Last Updated on October 13, 2026 by Jeff Tomas
If Dusit Poll and NIDA Poll report different political results, the gap may reflect how each survey was designed, not a simple disagreement about public opinion. A poll’s sponsor, question wording, sample, field dates, and geographic scope all shape what its findings can show.
Suan Dusit Poll publishes detailed general methods, including mixed in-person and online interviews, while NIDA Poll describes telephone surveys based on probability sampling. Those practices don’t make either poll automatically more accurate, and individual surveys can differ from an organization’s usual approach. To compare their findings fairly, start with the details each poll reports, then consider examples of NIDA and Suan Dusit survey results on party support.
Key Takeaways
- Suan Dusit Poll generally combines field interviews and online responses, while NIDA Poll conducts surveys by telephone using its nationwide Master Sample.
- These different modes and sampling approaches can shape who responds and how people answer, so results may vary even when polls ask similar questions.
- NIDA’s official poll description explains its telephone interview approach; check each release for its sample size, dates, and target group.
- Suan Dusit’s published methods and individual survey details should be checked separately, since an organization’s usual design may not apply to every poll.
- Neither poll is automatically more accurate; compare methods and remember that local trends and late voter shifts can affect election outcomes.
How Thailand’s Dusit Poll and NIDA Poll Differ
Suan Dusit Poll is a polling unit of Suan Dusit University, while NIDA Poll is part of the National Institute of Development Administration. Both publish public-opinion surveys, including political research, but the labels alone don’t tell you what a particular result measures. The question, target group, location, and field dates matter just as much.
What each poll is designed to measure
Suan Dusit Poll covers election and referendum questions as well as broader public concerns. Its monthly Thai Political Index tracks 25 indicators, including cost of living, salaries, wages, and welfare. A recent report on the 25-indicator index illustrates how its work can measure political conditions beyond voter preference. The poll also surveys everyday issues, such as household expenses, which can be compared with findings on Thai public concerns about living costs.
NIDA Poll also publishes political and election surveys. Its subject matter shouldn’t be defined only by a few examples; each release shows what the researchers chose to ask and whom they surveyed. NIDA’s official description of its poll provides background on the organization and its survey work, but readers should still check individual releases for the exact design.
The two polls can be compared when they ask materially similar questions of similar target groups, in comparable locations and during aligned field dates. For example, two national surveys about party support are more useful side by side when both measure voting intention among eligible voters within the same period. A national political-index score and a local election survey answer different questions, even if both mention politics.
Available documentation also varies by poll and method. Suan Dusit’s published materials describe its standard approach, while NIDA’s official pages outline its telephone survey process. Those descriptions don’t establish that every survey from either organization uses the same design. Check each release before attributing a difference in results to the polling organization itself.
What Dusit Poll and NIDA Poll reveal about their methods
A poll’s headline sample size or national label doesn’t tell the whole story. The design of each survey, including who was sampled, where interviews took place, and how responses were collected, determines what its results can support.
Dusit Poll’s sample sizes, coverage, and collection modes
Suan Dusit Poll’s methods page describes stratified sampling across five areas: North, Central, Northeast, South, and Bangkok and its surrounding area. It also describes field and online collection as a general approach, with the mix varying by survey.
That variation is clear in two January 2026 examples. A referendum and election survey conducted January 20 to 23 included 2,269 respondents and used both field and online collection, with regional stratification. A separate pre-election survey reported 26,621 respondents and field collection. The larger sample does not mean every Suan Dusit survey uses that size or mode; check the details for each release. The original referendum and election survey documents one specific design.
A stated nationwide scope also doesn’t provide a full list of every location surveyed. Regional strata describe how the sample was organized, but readers should look for site-level details before assuming every province or locality was included.
What the available NIDA Poll example can and cannot tell us
NIDA Poll’s Chiang Mai election survey offers a documented example of its sampling and interview process. Conducted January 9 to 14, 2026, it included 1,067 eligible voters age 18 and older. NIDA drew the sample from its Master Sample list, used multistage sampling, and interviewed respondents by phone. The Chiang Mai survey description reports these details.
However, this was a provincial survey, not a national poll. It can show how that Chiang Mai study selected and contacted respondents, but it doesn’t establish NIDA Poll’s typical national sample size, geographic design, interview mode, or publication schedule. For national comparisons, consult the methods stated in the particular NIDA release rather than treating this one example as the organization’s standard.
When comparing election findings, match the surveys’ target populations and field dates as closely as possible. Reports on Thailand’s 2026 general election polling offer broader context, but each poll’s own methods determine what its numbers measure.
Why poll results can differ even when they appear close in time
Polls published within days of each other can produce different figures because they may measure different choices among different groups of people. Before comparing percentages, check what each survey asked, who could respond, where researchers sampled, and how they collected answers.
