Parallel Vote Tabulation FAQ

What is Parallel Vote Tabulation?

Parallel Vote Tabulation (PVT) is an advanced observation methodology that builds on traditional observation. It involves deploying trained and accredited non-partisan observers to a representative random sample of polling stations. Using statistics, this enables CODEO to make statements about the conduct of voting and counting at all polling stations. It also enables CODEO to verify the accuracy of the tabulation process and the official results.

What is tabulation?

Tabulation is the process by which election officials add together the vote counts from individual polling stations to determine the official results. This takes place at constituency collation centers and at the Electoral Commission in Accra. It is not possible to directly observe tabulation as there is no way to know definitively if the vote counts that arrive at the collation centers or at the Electoral Commission are accurate or if the collation itself has been without blemish.

Isn’t the Election Commission responsible for announcing the official results?

Yes, only the Election Commission is responsible for announcing the official election results. CODEO will use the information gathered using the PVT methodology to be able to verify the conduct of the tabulation process and the accuracy of the results, but it remains the responsibility of the Electoral Commission to announce the official results.

Is a PVT the same as an exit poll?

No, a PVT is not an exit poll. An exit poll involves asking voters for whom they voted. CODEO observers do not ask voters for whom they voted. CODEO observers simply record the official vote count of the polling station after having observed voting and counting.

Does a PVT involve observers counting ballots?

No, CODEO observers do not count the ballots. Only the election officials count ballots. CODEO observers observe the voting and counting process and then record the official vote count as announced by the polling station officials.

Why should anyone trust statistics?

Statistics are a powerful and proven tool to draw accurate conclusions about a larger population based on a random sample of observations. The key is that the sample must be drawn at random. To draw a random sample one must have a complete list of all the objects to be sampled. Drawing a random sample of polling stations in Ghana is possible because the Election Commission maintains a list of all polling stations and has shared the list with CODEO.

How can I be sure the PVT sample is representative?

Experienced statisticians will draw the sample according to widely accepted sampling principles. The sample was stratified by region and constituency to ensure that it is representative of all of Ghana. Stratifying by region, for example, ensures that the percentage of polling stations in the sample for each region is equal to the percentage of total polling stations for each region. For example, since 11 percent of all polling stations are in Western region then 11 percent of the sample polling stations are also in Western region.

How was the PVT sample size determined?

Experienced statisticians selected the sample size based on statistically acceptable practices. A sample of approximately 1,000 polling stations was selected to ensure that it is large enough to draw conclusions about Ghana as a whole. Over 500,000 voters are likely to cast their ballots at the sampled polling stations. However, the sample is not large enough to draw conclusions about individual parliamentary races.

What will be the margin of error for the PVT?

All statistics have a margin of error. This is because statistics are only able to estimate the true result. The margin of error is based on the sample size, but is independent of the percentage of polling stations sampled. The final margin of error for the PVT will only be known when the information is collected from polling stations, but it will be within statistically accepted standards and will likely be between one and two percent.

How confident is CODEO in the findings of the PVT methodology?

The Confidence Level used for the PVT will be at least 95 percent. This is a very high level and is the accepted level used by statisticians around the world. Thus, there will be a 95 percent confidence that the true result lies within the margin of error of the estimated figure. Therefore, the people of Ghana and its political leaders should have very high confidence in CODEO findings from the PVT.

Who will conduct the PVT?

CODEO is deploying more than 4,000 domestic observers for the 2008 elections of which approximately 1,000 will be deployed using the PVT methodology. CODEO is responsible for all elements of the observation effort including the PVT, but has received financial support from the United States Agency for International Development (USAID) and technical assistance from the National Democratic Institute (NDI). Regardless of financial and technical support, CODEO maintains ultimate control of the design and implementation of all its projects.

Where else has the PVT methodology been used?

The PVT methodology is routinely used around the world by domestic observer groups to promote electoral integrity. It was first used by Citizens Movement for Free Elections (NAMFREL) in the Philippines in 1986. It has since been employed to varying degrees by observer groups in numerous countries, including: Bulgaria, Chile, Croatia, Guyana, Indonesia, Madagascar, Malawi, Montenegro, Nicaragua, Panama, Peru, Sierra Leone, Uganda, Ukraine, Zambia and Zimbabwe

Does the Election Commission approve of the use of the PVT methodology?
CODEO has informed the Electoral Commission that it intends to use the PVT methodology. No special approval or accreditation is required to use the PVT. All CODEO observers will be accredited by the Election Commission. The Election Commission has raised no objections to CODEO conducting a PVT.

How is the PVT sample different from the sample for a political poll?

