Five Stars for the Visit: What Online Reviews Can and Cannot Measure About Pest Control

1. Introduction: a rating written at the door

A customer asked for a review is usually asked soon after the visit, when the technician has been courteous, the price is known and the problem may already seem quieter. Whether the treatment worked is a question for later.

1.1 Pest control shares the property that matters

The customer cannot judge the result at the time of purchase, and often not for weeks.

That comparison is ours; §4 explains why we borrow evidence from medicine.

1.2 What this article argues

That reviews of a credence service measure the visit rather than the outcome, that the best-studied evidence supports that reading, and that rules against fake reviews, while valuable, do not address the problem. Sections 7, 15 and 22 are the case.

1.3 What this journal has already argued

Our article on pest control as a credence good reported laboratory experiments in which allowing sellers to build a reputation had little influence, and argued that an online reviewer reports an experience which, in a credence good, does not reveal quality. That was an argument from theory and the laboratory.

1.4 What this article adds

Field evidence on what ratings track in the credence market studied most, the data problems that make ratings thin and slow to change, a point about timing that the earlier article did not make, and the regulatory response to fake reviews together with what it leaves untouched.

2. Three kinds of goods

A standard distinction.

A search good can be judged before buying, an experience good after using it, and a credence good may never be fully judged by the buyer at all. The researchers behind the chronic care study note that prior research on online reviews was mostly based on search and experience goods and typically found reviews useful.¹

2.1 Credence is about what the buyer can verify

Not about whether quality exists, but about whether the buyer is in a position to see it.

2.2 Most services are a mixture

A haircut is almost entirely experience, a car repair partly credence, and medical care largely credence, which is why the classification describes proportions rather than categories.

The examples are ours.

3. Why pest control is mostly credence

Our argument.

A customer cannot see a residual deposit, judge whether bait was placed where cockroaches forage, or tell whether the population is falling or merely quiet. Our credence good article placed the trade on the same list as medicine, repair, legal services and financial advice, and our technician training article returned to the classification to explain why customers substitute what they can observe.

3.1 With experience elements

Arrival, courtesy, explanation, cleanliness and price are experienced directly, and they are real parts of the service.

3.2 The two are mixed in every visit

Which is why a single rating blends things the customer can judge with things they cannot.

3.3 And some outcomes are never observed at all

A preventive treatment that works produces nothing to see, and a customer who sees nothing cannot tell whether it worked or whether nothing would have happened anyway.

That point is ours, and it is the one our screwworm and recurring contract articles explored.

3.4 And the outcome is ambiguous even when it arrives

The credence good article made a further point. Pests that disappear are consistent with an excellent treatment, an unnecessary one or a seasonal decline; pests that persist are consistent with poor work, a hard case, reinfestation from outside or incomplete preparation. So even a customer who waits for the outcome does not learn the quality of the work directly from it.

Section 15 returns to what that means for the timing argument.

4. Where the evidence comes from

Borrowed, with reason.

We found no study of whether online ratings of pest control firms track treatment outcomes. The credence market in which online ratings have been tested against measured quality most often is medicine, and we use that literature as the closest available evidence.

4.1 An analogy, not a measurement

§29 records that this is the article's central limitation.

4.2 Why the medical studies could be done at all

The chronic care study was possible because hospitals keep outcome records: ten years of admissions and discharges, linked to the physicians who treated each patient.¹⁵ Pest control keeps nothing comparable. There is no shared record of whether an infestation returned, which firm treated it, or what happened next.

4.3 So the missing study is missing for a reason

Testing pest control ratings against outcomes would first require someone to record the outcomes, and the absence of that record is itself a feature of a credence market.

Sections 4.2 and 4.3, apart from the description of the medical data, are our reasoning.

