Public opinion surveys: trust in institutions and perceptions of reform

1. The problem: governing by noise

Public institutions, government agencies and the organisations that work with them receive continuous signals about what the population thinks: comments, petitions, articles, statements by opinion leaders. The signals are real, but they are not representative. Those who write are not those who stay silent, and those who stay silent are the majority. A reform can appear rejected in the public space and be accepted by two thirds of the population, or the reverse. The public opinion survey is the instrument that measures what the population thinks, not what its visible part thinks, on a probability sample with a known margin of error. It is the form of social survey with the greatest public exposure and, for that reason, the strictest methodological requirements.

2. What a public opinion survey measures

• Trust in institutions: government, parliament, the judiciary, police, local authorities, church, media, international organisations, measured comparably and repeatedly.
• The population's priorities: which problems are the most important, spontaneous and aided, nationally and locally.
• Perception of the situation: the direction the country is heading, assessment of the personal and national economic situation, expectations for the coming year.
• Perception of reforms and public policies: awareness, understanding, support, felt effect, for each reform of interest.
• Satisfaction with public services: health, education, local administration, infrastructure, based on recent experience.
• Attitudes on societal issues: integration, environment, equality, migration, digitalisation.
• Information sources and trust in them, which determine through which channels the institution can communicate.

Voting intention is a separate chapter, with its own rules of measurement and publication, and this guide focuses on the survey for public policy and institutional communication, not the electoral one.

3. The sample: the population, not the active public

A national public opinion survey has as its population all resident adults, usually 18+, and the sample is a probability sample, stratified by region, urban or rural residence, gender and age, and multistage: localities, households, persons. The standard size is 1,000-1,200 respondents, with a margin of error of about ±3% in total, sufficient for reporting by macro-region and by broad age group; reporting by district or by small groups requires larger samples. The reference method is the face-to-face interview, because it is the only one that evenly covers rural areas and the older population, where opinions about institutions and reforms differ systematically from those in cities. An opinion survey conducted online or by telephone describes the connected population, and the difference from the real population is, on political topics, usually considerable.

The structure of the achieved sample is compared with official statistics and weighted. The methodological report, published together with the results, contains the planned and achieved sample, the fieldwork period, the response rate, the method and the weighting variables. Without it, the survey cannot be evaluated by anyone.

4. Measuring trust in institutions

Trust is measured with a standard question, repeated identically from one wave to the next: "How much trust do you have in…", with a four-point scale, a great deal, quite a lot, not much, none at all, or from 0 to 10, for each institution on a list, in rotated order. The value of the measurement does not lie in a single number, because the absolute level depends on political culture and on question wording, but in two comparisons: between institutions, in the same wave, and over time, for the same institution. An institution with 35% trust is in completely different situations if the average of the others is 20% or 50%, and if last year it had 45% or 25%.

Trust is decomposed analytically by source: direct experience with the institution in the last 12 months, media exposure, the respondent's political affiliation and general attitudes towards the state. This decomposition tells the institution whether low trust comes from poor services, from communication or from political polarisation, and each calls for a different response.

5. The perception of a reform: four steps

The four steps are measured across the whole population and among the groups the reform affects directly, who may be few and require a booster sample. The difference between general support and the support of those affected is often the survey's most important finding.

6. Question wording: where the result is decided

In a public opinion survey, wording is the method. The same reform obtains 55% support when described through benefits and 35% when described through costs, and both figures are "true" for the questionnaire that produced them. Minimum rules:

• Neutrality: the reform is described in short, factual terms, without adjectives, checked by people with different positions on it.
• No technical terms: "fiscal decentralisation" is replaced with what it concretely means for the respondent, or the questionnaire first asks whether the term is known.
• Balance: the answer options include support and opposition in equal measure, and allow "don't know", which is a result, not a failure.
• Order: general trust questions precede questions about specific reforms, otherwise the reform discussed colours trust in the institution carrying it out.
• The open question "why?" after support or opposition, coded thematically, is the part with the greatest value for communication.

7. Reading the results: what they say and what they do not

A public opinion survey describes a state at one moment, with a margin of error. Differences smaller than the margin are not differences. Differences between subgroups have larger margins than the total. The change from one wave to the next is significant only if it exceeds the margin of the difference. "Don't know" at 30% means that a third of the population has no formed opinion, which for a new reform is the main information, not a measurement error. And, above all, opinion is not behaviour: stated support for a measure does not mean people will comply with it, and a survey that confuses the two misleads. The guide How to read a research report details these rules.

8. Using and publishing the results

Institutions use opinion surveys in three ways. For designing communication: what must be explained, to whom, through which channels, with which messengers. For adjusting the policy: the real reasons for opposition may reveal a design problem, not a communication problem. And for monitoring: the same questions, repeated half-yearly or annually, show whether trust and support are moving and in which direction.

Publishing results has rules of its own. A published survey contains the full methodology: who commissioned it, who conducted it, the population, the sample, the method, the period, the margin of error, the exact wording of the questions reported. Selected results, published without context and without the question wording, turn research into communication and destroy trust in the instrument itself. Organisations that publish surveys fully and consistently, including when the results do not suit them, build over time something more valuable than a favourable result: the credibility of the measurement.

9. What to ask the agency

• A national probability sample, face to face, with the multistage design described.
• A complete methodological report, delivered with the results and fit for publication.
• Question wording checked for neutrality, including by people with different positions.
• Qualitative testing of the questionnaire before fieldwork.
• Trust questions identical to previous waves, for comparability.
• The perception of reforms broken down into the four steps, in total and among the groups affected.
• Thematic coding of the open questions.
• Significance tests for differences between groups and between waves, marked in the report.
• The agency's independence from the result, recorded contractually: the report is delivered in full, not selectively.

The public opinion survey is the instrument through which an institution hears the whole population, not only the part that speaks loudly. Done properly and published honestly, it improves policy, communication and trust alike. Done to confirm, it erodes all three. For the data collection method, see the CAPI and CATI page.

See also: Social and political surveys CAPI and CATI Face-to-face interviews (CAPI) How to read a research report How focus groups work Methodology FAQ

The Perspicuus research team · 14 March 2026