When a false claim about drinking water spreads, the utility is usually the first organisation expected to correct it and the last one the audience wants to hear it from. We asked whether who delivers a correction matters as much as what the correction says. Our method was a survey experiment with 1,200 respondents, testing the same correction carried by different voices. The headline finding was stark. A correction from the utility alone reduced false belief by 6 points, while the same correction carried by a doctor together with a community group reduced false belief by 19 points.
I Discrimen
The organisation accused by a rumour is the weakest witness in its own defence
A correction is judged by its carrier before it is judged by its content.
The distinction our experiment drew was between the fact of a correction and the voice that delivers it. A water utility holds the technical evidence about its own supply, yet it also holds an obvious interest in the verdict. Audiences know that, and they discount accordingly, even when every word of the correction is accurate.
You may assume that publishing the test results settles the matter. Our data says the results settle very little until a voice the audience already trusts agrees to carry them. Readers weighing a similar messenger problem in a clinical setting will recognise the pattern in how to choose a PR agency for hospitals, where institutional voices face the same discount.
Accuracy is the entry fee, and credibility is the price of being heard.
II Machina
A survey experiment with 1,200 respondents isolated the messenger as the variable
Same words, different voices, and the voices were the experiment.
Our instrument was a survey experiment with a sample of 1,200 respondents. Participants encountered a false claim about the water supply, then received a correction. The wording of the correction was held steady while the messenger changed, so that any difference in effect could be read as a difference in who was speaking rather than in what was said.
One group heard the correction from the utility alone. Another heard it from a doctor together with a community group, a pairing chosen because each half of it holds a different kind of standing: professional knowledge on one side and local membership on the other. False belief was measured after exposure, and the gap between the two arms is the finding this post reports.
Hold the message constant, and the messenger becomes measurable.
III Exemplum
The trusted pairing cut false belief by 19 points against 6 for the utility alone
The study compared two carriers of one correction, and the carriers decided the result.
In the utility only arm, the correction reduced false belief by 6 points. That is a real effect, and it would be a mistake to call it failure. A utility that says nothing leaves the false claim standing unopposed, and 6 points of movement is better than silence.
In the doctor and community group arm, the same correction reduced false belief by 19 points, more than three times the movement in the utility only arm. The study record sets a Benchmark Correction Effect median of 9 points and a top quartile of 17 points. Read against those benchmarks, the utility only correction at 6 points sat below the median, while the trusted pairing at 19 points sat above the top quartile.
| Correction carrier | Reduction in false belief (points) | Benchmark median (points) | Benchmark top quartile (points) |
|---|---|---|---|
| Utility alone | 6 | 9 | 17 |
| Doctor plus community group | 19 | 9 | 17 |
| Survey experiment sample | 1,200 respondents | Not benchmarked in the record | Not benchmarked in the record |
fons: Aerovector research record for the experiment figures and the Correction Effect benchmarks in this table.
The lesson is not that utilities should stay silent. It is that a utility correction is the floor of what is possible, and partnerships are how an organisation climbs from that floor toward the top quartile.
Six points is what authority earns on its own, and nineteen is what trust adds to it.
IV Damnum
Every week a false water claim stands, it hardens into the local common sense
An uncorrected claim does not stay a claim for long, because repetition promotes it into a fact of neighbourhood life.
The cost of a weak correction is measured in behaviour as much as belief. Households that distrust the tap buy bottled water they may not be able to afford, avoid cooking practices they once trusted, and arrive at public meetings already persuaded that the utility is hiding something. Each of those responses makes the next rumour easier to believe and the next correction harder to land.
The Debunking Handbook 2020, written by Lewandowsky and co authors, names the mechanism the continued influence effect: misinformation keeps shaping reasoning even after a correction has been issued and accepted. A correction that moves belief by only a few points leaves most of that lingering influence intact.
Regulatory duty runs alongside the reputational one. The US Environmental Protection Agency Public Notification Rule requires water systems to notify customers when drinking water rules are violated or when water may pose a risk to health, and EPA notes that systems test regularly for approximately 90 contaminants. A utility already owes its customers formal notice in a crisis. Our study suggests it also owes them a messenger they will believe, a point that generalises to the regulated trust problems discussed in how to choose a PR agency for insurance companies.
A rumour corrected weakly is a rumour rehearsed for its second run.
V Genus
Messengers fall into kinds, and each kind holds a different licence to be believed
Trust is not one asset, because a community keeps several kinds of it in separate accounts.
Our experiment separated messengers into kinds defined by the basis of their standing. The utility speaks from authority and self interest at once, which is why its correction earned the smallest movement. A doctor speaks from professional knowledge applied to individual health, a basis audiences treat as personal rather than institutional. A community group speaks from membership: its members drink the same water, send children to the same schools and have no budget riding on the answer.
