Today, 9 days before the presidential elections, TNS opinion is presenting the results of its survey "Poll: Armenia on the Eve of Presidential Elections" describing the voting preferences and motivations of the Armenian voters, but also their views on the on-going campaigns of the various candidates. For this TNS opinion poll, implemented together with their local partner IPSC, 1 609 face-to-face interviews were conducted in all 10 regions (marzes) and 12 communities of Yerevan between 31 January and 5 February 2013, using the highest possible standards and extensive quality control measures, as outlined below. The poll was commissioned by European Friends of Armenia (www.EuFoA.org) in order to contribute to a factual debate ahead of the elections (see below for more details about our motivation and the organisations involved).

“Using a large number of quality checks, we carried out this poll in the middle of the campaign period and received very reliable and consistent data.” comments Dr Steve Schwarzer, Director for Methods and Statistics at TNS opinion, Brussels. “Now we see that President Sargsyan still has the most certain voters and almost twice as many as runner-up Raffi Hovhannisyan. However, the respondents say that so far Hovhannisyan’s campaign appears more convincing, focusing on programmatic campaigning around mainly economic topics. Strikingly, he now gains more support from the former Prosperous Armenia voters than the incumbent President, a clear change compared to our last poll.”

As a result, if presidential elections took place last week, Serzh Sargsyan would have scored 58,0% [-10,6% compared to our January 2013 poll] of valid votes (valid responses are without "Don't know" and "Refuse to answer"; raw data with "Don't know" and "Refuse to answer": 40,2% [-3,8%]). The poll now rates Raffi Hovhannisyan at 33,0% [+12,2%] of the valid votes (raw data: 22,9% [+9,6%]). To evaluate the likelihood of a second round, TNS opinion also assessed the certainty to vote and analysed the characteristics of respondents who did not reveal their voting intentions. Following this analysis, Hovhannisyan currently is positioned in a bandwidth between 26% and 34%, while Sargsyan scores between 58% and 66%, looking very likely to win in the first round – based on the situation two weeks before the elections.

The voters of Paruyr Hayrikyan, a candidate who was shot into the shoulder under unclear circumstances on 31 January (after the first fieldwork day of this opinion poll), are now much more certain that their decision is final and his awareness among the public has strongly increased. However, the people intending to vote for him remain stable at 4,9% [+0,1%] of the valid votes. Hrant Bagratyan now stands 1,6% points lower than in January, scoring 2,8% of valid votes. As in January, the remaining four of the eight official candidates are largely unknown to the poll respondents and altogether score less than 1,5% of valid votes.

“Apart from the candidates’ ratings, we were positively surprised by voters’ trust in these elections, an aspect which is important for Armenia’s democracy and its political weight. More than 2/3 of the population believes that the upcoming presidential elections will be organised in a better way than previous elections and a similar number of people perceives media to report in a more neutral and fair manner than in the past. The number of people who are certain that they will turnout to vote is almost the same as in the very competitive elections in May 2012.” says EuFoA Secretary General, Dr Michael Kambeck. 

The design of the random sample, fieldwork monitoring, survey implementation and quality control were developed together by TNS opinion and IPSC. Quality control measures were a particular focus, including call-backs and parallel visits, as well as extensive database consistency checks. For more details about this, please confer the background information below.

EuFoA has commissioned the poll for several reasons. In particular we want to:

• Raise public awareness about the elections, issues important to the people, and public attitudes towards politics, political parties and their leaders.
• Compare the findings of the poll on socioeconomic and political issues with earlier polls.
• Contribute to a more meaningful debate, based on political content more than party political propaganda and myths.
• Increase public transparency and trust in a realistic picture of political life in Armenia.
• Increase pressure for a good technical conduct of the elections by providing comparative data for the assessments of international observers. 

The complete poll report can be found in Armenian and in English on our website at www.eufoa.org/en/publications.

EuFoA is happy to provide commentary or background analysis in English, German, French, Italian, Polish, Russian and Armenian. Members of the Europe-Armenia Advisory Council may also be available for comments; for enquiries please contact our secretariat.

For more information on EuFoA and the Europe-Armenia Advisory Council, please visit our website at www.EuFoA.org.

High resolution picture material is always available on our website and upon request. It is free to use with a reference “Copyright:www.EuFoA.org”. 

Background on the poll: 

Methodology

• A multi-stage, random (probability) sampling design was used. In the first stage, primary sampling units (PSU) were selected from each of the administrative regional units. This was agreed as best methodology between TNS opinion and IPSC .
• 1 609 interviews conducted face to face between 31 January – 5 February 2013.
• Interviews were conducted in all 10 regions (marzes) of Armenia and in all Yerevan communities. The sample was distributed proportionally to reflect the population distribution in Armenia, with 34.4% of interviews conducted in Yerevan and 65.6% in the marzes.
• ·        Interviewers selected households using the random walking method to ensure that there is no selection bias. To ensure a random selection, the person interviewed in each household was the adult whose birthday was closest to the day of the interview.
• If a respondent was not immediately available, 1 to 2 call-back visits were done to conduct the interview later. If a call-back visit was not successful or if a respondent could not take part for other reasons, the interviewer approached the next randomly selected household according to the random walking method.
• The sampling procedure was monitored by using a detailed contact sheet for each interviewer. The data base was analysed by TNS opinion in order to ensure that interviewers followed the instructions.
• The data was weighted according to age and gender and is accurate to a maximum margin of error of ±2.4% for the overall sample. 

Quality control

• The design of the random sample, fieldwork monitoring, survey implementation and quality control were developed together by TNS opinion and IPSC.
• 22% of the interviews conducted were accompanied by IPSC quality control coordinators to ensure high quality and compliance with the standards agreed. All questionnaires were monitored by IPSC specialists to check the individual quality and logical consistency of the data. 6% of the questionnaires were checked by return visits. 25% of the entered paper questionnaires were double-checked in the database to ensure accuracy of processed data. Following all these checks, 31 interviews were rejected (1.9%) resulting in the aforementioned valid sample of 1 609 cases.
• Route records were checked by TNS opinion and IPSC independently to ensure that interviews had been selected randomly.
• 33% of the sample was called back by IPSC to confirm demographic details and that the interview took place in the conditions recorded. Additionally, 15% of interviews were randomly selected by TNS opinion and called back to confirm key demographic information and two opinion questions. This consistency check-back resulted in the required compliance level with the originally collected data.
• Extensive checks were conducted on the data file by TNS opinion to see if there were any logically inconsistent or inaccurate interviews, data points or invalid information. This was assessed positively and no case exclusions were made.