Introduction
As was communicated to the nation in the pre-election statement on Tuesday, December 6, 2016, the Coalition of Domestic Election Observers (CODEO) employed the Parallel Vote Tabulation (PVT) methodology to observe the 2016 presidential and parliamentary elections, just as it did in 2008 and 2012. CODEO employed the PVT to independently verify the accuracy of the presidential election results as declared by the Electoral Commission (EC). As of midnight on Election Day, Wednesday December 7, 2016, CODEO was in the position to verify if the presidential results reflected the ballots cast at polling stations.
The PVT Methodology
The PVT is an advanced and scientific election observation technique that employs well- established statistical principles and utilizes Information Communication Technology (ICT) to observe elections. The PVT involves deploying trained accredited observers to a representative random sample of polling stations starting at 6:00AM. Throughout the day they observed the process and transmitted reports in real-time about the conduct of voting and counting as well as the official vote count to a central election observation point using text messaging.
The PVT technique has been deployed successfully around the world to promote electoral integrity and help defend the rights of citizens to vote and to protect such rights when exercised. The first PVT was conducted in the Philippines in 1986. Since then, the methodology has been deployed in election observation in Bulgaria, Chile, Croatia, Guyana, Indonesia, Montenegro, Nicaragua, Panama, Peru and Ukraine. In Africa, the methodology has been deployed successfully in Madagascar, Malawi, Uganda, Sierra Leone, Zambia, Ghana, Nigeria, Zimbabwe, Burkina Faso, Cote d'Ivoire, Kenya and Tunisia.
Unlike pre-election surveys (which tracks intentions) or exit polls (which tracks recollections), the PVT is based on the actual behavior of electorates in an election. In other words, PVT Observers do not speak to voters about the choices they made at the polls to predict the outcome of an election. Rather after watching the entire voting and counting process, PVT Observers simply record the official vote count as announced by election officials at polling stations and rapidly transmit this information to an observation center via coded text message. Consequently, the PVT data is collected directly from polling stations.
The 2016 PVT Findings
The findings presented in this statement are based on information received from 1,491 PVT Observers located in all the 275 constituencies (including Jaman North constituency) by midnight on Thursday December 8, 2016. All PVT data was thoroughly checked for accuracy and internal consistency. A total of 553,872 Ghanaians cast their ballots at sampled PVT polling stations. The Election Commission's official results are based on 271 constituencies that had been announced as of Friday December 9, 2016.
The percentages of the valid votes garnered by each of the seven presidential candidates as announced by the Electoral Commission compared to the PVT estimates are as follows:
- Ivor Kobina Greenstreet (CPP): EC Official Result = 0.24%; PVT Estimate = 0.27%;
- Nana Konadu Agyeman-Rawlings (NDP): EC Official Result = 0.16%; PVT Estimate = 0.18%;
- John Dramani Mahama (NDC): EC Official Result = 44.40%; PVT Estimate = 44.32%;
- Dr. Papa Kwesi Nduom (PPP): Official Result = 1.00%; PVT Estimate = 1.07%;
- Nana Addo Dankwa Akufo-Addo (NPP): Official Result = 53.85%; PVT Estimate = 53.75%;
- Dr. Edward Nasigre Mahama (PNC): Official Results = 0.21%; PVT Estimate = 0.26%; and
- Jacob Osei Yeboah (Independent): Official Results = 0.15%; PVT Estimate = 0.15%.
For each candidate the official result closely match the PVT estimates. In general, the PVT ranking of the presidential candidates based on their share of the votes is largely similar to what was put out by the Electoral Commission.
Voter turnout rate as estimated by the PVT is 69.33% with a ±0.48% margin of error. This is very close to the 68.62% announced by the Electoral Commission. The official turnout is slightly outside the PVT estimate, but the turnout figures are consistent.
Finally, the PVT estimated rejected ballots stood at 1.60%, with a ±0.09% margin of error, is also quite close to the Electoral Commission's figure of 1.54%.