Pairing the second and third kinds produced the 19 point effect, and the pairing matters. The doctor supplied the expertise and the community group supplied the independence, so the correction arrived with both a reason to believe it and a reason to trust the motive behind it. Either voice alone would have carried only half of that licence.
Member owned organisations understand this instinctively, and the dynamics in how to choose a PR agency for credit unions show the same membership trust at work in a financial setting. The licence is local, personal and earned before the crisis arrives.
You cannot rent a trusted voice at short notice and expect it to arrive with its trust intact.
VI Regio
Corrections now compete inside an information environment built for speed, not care
The rumour travels by feed, and the correction has to learn to travel the same road.
The wider environment explains why messenger choice has become decisive. The Reuters Institute Digital News Report in 2024 drew on a survey of more than 95,000 people across 47 countries and documented rising public concern about what is real and what is fake online. When audiences start from suspicion, the identity of the speaker is the first shortcut they use to decide whether to listen at all.
National guidance has reached the same conclusion as our experiment. Reporting by the American Public Health Association in 2021 on the US Surgeon General advisory on health misinformation described health professionals partnering with community organisations as a core response, with trusted local voices amplifying accurate information. The World Health Organization treats the same flood of false and misleading material as an infodemic, a parallel problem to manage alongside the underlying health issue itself.
Digital channels decide how fast both claim and correction move, which is why the channel thinking in how to choose a PR agency for health technology companies belongs in a utility plan too. A water company now publishes into the same feeds that carry the rumour, and it publishes as the least trusted voice in the thread.
In a suspicious environment, the speaker is the headline and the statement is the small print.
VII Occasio
The trusted messenger partnership cut false claim engagement by 31 percent
The partnership was the action, and engagement was where its effect showed up in the wild.
The PR action our study tested was a trusted messenger partnership: the utility stopped correcting alone and instead worked through a doctor and a community group, who carried the correction in their own names and channels. The utility supplied the evidence, the testing record and the answers to technical questions, and then let the partners speak first.
The recorded outcome was that false claim engagement fell by 31 percent. Engagement is the behaviour that spreads a rumour, the shares, replies and forwards that carry it to the next street. A fall of that size means the claim did not merely persuade fewer people in a survey setting. It travelled less.
Treat the figure with the discipline it deserves. This was one utility, one claim environment and one partnership, and a different community will hold its trust in different hands. The direction of the result, though, matches the experiment exactly: belief moved furthest where the messenger was trusted, and behaviour followed belief.
Give the evidence to the voices people already believe, and then resist the urge to take the microphone back.
VIII Structura
A correction built to travel needs a fact, a named myth and a trusted carrier
Structure is what stops a good partnership dissolving into three separate statements.
Structure the correction before anyone publishes. Lead with the fact, because the Debunking Handbook 2020 advises that corrections open with the accurate account rather than the myth. Name the false claim once, clearly labelled as false, so audiences can recognise it when they meet it again. Close by restating the fact and the action residents should take, in plain language a tired parent can follow at midnight.
Research published through PubMed Central in 2023 on correcting vaccine misinformation on social media found that audiences judged corrections heavily on the trustworthiness of the source sharing them, which is the academic version of our 19 point result. Structure and carrier are therefore one decision, not two: the best written correction still fails if it leaves the building under the wrong name.
| Correction element | What it does | Who should carry it |
|---|---|---|
| Fact first statement | Gives audiences the accurate account before the myth | Doctor partner |
| Named and labelled false claim | Lets residents recognise the rumour when it resurfaces | Community group partner |
| Evidence and testing record | Answers the technical questions behind the claim | Utility, in support |
| Action for residents | States what to do about water use today | Doctor and community group jointly |
| Follow up measurement | Checks whether false belief and engagement are falling | Utility, reported through the partners |
fons: Aerovector research record for the partnership action; Debunking Handbook 2020 for the fact first structure; PubMed Central correction research for the source trustworthiness row.
The regulated communication discipline behind this table is the same one described in how to choose a PR agency for banks: prepare the structure in calm times, because a rumour will not wait while you draft it.
A correction improvised under pressure inherits the panic of the day it was written.
IX Consilium
Recruit the messengers before the rumour, and measure belief after every correction
The prescription is a standing partnership, a tested structure and a measured result.
Start your own correction planning the way this study ended. Identify the doctors and community groups your customers already trust, and build those relationships while the water is uncontroversial. Set your expected effect against the benchmarks in our record, a median correction effect of 9 points and a top quartile of 17 points, and treat a solo utility statement as the 6 point floor it proved to be. The wider set of disciplines this work draws together is laid out in our public relations services.