A January 2026 example shows why direct comparisons can mislead
Suan Dusit Poll surveyed 2,586 people nationwide from January 13 to 16, 2026, using online and field interviews. Its survey asked which parties had strengths on policy issues and which party respondents would choose in a party-list vote. For that party-choice question, the People’s Party received 34.11% and Pheu Thai received 18.37%.
NIDA Poll, by contrast, surveyed 1,067 eligible voters in Chiang Mai from January 9 to 14. Its questions covered preferences for prime minister and choices for constituency and party-list MPs. The results describe Chiang Mai voters, not the same nationwide population as the Dusit survey. The field periods overlapped, but the surveys did not ask the same questions of the same people.
Several design details can shift results. Question wording and answer options can affect which preference a respondent reports. Eligibility rules also matter: a survey of eligible voters may differ from one that includes a broader group. Geographic coverage is just as important, since local opinion can differ from national opinion. Meanwhile, field dates capture a particular moment, and even a few days can bring changes in voter attention or preference.
Sample size affects the statistical precision of an estimate, while collection method can affect who responds and how comfortable people feel answering. Suan Dusit used online and field interviews; NIDA used telephone interviews in its Chiang Mai poll. A margin of error describes sampling uncertainty under a survey’s design. It doesn’t correct for a mismatch in population, question, or survey method. The American Association for Public Opinion Research’s polling standards explain why sampling error is only one part of survey quality.
Treat these January results as examples of distinct surveys, not a head-to-head test of national party support. For more context, see January 2026 polling ahead of Thailand’s election.
How to judge which Thai poll is useful for your question
A poll is useful when its sample and questions match the issue you want to understand. Before comparing Dusit Poll and NIDA Poll headlines, read the full report and check what each survey actually measured. Two percentages can look comparable while describing different people, places, or choices.
Read the survey notes before comparing headline numbers
Start with the question wording and the response options. A question about a preferred prime minister is different from one about party-list voting, even if both appear in an election poll. Then check the field dates: the release date tells you when results became public, not when respondents answered.
Use this checklist to assess either poll:
- Target population: Did researchers survey eligible voters, adults, or a narrower group?
- Geographic reach: Was the poll national, provincial, or limited to a city or region?
- Sample size and selection: How many people took part, and how did researchers choose them?
- Interview mode: Did respondents answer online, by phone, or in person?
- Weighting and uncertainty: Does the report explain adjustments to the sample and disclose a margin of error or another measure of uncertainty?
Collection mode matters because online, telephone, and in-person surveys can reach different people and shape who chooses to respond. Weighting can adjust results to better match selected population characteristics, but it can’t bring in groups missing from the sample or correct every source of bias. The AAPOR journalist’s guide to understanding polls offers useful context on interpreting survey methods and results.
The right poll depends on the question. A Chiang Mai survey can help describe opinion among its stated local population, while a nationwide survey addresses a broader group. Neither automatically answers the other’s question. Likewise, a larger sample can reduce random sampling uncertainty in some designs, but sample size alone doesn’t prove better quality. Compare the population, wording, method, and dates first, then decide whether the result fits the claim you want to make.
Frequently Asked Questions
Check the details of each release before drawing conclusions from a headline. These answers clarify what an affiliation, sample, location, or schedule can tell you about a poll.
Are Dusit Poll and NIDA Poll government agencies?
No. Suan Dusit Poll is associated with Suan Dusit University, and NIDA Poll is associated with the National Institute of Development Administration. Both conduct public-opinion research, but neither is a government polling agency. Their surveys can cover different issues, including political opinion, as shown by this NIDA poll on political trust and support.
Does a larger sample always mean a more accurate poll?
No. A larger sample can reduce random sampling uncertainty in some survey designs, but it can’t fix a sample that misses important groups or questions that respondents interpret differently. Selection, population coverage, response rates, question wording, and weighting all affect how well results reflect the people a poll intends to describe.
Can I compare a provincial poll with a nationwide poll?
Only with care, because the two surveys describe different populations. A provincial poll can answer a question about its stated local population, while a national poll needs a sample designed to represent the country. Compare their methods and scope, but don’t treat their percentages as direct competitors.
Why do poll reports list a margin of error?
A margin of error estimates sampling uncertainty under stated assumptions, often for a probability sample. It doesn’t measure every possible source of error, so it won’t account for poor coverage, low response, or unclear questions. The AAPOR survey research best practices explain why researchers should report more than sampling error when describing a survey’s limits.
Does Dusit Poll publish surveys on a fixed schedule?
Its methods information describes schedules for certain recurring products, including weekly surveys and a monthly Thai Political Index. However, those schedules don’t apply to every subject or release, and a recurring product’s timing may differ from an individual survey’s field dates. Check the specific report for when researchers collected responses and when they published the results.
