Survey method, sampling, selection, and limitations

Keymakers Consulting Ltd is a private research organisation in Bangladesh. We conduct quantitative and qualitative research in Bangladesh on behalf of various national and international organisations.

Prothom Alo has entrusted us with conducting the National Public Opinion Survey–2025 on key socio-political issues. For this survey, from 21 to 28 October this year, we collected opinions from 668 women and 674 men, totaling 1,342 respondents. In any survey, selecting the target population (the intended or defined group) is a crucial aspect.

For public opinion surveys on socio-political issues, we considered two main criteria. First, respondents had to be eligible to vote in Bangladesh, meaning they had to be over 18 years old and citizens of Bangladesh. Second, respondents had to have at least basic literacy (even if they had no formal education), so they could understand the survey questions and provide answers. Therefore, one limitation of this survey is that it does not represent the opinions of the entirely illiterate population of the country.

In a national-level survey, a sufficient number of respondents must be included so that the trends in their opinions can represent the trends in the opinions of the entire population. No survey can perfectly or completely represent the opinions of a target population. However, the number of respondents in a survey must be large enough to ensure statistical significance during sample selection.

Maintaining statistical significance increases the reliability of the survey results and reduces the likelihood of errors. To represent a large population, at least 384 respondents should be surveyed. This ensures that the survey results have over 95 per cent reliability and a margin of error below 5 per cent, making the results statistically acceptable. This is internationally recognised.

Currently, Bangladesh has more than 127.6 million registered voters. Keeping this in mind, we selected a sample of 1,342 respondents to ensure a confidence level of over 99 per cent and a margin of error below 2.7 per cent. Therefore, the results of Prothom Alo’s National Public Opinion Survey–2025 on important socio-political issues can be considered fully reliable. Even when analyzed separately, the survey results for female and male respondents are reliable, with a 99 per cent confidence level and a margin of error below 3.8 per cent.

This survey is a representative survey of Bangladesh as a whole and is not specifically representative of any electoral constituency. For the survey, we selected urban areas in Dhaka, Chattogram, Khulna, Rajshahi, and Sylhet; the upazilas of Boalmari and Madhukhali in Faridpur; Gournadi in Barishal; Chandina in Comilla; Valuka in Mymensingh; and Pirganj in Rangpur.

In the selected areas, surveyors used random selection to choose respondents to avoid any bias. Opinions were collected only from those respondents who voluntarily agreed to participate and were part of the survey’s target population.

During data collection, respondents were given sufficient time to think and express their opinions. On average, it took about 40 minutes per respondent to complete the survey. Following international research guidelines, we did not provide Prothom Alo with any information that could identify individual respondents.

Surveyors made efforts to maintain gender balance among respondents. Ultimately, 49.78 per cent of respondents were women and 51.22 per cent were men.

The survey included 50 questions in total: 13 on respondents’ income, occupation, and habits, and 37 on important social and political issues.

Respondents patiently listened to all questions and provided considered answers. We sincerely thank and appreciate the respondents for their cooperation. We hope Prothom Alo will publish this survey with neutrality and fairness, honouring both the respondents’ cooperation and our efforts in conducting the survey.

*Khandker Swanan Shahriar is the Managing Director of Keymakers Consulting Ltd.