Table 1: EC Official Results and CODEO PVT Estimates for the 2016 Presidential Election
Candidate | Party | EC Official Result | CODEO PVT | |||
Estimate | Margin of Error | Range | Match | |||
Ivor Kobina Greenstreet | CPP | 0.24% | 0.27% | ±0.04% | 0.23% to 0.32% | ü |
Nana Konadu Agyeman-Rawlings | NDP | 0.16% | 0.18% | ±0.04% | 0.14% to 0.21% | ü |
John Dramani Mahama | NDC | 44.40% | 44.32% | ±1.25% | 43.07% to 45.57% | ü |
Dr. Papa Kwesi Nduom | PPP | 1.00% | 1.07% | ±0.16% | 0.91% to 1.22% | ü |
Nana Addo Dankwa Akufo-Addo | NPP | 53.85% | 53.75% | ±1.29% | 52.47% to 55.04% | ü |
Dr. Edward Mahama | PNC | 0.21% | 0.26% | ±0.10% | 0.16% to 0.36% | ü |
Jacob Osei Yeboah | IND. | 0.15% | 0.15% | ±0.03% | 0.11% to 0.18% | ü |
Note (1): 1,491 of 1,500 PVT sample polling stations reporting. Note (2): Margin of Error and Range based on 95% Confidence Level. |
Conclusions
Given that the official result announced by the Electoral Commission for each candidate falls within their respective PVT estimated confidence ranges, CODEO can confidently confirm that its PVT estimates are consistent with the official presidential results.
Not only do CODEO's PVT estimates match the official results of the EC, our observers also reported that the main two parties, NDC and NPP, had party agents at over 99% of polling stations and that at over 99% of polling stations, both NDC and NPP party agents signed the official declaration of polls and were given copies of that document.
Further, evidence from CODEO's PVT Observers showed that voting and counting were conducted in accordance with CI.94, the electoral laws, and the Constitution.
Thus, CODEO is confident that the results of the 2016 presidential polls declared by the Electoral Commission are the accurate reflection of how Ghanaians voted in the December 7 polls. Accordingly, CODEO advises the contestants in the presidential election, their supporters and the general public to place confidence in the official presidential election results as declared by the Electoral Commission.
CODEO also commends, His Excellency, John Dramani Mahama, and all other contestants for the presidential elections for conceding defeat, and congratulating the president elect, Nana Addo Dankwa Akuffo-Addo for his victory.
CODEO further commends the maturity Ghanaians have shown by remaining calm and peaceful throughout the voting, counting and collating processes.
CODEO reiterates its earlier call for the need for political parties, their leaders and the entire political class to take adequate measures to bridge political polarization that characterized the 2016 election campaign, especially between the NDC and the NPP, and to reduce the winner- takes-all tendencies in our politics. We need to do this for the sake of peace, unity, good governance and sustained socio-economic development of our dear country.
On behalf of the Advisory Board of CODEO and all its member organizations, we would like to thank the thousands of Ghanaians across the country that volunteered to serve as nonpartisan election observers and worked tirelessly to protect the integrity of the December 7, 2016 polls.
CODEO also commends the security agencies for their professionalism in maintaining peace during the period of voting, counting and collation of results.
The Coalition of Domestic Election Observers (CODEO) would also like to acknowledge the generous support of the American people through the United States Agency for International Development (USAID) and the Embassy of the Kingdom of Netherlands, which made it possible for the Coalition to observe the December 7, 2016 presidential and general elections.
In this election, it is Ghana and Ghanaians who have WON. It is not an exclusive victory for any political party or candidate.
GOD BLESS OUR HOME LAND GHANA
Signed:
Professor Miranda Greenstreet and Justice VCRAC Crabbe
Joint Chairpersons
(For and on Behalf of the Advisory Board)
CODEO Secretariat
Saturday December 10, 2016
ANNEX
2016 PARALLEL VOTE TABULATION SAMPLING METHODOLOGY
Introduction
In 2016 PVT, CODEO employed a nationally representative sample of 1,500 polling stations drawn from the official list of polling stations provided by the Electoral Commission. The sampled polling stations are spread across the 275 constituencies located with the ten regions of the country. To ensure national representativeness, the sample was stratified at the regional and constituency levels.