5. How much people rely on ratings

In a national survey of American adults, 65 per cent were aware of physician rating sites, fewer than were aware of ratings for cars, and 19 per cent considered ratings very important when choosing a physician, against 89 per cent for whether the physician accepted their insurance.¹³

About 23 per cent had used such a site in the past year.¹⁴

5.1 Among those who used them, ratings changed choices

Of past-year users, 35 per cent reported selecting a physician because of good ratings and 37 per cent had avoided one because of bad ratings.¹³

5.2 How these figures travel

A later systematic review reports the same two figures as shares of patients.³ The original survey reports them as shares of respondents who had sought ratings in the past year, a much smaller group, and we use the original denominator.¹³

A figure moving from a subgroup to a whole population as it passes between papers is a data-methods point in its own right.

5.3 Some users trust ratings highly

The review also reports a study finding that 80 per cent of users trust online physician ratings as much as personal recommendations, and that more than 30 per cent of consumers compare physicians online before choosing.³

5.4 While many others do not

Of respondents who had not sought online ratings, 43 per cent reported a lack of trust in the information on the sites.¹³

6. What ratings track

Consistently.

A systematic review of 32 studies found that online patient reviews correlated positively and consistently with measures of patient experience, both for whole organisations and for individual providers.⁴

6.1 Which is what a reviewer actually experienced

The consistency is unsurprising: a rating of an experience agrees with other measures of that experience.

6.2 At two levels

For whole organisations and for individual providers.⁴ That suggests to us the link with experience does not depend on how ratings are grouped, though the review does not put it that way.

7. What they track inconsistently

The part that matters.

The same review found that associations between OPRs and quality measures were mixed.⁴

7.1 Experience and quality diverge

A provider can be pleasant and ineffective, or brusque and excellent, and a rating captures the first dimension far better than the second.

That illustration is ours.

7.2 How later researchers summarise it

A 2023 study of reviews of cancer specialists described the review as showing that web-based patient reviews are a valid measure of patient experience, although not necessarily of care quality.¹⁹

8. The chronic care study

A direct test.

Researchers linked online reviews of physicians to their patients' clinical outcomes using a granular admission and discharge data set, and found no clear relationship between online reviews of physicians and their patients' clinical outcomes, such as readmission risk or emergency room visits.²

8.1 Stars and text alike

The researchers scored the sentiment expressed in each written review as well as using its star rating,¹⁵ and both were found equally uninformative about the actual quality of care.¹

8.2 Ten years of records

The data covered ten years of admissions and discharges of patients with chronic obstructive pulmonary disease at hospitals in North Texas, tracking each patient across multiple physicians.¹⁵

Hard outcome records rather than patient opinion are what allow the comparison to test quality rather than satisfaction.

8.3 And a reason the finding matters to employers

The lead author noted that some hospitals link physicians' compensation to the quality of patient care, which makes the validity of any quality signal consequential.¹

9. Its conclusion about credence goods

Stated by the authors.

The paper's abstract explains that chronic diseases require continuous care, monitoring and multiple treatments over long periods, so patients find it hard to judge accurately how effective a particular physician is, and it describes chronic disease care as a credence service on that basis.²

9.1 The researchers were surprised

The lead author called the result a surprise, because prior research on search and experience goods had typically found reviews useful to prospective consumers.¹

9.2 What reviews did tell patients

He recommended that consumers reduce their reliance on physician review sites, noting that reviews may provide some information, such as whether the staff is courteous, but are not necessarily reliable indicators of the quality of care.¹

9.3 Why the patients could not tell

A chronic disease patient may see several physicians over years, and social and economic factors often influence treatment outcomes, which the researchers said makes writing an accurate review even harder.¹⁵

9.4 The same confounding exists here

A pest problem's outcome depends on reinfestation from outside, the season and the household's preparation as well as on the technician, which are the confounders our credence good article listed.

Applying it to reviews is ours.

10. The split in the literature

A 2023 systematic review of the credibility of physician rating websites included 28 studies. Seven publications supported the credibility of PRWs, six publications found no correlation with alternative data sets, and fifteen reported mixed results.⁵

10.1 Mixed is the largest group

More than half the studies found ratings agreeing with some comparators and not others, which is why the review's conclusion turns on what ratings were compared with.⁵

11. What was being compared

Which explains the split.