A PVT involves sampling polling stations and then collecting official vote counts from those sampled polling stations. It is possible to draw a truly random sample of polling stations because the Electoral Commission maintains a list of polling stations. Each polling station is assigned a unique number and then random numbers can be selected (like the numbers in a lottery) to determine which polling stations are included in the sample. Thus, the public and political contestants should have a great deal of confidence in quality of the sample used for the PVT and the results from the PVT.

In contrast, political polling involves sampling individuals rather than polling stations. It is much more difficult, though, to have a truly random sample of individuals. For a truly random sample of individuals one would need to have before Election Day a list of all the people who will vote on election day. However, no such list exists. The voters list is an approximation of the list of everyone who will vote on Election Day, but many registered voters will not vote. Even if individuals were selected at random from the voters list, it would be very time consuming to contact the people selected. Polling in other countries is done by telephone to reduce the time and cost of contacting selected individuals, but the voters list in Ghana does not contain telephone numbers and individuals selected would have to be physically found and interviewed in person. This is time consuming and expensive. Instead, political polling in countries such as Ghana typically does not involve selecting individuals at random from a voters list, but selecting areas of the country at random (such as electoral areas) and then using various techniques to simulate choosing people at random in those areas (such as going to a local market and interviewing every fifth person alternating between men and women). While this methodology provides valuable information and such polling is very useful, it is not as robust as true random sampling as is used in a PVT.

Further, political polling involves asking people questions to which their answers may or may not be accurate on Election Day. Respondents may say that they are going to vote and they may intend on voting, but they may not actually vote on Election Day. Similarly, respondents may say that they intend on voting for Candidate A, but later may change their mind and vote for Candidate B. These issues do not negate the value of political polling to estimate popular sentiment, but it is not as accurate an estimate as adding together the official vote counts of how voters actually voted from individual polling stations as is done in a PVT.

Thus while political polling is very useful, information from a PVT is more accurate and reliable than polling data because it is based on a truly random sample and does not involve asking people how they voted but recording the official vote counts which reflect how people actually voted.

What is the PVT sample size?

The sample includes approximately 1,000 polling stations located across all of Ghana. There are sampled polling stations in every region and every constituency of the country. It is anticipated that over 500,000 voters are likely to cast their ballots at these polling stations on Election Day. Prior to election day it is not possible to give the exact number of location of the sampled polling stations for security purposes. If the number and location of the sampled polling stations were known in advance then it would be possible for some group to manipulate the results by only focusing only on non-sampled polling stations. Until the official results have been announced and accepted by all parties it is necessary to keep the number and location of the sampled polling stations confidential. Afterwards CODEO will make the sample public.

Can we have the list of PVT polling stations?

A list of the sampled polling stations can only be provided after the election. It is critical to the success of the exercise that the number and location of the sampled polling stations be kept a secret until after election day. If the number and location of the sampled polling stations were known in advance then it would be possible for some group to manipulate the results by focusing only on non-sampled polling stations. Until the official results have been announced and accepted by all parties it is necessary to keep the number and location of the sampled polling stations confidential. Afterwards CODEO will make the sample public. To ensure the integrity of the exercise a sealed copy of the list of sampled polling stations has been given to [insert name here].

What is the goal of the PVT?

The PVT is intended to help give political contestants and the Ghanaian people greater confidence in the accuracy of the official results. By independently collecting and adding together the official vote counts from a representative random sample of polling stations CODEO will be able verify the accuracy of the official results. If two independent groups both add together the same figures and arrive at the same totals then everyone, political contestants and voters alike, should have greater confidence in the accuracy of the official results. Recently in Zambia where a presidential by-election was decided by less than 35,000 votes, the results a PVT conducted by the local domestic observer group FODEP closely matched the official results as announced by the electoral officials. The release of the PVT results gave greater confidence that despite the closeness of the election that the official results accurately reflected the will of the Zambian people.

Will the PVT results exactly match the official results?

No, the PVT results will not match the official results nor should they. The PVT results are an estimate based on a representative random sample of polling stations while the official results and based on the vote counts at every polling station. However, the PVT results and the official results should match in terms of the order of the candidates and the magnitude of the vote received by each candidate. For example, for the recent presidential by-election in Zambia the results of a PVT conducted by the local domestic observer group FODEP placed the candidates in exactly the same order as the official results and the magnitude of the vote for every candidate from the PVT closely matched the official results.

Zambia 2008 Presidential By-Election
Candidate         Official Results             PVT Results
Banda               40.1%                           39.2%
Hichilema          19.7%                           20.8%
Miyanda            0.8%                             1.0%
Sata                 38.1%                           37.4%
Rejected           1.3%                             1.7%