Undecideds Swing New Poll Numbers: Armenia’s Parliamentary Elections Take Shape

2026-02-26 - 10:54

Listen to the AI generated audio article. Your browser does not support the audio element. As Armenia prepares for the June 7, 2026 Parliamentary elections, the first wave of newly-released voter behavior survey by EVN Report, conducted by the Armenian Election Study (ArmES), highlights the nuanced preferences of the electorate, casting Armenian society as cautiously optimistic on security and the economy, while generally ambivalent about their vot e preference of which party should lead the country. The first survey of its kind to be produced in Armenia, with three more waves of polling to be conducted until the elections, the results provide disaggregated data on non-committed voters, generally referred to as undecided voters. Considering the important role that non-committed voters played in the 2021 Parliamentary elections, this survey offers a methodologically sophisticated modeling of non-committed voters, discerning both their policy preferences and the political party that their potential vote leans towards. The overall results demonstrate that Civic Contract, led by Prime Minister Nikol Pashinyan, would win the plurality of the votes if the elections were held now, but pending on voter turnout, may fall short of securing a single-party government. The main opposition in the current Parliament, the Armenia Alliance led by former President Robert Kocharyan, is overtaken in the poll as the runner-up by the newly-formed Strong Armenia Party, led by Narek Karapetyan. While Civic Contract enjoys a substantive lead in comparison to the other parties in the electoral landscape, it remains the only pro-Western, pro-Velvet party that is above the threshold to enter Parliament. Substantively, the results demonstrate that Armenia’s democratic opposition, and possible coalition partners to Civic Contract, are absent from the electoral domain as electorally-viable political parties. The polling results indicate five important developments: 1) Prime Minister Pashinyan’s job approval is under water, but he is performing fairly well with non-committed voters; 2) Strong Armenia has been absorbing the Armenian Alliance votes, as it is mostly consolidating the electorate that voted for Robert Kocharyan in 2021; 3) majority of Armenian voters consider Armenia’s security situation to have improved, with non-committed voters displaying higher marks on the improved security environment; 4) Armenian society remains positive on the future economic trajectory of the country, while displaying reservations on the state of economy for the previous 12 months; and 5) regardless of vote intention, majority of Armenians think that Civic Contract will win majority. Methodology and Design The methodological approach to this first wave of polling was primarily based on the American National Election Studies (ANES), specifically the ANES 2020 Time Series Study. Widely regarded as “the gold standard” of political behavior surveys, ANES is a pre- and post-electoral survey that is administered using a mixed-mode design (telephone + internet). In most aspects, ArmES aimed to resemble its American counterpart. Both ArmES and ANES rely on a combination of panel and fresh sample design, and both are nationally representative. Minor differences include: (1) the use of mixed-mode by ANES and telephone mode by ArmES and (2) a pre- and post-electoral aspect by ANES, while ArmES is strictly pre-electoral. The first wave of the pre-election survey commenced on December 11, 2025. The telephone survey relied on a random digit dialing technique and sampled respondents in accordance with the population size and regional distribution of registered voters. Each question was theoretically grounded to understand the electoral preferences of the Armenian voter through a sociological and socio-psychological lens. The survey consisted of the following sections: (1) Direction of the Country and Positional Issues (2) Partisanship (3) Election & Candidate Evaluation (4) Sociodemographic & Socioeconomic Indicators (5) Citizen Forecasting Respondents were asked a total of 23 questions with 14 questions related to the upcoming elections. The survey also asked questions related to the respondent’s socioeconomic and sociodemographic characteristics. The survey concluded on February 22, 2026, with a sample size of 820 respondents. At the 95% confidence level, the margin of error is 3.4%. Results The job approval ratings of the Prime Minister produced three interesting results. First, the strong disapproval stands at 32%, which is exponentially higher than the “somewhat disapprove” preference at 13%, indicating that a third of the public has consolidated against the Prime Minister. Second, the collective approval rating stands at 36%, commensurate with findings in previous studies that demonstrate the Prime Minister’s consistent base of support to be precisely in the 35% range. And third, as displayed below, the non-committed, or so-called undecided voters, break even on the Prime Minister’s job performance, with a slight edge of 35.4% approving and 33.5% not approving. The 30% of the non-committed voters that make up the “refuse to answer” and “don’t know” categories are indicative of low-propensity voters, where the probability is quite high that such voters will not turnout to vote on election day. The survey asked two questions specific to the national economy, one retrospective and one prospective. The retrospective question asked respondents how do they qualify the current state of the economy when compared to the previous year, while the prospective