The review grouped comparators into patient-reported experience measures, objective quality criteria and peer review, and concluded that results seem to depend to a significant extent on the selected standard of comparison.¹²

11.1 Agreement with perception, not with quality

Its authors found ratings credible when the comparison relied mainly on patients' perception, but inadequate to represent values such as the medical quality of physicians.⁵

11.2 And a direct warning about outcomes

They advised that decisions concerning aspects beyond patient perceptions, such as long-term treatment outcomes, should not be made with the support of rating website data.¹²

11.3 Which is the whole question in pest control

A customer choosing a firm is making exactly that kind of decision: one about a treatment outcome, using data that track perception.

That application is ours.

11.4 Consistent with §§6 and 7

Studies comparing ratings with experience surveys will tend to find agreement, and those comparing them with outcomes will not, which is the pattern the other review reported.⁴

That reconciliation is ours.

12. Low volume

A separate weakness.

One study found that 57 per cent of doctors had received only one to three ratings, which the review says casts doubt on the ratings' credibility, especially where anonymous rating is allowed.³

12.1 A small firm faces the same arithmetic

A handful of reviews can be moved by one unusually happy or unhappy customer.

12.2 And readers wait for volume

Previous research is described as showing that users did not trust ratings until a minimum number was available, and sought other sources until then.³

12.3 Many have none at all

Several papers found that more than half of physicians had no ratings or reviews, and argued that no meaningful decision could be made about them.³

12.4 Negatives are scarce

A co-author of the data quality review said the main hurdle affecting accuracy was the glaring absence of negative ratings.¹⁶

12.5 Ratings cluster high

An analysis of 4,999 online physician ratings found an average of 3.84 out of 5 on five-point sites and 77 out of 100 on hundred-point sites, and concluded that most patients give physicians a favourable rating.¹⁷

12.6 Partly because reviewers fear the consequences

In the national survey, 34 per cent had concerns about their identity being disclosed if they left a negative comment, and 26 per cent were concerned the physician would take action against them.¹³

12.7 And few people write at all

The ratings analysis cites earlier survey work finding that 7 per cent of people who sought information about their provider had posted a review online.¹⁷

12.8 The pest control version

A customer who expects to need the firm again has a reason of their own to soften a complaint, and a customer whose problem quietly resolved has little reason to write at all.

Both points are our reasoning and are untested.

13. Early reviews shape later ones

A dynamic.

The data quality review reports that early negative reviews beget more negative reviews.³

13.1 Which weights a small firm's first reviews

If early reviews shape later ones, the first few reviews of a new or small firm carry more weight than their number suggests, in either direction.

That extension to small firms is ours.

14. And persist after quality changes

The other direction.

The review adds that doctors with very good early reviews may go on benefiting from them even after their clinical quality has declined.³

14.1 A rating is a history

Which means it describes a provider as they were when most reviews were written, not necessarily as they are.

15. The timing problem

Our argument.

A cockroach treatment is judged by what happens over the following weeks, a bed bug treatment by whether bites resume after a month, and a rodent job by whether activity returns in the next season. A review requested as the technician leaves precedes all of that.

15.1 So even an honest review measures the wrong moment

The customer reports accurately on what they know, which is the visit.

15.2 Delay fixes one problem, not two

A review written after the outcome window can at least report the outcome. But, as §3.4 recorded from our earlier article, the outcome does not identify the quality of the work, because disappearance and persistence each have several explanations.

15.3 So there are two layers

A review at the door cannot see the outcome; a review a month later can see it but cannot fully interpret it. The first layer is a matter of timing and could be changed. The second is the credence problem proper and could not.

15.4 An example from this journal

Our article on residential tick control trials described a treatment that cut questing ticks substantially without reducing tick encounters or disease in the treated households. A customer who reviewed it after seeing fewer ticks would report something true that was not the outcome they were paying for.

Sections 15.2 to 15.4 are our reasoning.

16. What a reviewer can see

Our summary of what is observable when.

