How did the authorities come to cite figures from the IRI survey before its publication? — Armen Khachikyan

The credibility of sociological survey results largely depends on the methodology applied. First and foremost, the way the sample is designed and how representative it is are of critical importance. The socio-demographic composition of the sample should closely reflect Armenia’s actual demographic profile based on data from the Statistical Committee of the Republic of Armenia. If there are deviations in the sample structure—for example, in gender, education level, place of residence, or age-group distribution—these may systematically affect the results, said Armen Khachikyan, a member of International Sociological Association (ISA), the Research Committee on Political Sociology, the European Sociological Association, and Director of ARAR Foundation, in an interview with Aysor.am.

“Another major issue in surveys is non-response bias. Certain social or political groups often participate less in surveys, which may result in a sample that does not fully reflect actual political attitudes.”

According to Khachikyan, the second important factor is the structure of the questionnaire. In sociological methodology, it is well established that the sequence of questions, the content of preceding questions, and subtle differences in wording can all influence responses. The so-called question order effect and framing effect may alter response distributions even when the same question is asked.

A third important component concerns data analysis and weighting models. Different analytical approaches or weighting mechanisms may significantly influence the final presentation of the data, particularly when the sample structure contains imbalances.

“If we look at the latest survey by International Republican Institute, the fact that Civil Contract remains the leading political force is also reflected in several other studies and in broader analyses of Armenia’s political landscape.

However, according to various sources, the gap with the second-largest political force is most likely not as wide as indicated, while support for other political actors is probably not as low as presented in this survey.

One notable detail may point to this. When respondents were asked whom they voted for in the 2021 elections, the distribution of answers was as follows:

47% — Civil Contract
8% — Armenia Alliance
2% — Aram Sargsyan’s Republic Party

Yet the actual election results were:

Civil Contract — 54%
Armenia Alliance — 21%
I Have Honor Alliance — 5.2%
Prosperous Armenia — around 4%

This may suggest certain distortions either in sample composition or in response distribution—issues that are commonly regarded in sociological research as indicators of representativeness problems.

A similar pattern can also be observed in questions related to media consumption.”

The sociologist also emphasized that when respondents were asked which media outlet they considered most trustworthy for political information, 30% named Public Television Company of Armenia (H1), while among respondents aged 18–35, that figure stood at 25%. However, numerous studies conducted in Armenia suggest that a significant share of young people do not consider H1 their primary source of political information, which may again indicate certain distortions in sample composition.

At the same time, it should be noted that the fieldwork for the survey was carried out by Breavis, which is generally regarded as a reliable professional organization. The methodological design and analytical component, however, were conducted by IRI.

“In this context, another important question arises from the perspective of public trust: how did representatives of the authorities appear to be confidently citing figures from this survey in their speeches even before its official publication?

Overall, these observations may point to certain methodological issues and raise broader questions about public confidence in the survey’s results. This, however, does not mean the survey should be dismissed entirely; it is more appropriately viewed as one possible measure of political sentiment, to be considered alongside other independent research and long-term data trends.