The Democratic Reforms Party questions PA Consulting's exit-poll resuts

Exit polls conducted on the Election Day by different organizations inspired heated debates in the Azerbaijani society. Thus, the Democratic Reforms Party (DRP) expressed concern and doubt in the accuracy of the exit poll results conducted by the PA Consulting Group. In order to clarify the situation the DRP commissioned an expert evaluation of polling results.

The example of the 22nd Nasimi constituency, from which the chairman of the party Asim Mollazade was running for parliament, revealed a number of inaccuracies and violations in the process, says the party’s official statement.

There are total of 37,353 voters officially registered in the district. According to the official data, the turnout for that constituency was 34.53% or about 13,002 voters. However, the PA Consulting claims their polling covered 24,388 voters. In experts opinion, if we sum the number of votes given for all 14 candidates from the given constituency, we get 24386 people. Thus, 2 people are just "lost" in a process…

PA Consulting conducted questioning in 13 out of 28 polling stations. On the website with the polling results (http://www.usaidazexitpoll.com) one can find a percentage of people who voted in each polling station. In one case (polling station #2 in school # 152), this number is at 110%. Thus, instead of 1,277 people who were registered to vote at that precinct, 1,405 people took part in the exit poll.

Experts also noted that if we take the official voters turnout rate for each polling station, we will see that 500 people voted there. But it remains unclear where the rest 900 people came from, with the 110 % turnout rate. Even if the turnout that day was 100%, we still have about 128 additional voters (10% of the total).

Moreover, according to experts, when we compare the PA Consulting figures on the number of people questioned on each polling station with the Central Election commission (CEC) data on the number of voters, it turns out that 10,560 people were questioned.

24,386 people according to the PA Consulting took part in the polling (sum of all the voters who voted for all the candidates), and the number of voters in 13 out of 28 polling stations was 10,560 people. Thus, about 14,000 people who voted, according to this exit poll, did not come to the polling station – if we compare this with the turnout rate for each polling station.

There was probably a certain interval used during polling by the PA Consulting. But it seems that they have weighted the voting at the 13 polling station covered by their questioning, the same way as the 15 polling stations at which they were not present. However, it is not mentioned anywhere in the methodology, what formula was used to calculate the total result.

We can guess that there was a certain coefficient. According to the PA Consulting Group, 24,386 people voted. Before the elections, this polling firm announced that they will question every 4 person coming out of the polling station. If we multiply 4 by 24,386 we get 97,500 people questioned by the PA Consulting pollsters (with the total number of voters in that district – 37,653) .

According to the PA Consulting exit poll results for the 22nd Nasimi constituency, Hajimurad Sadadinov is leading with 33.6% (8187 voters), Yagub Mamedov is on the second place with 28.6% (6969 voters), Asim Mollazade got 17.2% (4204 voters) and the other candidates - less than 4% in total.

Within the margin of error, Mollazade could have received 20% of votes, Hadjimurad Saddadinov – 28% and Mamedov – 22% accordingly. However, there is no information on the website on how this margin of error was calculated and why it is different for each candidate. He margin of error for Mamedov is 6.63% and for Sadadinov – 5.63%. But if the latter received more votes, then the margin of error should have decreased accordingly.

According to the CEC, Mollazade received 33.57% of votes (4635 voters). So if we accept the CEC statistics, other candidates could have received the total of 8,637 votes. Let’s assume that no one voted for other than Mollazade, Mamedov and Saddadinov. Then 8,637 votes were given for both Sadadinov and Mamedov. With the max. margin of error (according to the PA Consulting), it turns out that 5,600 people voted for Saddadinov (minimum) and 6,820 for Mamedov (min.) – 12,420 total (though officially we have only 8,637 voters)
Again we have 4,000 additional votes – all this we guess that not a single voter has voted for some of the other 11 candidates – which is in principle impossible.

According to the party’s chairman, monitoring by DPR observers revealed that some of political opponents used the "carousel" method (multiple voting) at some polling stations. "Our observers were able to stop this at some polling stations - but they used the same method for exit polls," he says.

"Exit polls can be used not only as the tool to forecast election results, but also as a mechanism for society to control the process and make sure it is run fairly. However, this is possible only if the people who are involved in polling do their job conscientiously and without bias, then the results won’t depend on political affiliation of the staff doing the polling," Mollazade believes.

Day.Az