16.1 Not every job is equally credence

A wasp nest removed from a soffit either stops producing wasps within a day or does not, so a review written that week can report the outcome. A cockroach baiting programme, a bed bug treatment or a rodent exclusion job does not show its result for weeks or a season.

16.2 So reviews are more informative for some services than others

A firm's rating is an average across all of them, and a customer buying the slow-outcome service is reading a number partly made of fast-outcome jobs.

16.3 Which a buyer can partly correct for

By reading reviews that name the same service they are buying, rather than the overall average.

Sections 16.1 to 16.3 are our reasoning; we found no data on how review content is distributed across pest control services.

17. The rule against fake reviews

A regulatory response to a different problem.

The American trade regulator's final rule, effective 21 October 2024, prohibits among other things selling or purchasing fake reviews, buying positive or negative reviews, undisclosed insider reviews, a company-controlled review site posing as independent, certain review suppression practices and fake indicators of social media influence.⁷

17.1 Including machine-written reviews

The rule addresses reviews by someone who does not exist, such as AI-generated fake reviews, or by people without actual experience of the business.⁶

17.2 With penalties attached

One law firm reports that the rule allows the regulator to pursue penalties of 51,744 dollars per violation as well as consumer redress.⁸

17.3 Why a rule was needed

The regulator said case-by-case enforcement without civil penalty authority might not deter clearly deceptive review practices.⁶

17.4 What it does not cover

Another law firm notes that the prohibitions on buying fake reviews and disseminating undisclosed insider reviews do not apply to businesses that simply host consumer reviews, even if they prompt submissions or aggregate star ratings.¹⁰

18. Bought sentiment

The most common temptation.

The rule prohibits compensation or other incentives conditioned on the writing of reviews that express a particular sentiment, positive or negative.⁶

18.1 Stated or implied

The regulator clarified that the condition may be conveyed expressly or by implication, so an offer need not say the words to be caught.⁶

19. Suppression

Hiding the bad.

The rule prohibits suppressing negative reviews through tactics such as groundless legal threats, physical threats, intimidation or false public accusations.⁸

19.1 And misrepresenting what is shown

A business also may not misrepresent the review section of its own website when reviews have been suppressed because of negative sentiment or low ratings.⁸

20. Insiders

Reviews from the inside.

The rule prohibits reviews by company insiders that fail to disclose clearly the reviewer's material connection to the business, prohibits such reviews by officers or managers, and imposes requirements when officers or managers solicit reviews from their own relatives, employees or agents.⁶

20.1 A small-business version of the problem

In a small trade business, relatives, friends and early acquaintances are a natural first source of reviews, and the American rule treats solicitation from relatives and employees as something requiring disclosure rather than as neutral.⁶

The observation about small firms is ours.

21. The Canadian position

As law firms describe it.

Canada's competition regulator addresses false or misleading online reviews in its deceptive marketing practices digest, describes the problem as international, and takes note of FTC guidance in this area.⁹

21.1 And the rule may reach Canadian firms

The same firm notes that the rule may apply directly to Canadian entities advertising and selling to American residents, and expects it to have a significant influence on the Canadian regulator's view of compliance.⁹

21.2 Canadian penalties are larger

Another law firm reports that the civil provisions of the Competition Act allow administrative monetary penalties of up to the greater of 10 million dollars, 15 million for each subsequent violation, or three times the benefit derived from the deceptive conduct.¹¹

21.3 And the regulator's advice to businesses

The same firm reports that the regulator recommends a compliance program to prevent misleading reviews by employees.¹¹

22. What the rules do not reach

Our argument.

Every prohibition concerns a distorted review: fabricated, bought, hidden or written by an insider. None concerns an honest review written before the outcome was known, because there is nothing deceptive about it.

22.1 So a perfectly clean rating can still mislead

About the one thing a buyer of a credence service most needs to know.

22.2 Nor the scarcity of negatives

Suppression by the business is prohibited, but a customer who never writes a complaint, out of fear or because they did not yet know there was anything to complain about, produces the same silence without any deception.