Then rehearse the structure: fact first, myth named once, action stated plainly, partners speaking first and the utility supplying evidence behind them. Measure false belief and false claim engagement after each use, because the 31 percent engagement fall in this study is the kind of figure that tells you whether the partnership is real or decorative.
Partnerships built during a rumour look like recruitment, and audiences can tell the difference.
X Quaesita
Questions
What did the misinformation correction experiment measure?
We ran a survey experiment with 1,200 respondents who encountered a false claim about the water supply and then received a correction. The correction wording stayed the same while the messenger changed, and we measured how much false belief fell in each arm.
How much did the utility only correction reduce false belief?
The correction delivered by the utility alone reduced false belief by 6 points. Against the benchmarks in our record, that sits below the median correction effect of 9 points.
How much did the doctor and community group correction reduce false belief?
The same correction carried by a doctor together with a community group reduced false belief by 19 points. That sits above the top quartile benchmark of 17 points in our study record.
What are the Correction Effect benchmarks in the study record?
Our record sets a Benchmark Correction Effect median of 9 points and a top quartile of 17 points. The two arms of the experiment landed on opposite sides of both markers, which is what makes the messenger effect so plain.
What PR action did the study test?
The action was a trusted messenger partnership, in which the utility supplied evidence and technical answers while a doctor and a community group carried the correction in their own names. The utility moved from lead speaker to supporting evidence provider.
What outcome was recorded after the partnership?
False claim engagement fell by 31 percent after the trusted messenger partnership. Engagement covers the shares and replies that spread a claim, so the fall describes the rumour travelling less, not only persuading fewer people.
Should a utility stop issuing its own statements?
No, because the utility holds the testing record and carries formal notification duties when rules are violated. The finding is about order and voice: partners lead the correction, and the utility stands behind them with evidence that anyone can inspect.
How this guide was built
This post reports Aerovector's own research record for a misinformation correction experiment for a water utility: a survey experiment with 1,200 respondents, the correction effects and benchmarks stated in the text, the trusted messenger partnership as the PR action, and the false claim engagement outcome recorded after it. The study figures are Aerovector's own research record, and identities are withheld, so the organisation is described only as a water utility and no partner is named. The study covers a single utility and a single claim environment, its belief measures come from a survey experiment rather than from observed behaviour alone, and its benchmarks come from the same research record. Third party sources were used only to set context on correction science, drinking water notification duties and the wider trust environment, and each is named with its year where it appears.
Sources
- Aerovector, Misinformation correction experiment research, 2026: the first-party study reported in this post, figures from the practice research record, identities withheld.
- US Environmental Protection Agency, Public Notification Rule: drinking water public notice requirements. https://www.epa.gov/dwreginfo/public-notification-rule
- US Environmental Protection Agency, Safe Drinking Water Act overview. https://www.epa.gov/sdwa
- US Environmental Protection Agency, Basic information about lead in drinking water. https://www.epa.gov/ground-water-and-drinking-water/basic-information-about-lead-drinking-water
- US Environmental Protection Agency, Flint drinking water frequently asked questions. https://www.epa.gov/flint/flint-drinking-water-faqs
- Lewandowsky and co authors, The Debunking Handbook 2020, 2020. https://skepticalscience.com/docs/DebunkingHandbook2020.pdf
- American Public Health Association, Surgeon General advisory on health misinformation, 2021. https://www.apha.org/publications/public-health-newswire/articles/2021/07/19/surgeon-general-advisory
- National Library of Medicine, PubMed Central, Correcting vaccine misinformation on social media: effect of social correction methods, 2023. https://pmc.ncbi.nlm.nih.gov/articles/PMC10200812/
- World Health Organization, Infodemic: health topic overview. https://www.who.int/health-topics/infodemic
- Reuters Institute, Digital News Report 2024, 2024. https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2024
XI Incipe
When the rumour outruns the utility, Aerovector puts the correction in trusted hands
Our work on a claim like this starts with a Media Map of who a community actually believes, then builds a Message House the trusted partners can carry in their own words, with the utility evidence organised behind them. Coverage Reporting follows false claim engagement as the partnership works, so the next decision rests on measured movement rather than on how the statement felt in the room.
Aerovector works to a fixed chain for clients: ranking for AI search results first, media coverage under guarantee second, and leads generated by the two together. For a utility, that means the accurate account your partners carry becomes the account search tools and local reporters find first. If a false claim about your supply is circulating now, send us the claim and the voices your customers trust through the form on our contact page, and we will show you how a partnership correction would be assembled.
Aerovector gives the correction to the voices the community already believes.
Additional Details
Description: Our Research: Misinformation correction experiment for a Water utility: what the study measured, what it found, and what the finding changes for teams.
Published: 2026-10-25