Stratification at regional level: At the regional level, the percentage shares of regions in the total number of polling stations in the country were used to distribute the 1,500 PVT polling stations across the regions. This ensured that each region's percentage share in the 1,500 PVT sampled polling stations closely matches its percentage share in the total number of polling stations in the country (Compare percentage shares of regions in columns 3 and 5 in Table 1).
Table 1: Distribution of sampled polling stations across regions in Ghana
Polling Station | % Polling Station | PVT Polling Station | % PVT Polling Station | |
Western | 2,941 | 10.1% | 152 | 10.1% |
Central | 2,482 | 8.6% | 128 | 8.5% |
Greater Accra | 4,762 | 16.4% | 246 | 16.4% |
Volta | 2,506 | 8.6% | 130 | 8.7% |
Eastern | 3,243 | 11.2% | 168 | 11.2% |
Ashanti | 5,190 | 17.9% | 269 | 17.9% |
Brong Ahafo | 2,963 | 10.2% | 153 | 10.2% |
Northern | 2,688 | 9.3% | 139 | 9.3% |
Upper East | 1,226 | 4.2% | 64 | 4.3% |
Upper West | 991 | 3.4% | 51 | 3.4% |
Total | 28,992 | 100.0% | 1,500 | 100.0% |
Stratification at regional level: The constituency level stratification ensured that the percentage share of a constituency within a particular region in the number of PVT polling stations allocated to that region closely matches its percentage share in the total number of polling stations in that region (Compare percentage shares of constituencies in columns 3 and 5 in Table 2).
Table 2: Distribution of sampled polling stations across constituencies in Upper West Region
Polling Station | % Polling Station | PVT Polling Station | % PVT Polling Station | |
Wa Central | 150 | 15.1 | 8 | 15.7 |
Wa West | 118 | 11.9 | 6 | 11.8 |
Wa East | 104 | 10.5 | 5 | 9.8 |
Nadowli/Kaleo | 93 | 9.4 | 5 | 9.8 |
Daffiama/Bussie/Issa | 46 | 4.6 | 2 | 3.9 |
Jirapa | 115 | 11.6 | 6 | 11.8 |
Lambussie-Karni | 70 | 7.1 | 4 | 7.8 |
Lawra | 80 | 8.1 | 4 | 7.8 |
Nandom | 68 | 6.9 | 3 | 5.9 |
Sissala West | 67 | 6.8 | 4 | 7.8 |
Sissala East | 80 | 8.1 | 4 | 7.8 |
Total | 991 | 100.0 | 51 | 100.0 |
Training of CODEO observers (including PVT observers)
All CODEO observers were given extensive training and were provided with a manual on their role and responsibilities as observers, and particularly, the “dos” and “don'ts” of a CODEO observer prior to their deployment. As a pledge of neutrality, non-partisanship and transparency in their election observation work, all CODEO observers were made to sign an undertaking at the end of the training. Furthermore, all CODEO observers participated in 2 full simulation exercises on December 3 and 5, 2016, respectively, to test all aspects of the communication system ahead of the elections on December 7, 2016. During the exercise, all CODEO observers sent text messages to the CODEO National Information Center (CNIC) at the Kofi Annan International Peacekeeping Training Centre (KAIPTC).
CODEO observers (including PVT observers) work on Election Day
Data for the PVT was collected by 1,500 well-qualified, reliable and experienced Rapid Response Observers (RROs) out of 7,000 stationary CODEO observers deployed on the Election Day. CODEO observers (including those in PVT polling stations) sent reports, including vote count figures from their assigned polling stations using specially coded text messages. These were received directly into the computer database at the CODEO National Information Center (CNIC) at the Kofi Annan International Peacekeeping Training Centre (KAIPTC). Upon receipt of text messages from observers, CODEO trained Data Entry Clerks at the CNIC at KAIPTC called observers to verify and confirm the information received. The vote count figures received into the database were used to estimate candidates' share of the valid vote count to determine the winner of the presidential election.