That is our reasoning, drawing on §§12.4 to 12.6.

22.3 None of this is an argument against the rules

Removing fabricated and bought reviews improves what is left. The point is only that an honest rating system for a credence service still measures what its reviewers can see.

23. What reviews are good for

Which is real.

Whether a firm shows up, explains what it is doing, treats the home with care and charges what it quoted are genuine parts of the service, and ratings track experience consistently in the literature we read.⁴

23.1 And experience can bear on outcome

A technician who explains preparation clearly may get better cooperation, and our article on the psychological burden of infestation found preparation to be a substantial determinant of whether treatment works.

That link is our inference.

23.2 Negative reviews carry information too

A detailed complaint about a missed appointment, an unexplained charge or an unanswered callback describes things a customer can verify, and those are the attributes ratings track well.

That is our reading of §6.

24. Reading reviews of a credence service

Our suggestions, which follow from §§15 and 22. The last repeats advice from our credence good article, which reported that liability rather than reputation was the remedy with an effect in the experiments, and suggested asking whether a guarantee obliges attendance or resolution.

25. What a firm could do

Also ours.

Ask for a review after the outcome window, a month after a bed bug or cockroach treatment rather than on the day, and publish the result whatever it says. That would make ratings measure more of what customers are buying.

25.1 At a cost

A review requested after a callback or a recurrence will be lower than one requested at the door, which is why the practice is unlikely to spread on its own.

25.2 Asking everyone, not a selection

A firm that asked only the customers it expected to be satisfied would bias the sample in the same direction as suppression, whether or not a particular rule covers the practice.

That is our reasoning; we did not establish how the American rule treats selective requests.

25.3 Without tying anything to the answer

Any incentive for a review must not depend on its sentiment, which the American rule prohibits whether the condition is stated or implied.⁶

25.4 And it still would not settle quality

For the reason in §15.2, a delayed review improves on a prompt one without resolving the credence problem.

26. What we take from it

Three things.

Ratings of a credence service track the visit more reliably than the outcome. Sections 6 and 7.⁴

Timing, not only honesty, is part of the problem, and delay fixes only that part. Section 15, which is our argument.

And rules against fake reviews do not address it. Section 22.⁷

27. Our own position

The disclosure.

This article concerns how customers choose between firms, and this company is one of the firms they choose between.

28. The Manitoba position

28.1 The Canadian guidance applies here

The competition regulator's guidance is national, so it governs firms in this province, though we read only a law firm's description of it.⁹

28.2 What we could not find

Any study of online ratings of pest control firms against outcomes, in Canada or elsewhere, and any data on how local customers use reviews to choose a firm.

28.3 Canada is thin even in medicine

A nationwide analysis of online hospital reviews noted that only one earlier study of online hospital reviews had included Canadian hospitals.¹⁸

29. Limitations and open questions

The evidence is borrowed from medicine. Every finding about whether ratings track quality comes from studies of physicians and hospitals, and pest control differs in cost, frequency, stakes and how soon outcomes appear.²⁴⁵

That is the most important limitation because the article's central claim, that ratings of pest control track the visit more than the outcome, is an inference from an analogous market rather than a finding about this one.

We read abstracts, summaries and news accounts. The chronic care study reaches us through its abstract and two news accounts, and the systematic reviews through extracts, so the details of their methods are not in front of us.¹²¹⁵

The survey data are old. The national survey was conducted in 2012 and the ratings analysis earlier, before the review platforms customers use today took their present form.¹⁴¹⁷

One primary source was read from a copy. The national survey letter was read from a copy hosted on a university server rather than from the journal.¹³

The Canadian position is second-hand. We read law firms' accounts of the competition regulator's guidance and penalties rather than the guidance itself.⁹¹¹

Legal summaries are not the rule. Several points about the American rule come from law firm summaries, which we cite alongside the regulator's own text.⁷⁸¹⁰

Sections 1.1, 3, 6.2, 7.1, 9.4, 11.3, 11.4, 12.8, 4.2, 4.3, 13.1, 15, 16, 20.1, 22, 23.1, 23.2, 24 and 25 are our reasoning. The classification of pest control as mostly credence, the timing argument and its two layers, the reading of the rules' reach and the suggestions for buyers and firms are ours rather than sourced positions.