question asked what their expectations are one year from the present. On the retrospective question, 23.2% of respondents, by a very slight margin, noted the economy was better now than it was a year ago, while 22.9% said it was much worse. Collectively, 32% held the economy had improved, 19% noted that the economy had stayed the same, and 37.9% perceived the economy as having worsened. The more nuanced observation here is the 19% who did not respond in the negative, but neither did they respond in the positive, which poses the question: is this 19% content that things have not gotten worse, or are not content with things staying the same? If the trend is toward the former, then the non-negative perception of the economy stands near 50%, but if it’s the latter, this means that this 19% will have expectations of the government to perform much better in the economic domain. To better understand how these expectations can be qualified, the responses to the prospective question below offer insight. When asked, “In your opinion, a year from now, what will be the state of Armenia’s economy,” nearly 28% responded they did not know, which is both expected and not uncommon in prospective sociotropic questions. But 21% noted the economy will be better, with 17% thinking it will be much worse. Collectively, 35.3% were of the opinion that the economy will be better in a year, while 23.2% believed it will get worse. In comparative terms, the negative perception declined from 37.9% to 23.2%, while the positive perception increased from 32% to 35.3%. Only 10.7% thought the economy will stay the same a year from the present, an 8% decrease from the retrospective response. But the most telling is the increase in uncertainty, as the “don’t know” response increased from 7.3% to 27.9%. In sum, the trajectory suggests a growing level of cautious optimism, with big portions of those who held negative perceptions transitioning to a more moderate uncertain opinion. Respondents were asked to gauge the direction in which the country was heading, with 37.9% stating that the country was headed in the wrong direction, 34.9% responding that it was heading in the right direction, and 26.5% being undecided or non-committed. The distribution of the non-committed voters was similar to the distribution of the overall sample, as 29.2% of such voters deemed the country to be moving in the wrong direction, while 25.9% viewed it as moving in the right direction. The roughly 26% of the positive direction vote clearly demonstrate these non-committed voters leaning towards the incumbent government, while the distribution of the negative response among the numerous parties remains unclear. Given the large percentage of non-committed respondents in our 2021 pre-election survey, we anticipated a similar pattern for this wave. To avoid conceptualizing the non-committed electorate as a “black box,” we constructed a series of positional questions and employed respondents’ answers as a heuristic instrument to classify and predict their prospective vote choice. Positional questions are used in developing countries — particularly in the Afrobarometer — to deduce electoral behavior. In this wave, we asked questions to qualify citizen perceptions on justice and accountability, Armenia’s security situation, and the TRIPP (Trump Road for International Peace and Prosperity) project. Respondents were presented two varying policy positions, with one position aligning with the incumbent government and the other with the opposition. The first question item asked respondents to specify their position on the following questions: The current government has or has not undertaken sufficient work to hold previous government officials responsible for illicit activities? A healthy majority of respondents, 57.4%, held the position that the current government, in the domain of justice and culpability, has not done a sufficient job of holding officials of previous governments accountable for abuse of power, corruption, and other such illegal behavior. This response indicates broad public support for more comprehensive prosecutorial actions by the government in holding the political elite of the previous regimes accountable for perceived illicit activities. The demand for accountability also remains robust with non-committed voters, as 51.5% of so-called undecided voters believe that the current government has not done a sufficient job in bringing to justice officials from previous governments, while 21.2% is of the belief that the government’s approach has been sufficient, and deductively speaking, prosecutorial actions against previous regime officials should cease. The second question item asked citizens to disclose one of two positions on the following question: Has the security situation in Armenia improved or not improved when compared to one year ago? A majority of citizens responded in the positive, viewing Armenia’s security environment to have improved when compared to the state of the country’s security landscape at the beginning of 2025. From an analytical perspective, the developments after the August 8, 2025 Washington Accords, the absence of shootings or internecine violence on the border since and prior, and the diminishing probability of war due to these and other factors, appear to have altered the public’s perception of Armenia’s security environment. The results simply demonstrate that the majority of the Armenian public feels Armenia is a safer and more secure country when compared to 12 months ago. The disaggregated results of the security question among non-committed voters also demonstrates majority approval in the collective improvement of the country’s security environment. 