracking Monitors and ReportsTracking Monitors and Reports
Analyzing DataAnalyzing Data
Drilling DownDrilling Down
Verifying ResultsVerifying Results
SMS CodesSMS Codes
Outgoing SMS MessagesOutgoing SMS Messages
Response RateResponse Rate
CODEO StatementCODEO StatementConclusions CODEO can confidently confirm that the results of its PVT are consistent with the official presidential results as announced by the Electoral Commission. The order of the candidates in terms of voting results is the same between the PVT data and the official results. The percentage of the vote received by each candidate is also consistent between the PVT data and the official results. Our estimate of the voter turnout which is 70.0 percent also matches with the official figure of 69.5 percent. Similarly, our estimated percentage of reject ballots is 2.2 percent, which is quite close to the Electoral Commission’s figure of 2.4 percent. The official result for every candidate falls well within the PVT estimated range given the margin of error. This range is based on a 95 percent confidence level. It should be noted however, that because the PVT data is based on a sample of polling stations, the PVT results should not necessarily match the official results, but rather the official results should fall within a range of values statistically estimated from the PVT data. It would have been of concern if any official results fell significantly outside the PVT estimated range. Clearly, this is not the case for the 2008 presidential election result (see Table 3). Table 3. PVT Estimates and Official Results (as of December 10, 2008)
Candidate Party PVT
Estimate Official Results
Nana Addo Dankwa Akufo-Addo NPP 49.8 49.1 Prof. John Evans Atta Mills NDC 47.4 47.9 Dr. Paa Kwesi Nduom CPP 1.3 1.3 Dr. Edward Nasigrie Mahama PNC 0.8 0.9 Emmanuel Ansah-Antwi DFP 0.3 0.3 Kwesi Amoafo-Yeboah Independent 0.2 0.2 Kwamena Adjei RDP 0.1 0.1 Thomas Ward Brew DPP 0.1 0.1 Note: PVT estimate range is based on a 95% confidence level
ZESN Statement – 31 March 08ZESN Statement – 31 March 08While it is the responsibility of the Zimbabwe Electoral Commission (ZEC) to announce the official results of the election, it is the legal duty of election observers to provide the people of Zimbabwe with independent non-partisan information on all aspects of the electoral process. This information helps build confidence in the process. Based on official vote counts publicly displayed at polling stations as required by the Electoral Act, ZESN collected data from a representative random sample of 435 polling stations spread across all 10 provinces and after a careful and thorough analysis of the data, ZESN is able to make the following projections about the presidential elections: Morgan Tsvangirai is projected to receive the highest number of votes in the March 29, 2008 presidential election; Robert Gabriel Mugabe is projected to receive the second-highest number of votes; and, Herbert Stanley Simba Makoni is projected to receive the third-highest number of votes The projections are follows:
Presidential Poll Projections – 29 March 2008 Harmonized Elections
Candidate Projected Percent
Vote Margin of
Error 95% Confidence
Interval Makoni, Herbert Stanley Simba 8.2% 1.1% 7.1 to 9.3 Mugabe, Robert Gabriel 41.8% 2.6% 39.2 to 44.4 Towungana, Langton 0.6% 0.1% 0.5 to 0.7 Tsvangirai, Morgan 49.4% 2.4% 47.0 to 51.8
Official results were finally announced on 02 May 08 (32 days after ZESN and 34 days after the election). Official, Mukoni received 8.3%, Mugabe 43.2%, Towangana 0.6% and Tsvangirai 49.7%.