METHODOLOGY
What is the scope of the PANORAMATR Report?
PANORAMATR has two primary parts:
How is the PANORAMATR Report Prepared?
Public Opinion Survey Methodology:
Public opinion polling is an example of survey research as a methodological approach. In survey research, a representative sample of participants is drawn from the target population to allow for generalizations. The results from this sample are then extrapolated to the target population using statistical methods. Public opinion polling is often called a “survey” or “questionnaire study” because it uses a standardized set of items given to all participants.
Survey research is a commonly accepted method for monitoring and tracking public perceptions over time. Nonetheless, it captures public attitudes only at the moment it is conducted and does not make predictions.
Risk Analysis Methodology:
Within the scope of risk analysis, political, economic, and geopolitical developments are continuously monitored and evaluated using a combination of methodological approaches. Based on this comprehensive assessment, forward-looking projections and scenario-based estimations are developed. In addition to quantitative economic indicators and public opinion survey data, a broad range of analytical techniques, including structural policy analysis, trend analysis, content analysis, and discourse analysis, are employed. An integrated approach is used to analyze both qualitative and quantitative data sources, enabling the generation of well-informed projections.
What Is the Target Population of the PANORAMATR Public Opinion Survey, and How Is the Sample Selected?
Since PANORAMATR monitors public perceptions and voter tendencies in Türkiye, its target population consists of Turkish citizens residing in Türkiye who are 18 years of age or older and eligible to vote. Within the scope of the research, the sample is constructed based on a fully random selection principle, whereby each individual in the target population has an equal probability of being included in the study. Participants are recruited nationwide via mobile phone, without any provincial or regional restrictions.
Mobile phone numbers are produced through computer algorithms and are dialed randomly via computer-assisted random digit dialing (RDD). This method is designed to guarantee the sample’s randomness and reduce possible sampling biases.
How Is the Generalizability of the PANORAMATR Public Opinion Survey Ensured?
The main approach to ensuring that the survey results apply broadly to Türkiye’s population is through the entirely scientific and random selection of participants from the specified target group, with no restrictions. This method seeks to include every individual or group within the research scope, reducing potential bias in the sample and findings. However, differences between the sample characteristics and those of the population may still happen. To address this, post-stratification weighting procedures are applied using official data from the Turkish Statistical Institute (TURKSTAT) for the population aged 18 and over, specifically NUTS Level 1 (Statistical Regions Classification) regional population figures, as well as age, gender, and education distributions. In addition, official results published by the Supreme Election Council (YSK) for the 14 May 2023 General Election (28th Parliamentary Election) and the 28 May 2023 Presidential Runoff Election are incorporated into the weighting process.
In summary, after confirming that the sample accurately reflects the target population, analyses are conducted in a manner that enables valid generalizations.
How Should the Findings of a Public Opinion Survey Be Interpreted? What Does the Margin of Error Mean?
The proportions or percentages obtained in a public opinion survey reflect the distribution of responses within a particular group at the time the survey was conducted. It is important to note that if major social or political events happen after the survey fieldwork, public perceptions or voting preferences might shift accordingly. Because public opinion surveys rely on a sample from the population, their results might not perfectly reflect the actual/true population value. The margin of error indicates how close the observed sample result is likely to be to the true, but unknown, population parameter. Therefore, when analyzing results from a scientifically conducted public opinion survey, it is essential to consider both the confidence level and the margin of error.
For example, imagine that 52 percent of survey respondents report being happy. Assume further that the survey has a 95% confidence level and a margin of error of ±3 percentage points. Considering these factors, the finding should be interpreted as follows: if the same survey were repeated independently 100 times, the true population value would fall between 49 percent (52–3) and 55 percent (52+3) in about 95 of those repetitions. In 95 out of 100 instances, the true population value would fall within the interval from 49 to 55 percent. However, in the other 5 cases, this interval would not contain the true value, leading to incorrect inferences about the population.
In summary, any reported finding should be considered in light of its margin of error. Keep in mind that this interval results in a correct inference roughly 19 out of 20 times (95 percent of cases).
Is Bias Present Beyond the Margin of Error?
A study analyzing 4,200 public opinion polls against actual election results found that surveys can show biases that go beyond the expected statistical margin of error. This bias probably results from structural issues common to most public opinion surveys, like challenges in reaching specific population groups by phone and the unreliable nature of predicting voter turnout. The study shows that, on average, public opinion polls have an extra bias of about 2 percentage points beyond the reported margin of error. [1]
Thus, while this bias is neither desirable nor easy to detect, it is important to recognize and consider the potential theoretical risk posed by this additional source of error when analyzing survey results.
[1] Shirani-Mehr, H., Rothschild, D., Goel, S., & Gelman, A. (2018). Disentangling bias and variance in election polls. Journal of the American Statistical Association, 113(522), 607–614. https://doi.org/10,1080/01621459,2018,1448823