50.7% of so-called undecided voters hold the position that the security situation has improved, while 36.9% hold the position that the security environment has not improved. The trend among non-committed voters demonstrates deeper approval of the current government’s handling of the security landscape, as the non-improvement vote decreases when compared to the overall sample size. The third positional question asked respondents what their opinion is of TRIPP, with the two positional statements being framed as follows: TRIPP is beneficial to Armenia because it will bring economic growth and stability or TRIPP is not beneficial to Armenia because it will not bring economic growth or stability? Plurality of respondents view TRIPP as being beneficial to Armenia, while 34.4% do not view that TRIPP will bring economic benefits or stabilize Armenia’s geopolitical situation. Approximately 24% of respondents did not have a response, indicating either partial or very little knowledge of TRIPP or what it entails. The distribution of these numbers also hold constant with non-committed voters, as displayed below, as 37.2% of undecideds consider TRIPP to be beneficial for Armenia, while 29.2% do not. Among the non-committed voters, the number of respondents not having a response is at 33.6%, higher than the general population sample, indicating, again, absence of sufficient knowledge or information. To gauge both voter turnout and vote intention, respondents were asked if they would be voting for the June 7 Parliamentary elections, and if so, which party or alliance they would vote for. Approximately 82% of respondents noted their intention to vote in the upcoming elections, which means that the produced results are modeled off of the premise of an 82% voter turnout. This is consistent with a plethora of voting behavior scholarship analyzing the “turnout gap” that exists between self-reported turnout in surveys and actual turnout data. As such, respondent’s intent in voting and actual voter turnout usually differ by a margin of 10% to 30%. The 2024 U.S. presidential election, for example, had a turnout at 65%, while intended voters in most of the polls were in the 80% range. In Armenia’s case, in the 2021 Armenian Parliamentary elections, voter turnout was at 49.31%. The higher turnout percentile in the model allows for a more rigorous and conservative estimate of front-runner or dominant party results. The lower the turnout rate, the higher the electoral performance of the dominant party, and as such, a higher turnout rate in the model allows for a more cautious estimation of dominant party performance. Thus, if the model was gauging voter turnout at 70%, the dominant party’s performance would be stronger, or at 60%, it would be even more stronger, so on and so forth. With respect to vote intention, initial results, without including the directional leaning of non-committed voters, suggests plurality dominance by Civic Contract at 26.1%, yet exceedingly short of a majority, while Strong Armenia comes second with 11.9%. Prosperous Armenia Party (PAP) ranks third with 6%, while the remaining parties, including the current opposition, Armenia Alliance, do not pass the 4% threshold of entering Parliament. In the domain of electoral competition, none of the parties representing the pro-West factions, such as the Republic Party for example, meet the threshold. Similarly, the Armenia Alliance, which received 21% of the votes in the 2021 elections and has absorbed the opposition landscape, has bled most of its votes to the Strong Armenia party, an ideologically-similar pro-Russia, anti-Velvet Revolution party. The Armenia Alliance also appears to have leaked votes to the PAP, led by Robert Kocharyan’s ally Gagik Tsarukyan. Collectively, the so-called anti-Pashinyan vote has consolidated within Strong Armenia and PAP, respectively. When compared to the Armenia Alliance’s 2021 election results, the numbers from this poll suggest distributive commensurability: 11.9% Strong Armenia, 6% Prosperous Armenia Party, and 3.3% Armenia Alliance, with a total of 21.2%. Armenia Alliance received precisely 21.11% in 2021, clearly indicating how the distribution of its votes has been absorbed by Strong Armenia and PAP. The most important result in the vote intention category, however, is the 18.9% that responded “don’t know” and the 17.8% that responded “refuse to answer.” These two categories, collectively, make up the non-committed in the Armenian electorate, which stands at 36.7%. The distribution of these votes, in essence, will determine the outcome of the elections as was the case in the 2021 Parliamentary elections. By discerning the directional leaning of the non-committed/undecided voter, this survey, for the first time, introduces insights and results on the likelihood of how non-committeds/undecideds will vote. This was produced through our application of an empirically-grounded technique called additive index modeling. This approach is common in voting behavior studies, particularly among scholars who seek to analyze the political sophistication of the electorate. In our case, we created a 0-3 additive scale based on responses to the three positional items to denote the proximity of the respondent’s policy positions to the Prime Minister. A score of 3 indicates the highest level of issue proximity, while a score of 0 indicates the lowest. If we posit that the Armenian electorate is gradually casting their ballot based on issues (preliminary results from our first wave provides evidence of this), then our additive index can provide predictability into their vote calculus. The results demonstrate that 32% of non-committed voters are “more likely” to vote for Civil Contract, while 14% are “most likely” to vote for the incumbent party. The directional likelihood of this category of non-committed voters voting for the current government remains statistically significant, and as such, robust. Collectively, 46.3% of the non-committed in the electorate will be “more likely” or “most likely” to vote for the Prime Minister’s party. Proportionally, 46.3% of non-committed voters make up 14.5% of total voters, demonstrating the extreme importance of how such voters directionally lean, and the extent to which this increases the incumbent party vote. On the other hand, 43.2% of non-committed voters are “less likely” to vote for Civil Contract, but this does not indicate that they are categorically unlikely. However, in the domain of probability, the likelihood scaling threshold is specific to “more likely” or “most likely” voters, and our model sought caution in not qualifying any portion of “less likely” voters as leaning toward the incumbent party. 10.5% of non-committed voters are “least likely” to vote for Civic Contract, making up the category of undecided voters that will categorically refrain from casting their ballots for the incumbent. The findings from the additive index allow for predictive vote intention modeling, as displayed above, demonstrating what the vote for Civil Contract will look like, at the 82% turnout rate, with 36.6% of non-committed voters most or more likely to vote for them. With a conservative estimate of only the “most likely” non-committed voters intending to vote for Civil Contract, the incumbent party’s vote share increases to 30.1%. With the total number of “most likely” and “more likely” non-committed voters intending to vote for Civil Contract, the incumbent party’s vote share increases to 40.5%. Thus, statistically speaking, Civil Contract, currently, has a vote share of 40.5%; meaning, if the elections were held now, and 82% of registered voters turned out to vote, they will approximately receive 40.5% of the vote. If the voter turnout is lower, the vote share of Civil Contract will proportionally increase. The first wave of the Armenian Election Study also introduces a fairly new concept to the audience of Armenian voting behavior: citizen forecasting. Broadly defined, citizen forecasting is an approach of asking the public on who they believe will win an upcoming election. A burgeoning approach in the election forecasting research agenda, this model foregoes the debates of which forecasting model is most accurate and instead relies on a parsimonious approach of aggregating public responses to the question of who will be victorious. Interestingly, citizen forecasts have been quite accurate in predicting electoral outcomes in the United States. Our citizen forecasting question produces a surprising result. Despite being well below the majority support threshold, when asked who would win the 2026 parliamentary election, a clear majority of respondents indicated that Civil Contract would secure the majority of seats and thus forgo the process of forming a coalition government. Only 8% of respondents disclosed the belief that Civil Contract would fall short of a single-party majority. On the other hand, approximately 15% posit a majority victory by any one of the opposition parties of blocs. In all, this citizen forecast presents a challenge to the incumbent party. If a significant number of Civil Contract supporters expect a majority victory, then this assumption may influence their decision to turnout on June 7. What are we to make of these results? First, it is important to note the limitation. Surveys are a snapshot in time and the fact that the first wave was largely conducted at least 120 days prior to the election should caution many on reaching broad conclusions about electoral results on the night of June 7. As the country shifts toward campaign season and the public is increasingly primed with political information, external validity will naturally increase. Second, with such a large percentage of non-committed respondents, including positional items, can provide researchers with the ability to better understand how they would lean come election day. Future polls should not simply take this statistic as a given and instead implement proxies to better understand the political behavior of the Armenian voter. Finally, some of our results present a divided electorate. The country’s next caretaker should pay close attention to this phenomenon and address polarization throughout their tenure. Through much of post-Soviet Armenia, the electorate has largely been anti-incumbent and the country did not experience the political polarizations of the West. Interestingly, despite the heightened polarization in the United States in the lead up to the 2024 Presidential elections, nearly 7 out of 10 Americans stated the country was headed in the wrong direction. The result was a sweep of battleground states by President Trump. Our results note 4 out of 10 respondents who stated the country is going in the wrong direction, a more cautiously optimistic perspective when compared to the U.S. outcome. That being said, two-party presidential electoral systems are obviously different from proportional parliamentary systems, and the directional leaning of non-committed voters remains more acute in effecting the balance of outcome in parliamentary systems. Nonetheless, polarization remains an acute problem for the Armenian electorate, which limits the cogent development of a healthy political system, institutionalized political parties, and public debate on salient issues.