30. Conclusion

Pest control mixes things a customer can judge on the day, such as punctuality, courtesy, explanation and price, with the thing they are paying for, whether the population is eliminated, which becomes clear weeks or months later and, as our credence good article argued, is ambiguous even then. Reviews are usually written close to the visit. We found no study testing online ratings of pest control firms against outcomes, so we borrowed the best-studied credence market, medicine. There, a systematic review of 32 studies found online ratings consistently associated with patient experience and inconsistently with quality measures.⁴ A study linking reviews to ten years of hospital records for a chronic disease found no clear relationship with clinical outcomes, for star ratings and written reviews alike, and its lead author said reviews may tell patients whether the staff is courteous but not the quality of care.²¹⁵¹ A review of 28 credibility studies found more mixed results than clear ones, and advised against basing decisions about long-term treatment outcomes on rating data.⁵¹²

The data are also thin and path-dependent: most physicians in one study had one to three ratings, negatives were scarce, ratings clustered high, some reviewers feared the consequences of criticism, and good early reviews persisted after quality declined.^3161713

Regulators have acted on a different problem. The American trade regulator's rule, in force since October 2024, prohibits fake and machine-written reviews, reviews bought for a sentiment, undisclosed insider reviews and several forms of suppression, and Canada's competition regulator takes note of that work and has larger penalties of its own.⁷⁶⁹¹¹ Those rules make ratings more honest. They cannot make an honest rating measure an outcome that had not happened when it was written, and they do not touch the silence of customers who never write.

Reviews remain useful for what they measure: whether a firm shows up, explains itself and treats a home with care. For the rest, a buyer is better served by reviews written weeks after a treatment, by how complaints and callbacks were handled, and by what a guarantee obliges. A firm that wanted its ratings to measure more of what customers buy could ask every customer for a review after the outcome window and publish what came back, knowing that even this fixes the timing layer and not the credence problem beneath it.

References

1. University news release on a study asking whether online physician reviews indicate clinical outcomes, published in November 2020. Institutional publicity material describing peer-reviewed work. Source for the finding that for chronic diseases online reviews do not reliably indicate quality of care as measured by readmission risk and similar outcomes; that star ratings and text reviews were equally uninformative; the lead author's statement that the result was a surprise because prior research on search and experience goods typically found reviews useful; his statement that the efficacy of reviews of search and experience goods does not extend to credence goods such as chronic disease care; his recommendation that consumers reduce reliance on review sites, which may show whether staff are courteous but are not necessarily reliable indicators of quality; and his note that some hospitals link physicians' compensation to quality of care. https://news.utdallas.edu/business-management/online-reviews-of-physicians-2020/
2. Are online reviews of physicians reliable indicators of clinical outcomes? A focus on chronic disease management, published in an information systems journal in 2020 and read as its abstract. Peer-reviewed material. Source for the finding of no clear relationship between online reviews of physicians and their patients' clinical outcomes, such as readmission risk or emergency room visits; for the statement that chronic diseases require continuous care, monitoring and multiple treatments over extended periods, making it hard for patients to assess a physician's effectiveness accurately; for its description of chronic disease care as credence in nature; and for its use of a granular admission and discharge data set. https://pubsonline.informs.org/doi/10.1287/isre.2020.0945
3. Systematic review of data quality issues with physician rating websites, published in a medical internet research journal in 2020 and read through a public biomedical archive. Peer-reviewed review material. Source for its reports that more than 30 per cent of consumers compare physicians online before choosing, that one study found 35 per cent of patients selected physicians on good ratings and 37 per cent avoided physicians with bad ratings (see reference 13 for the original denominator), and that 80 per cent of users trust online physician ratings as much as personal recommendations; and for the findings that several papers found more than half of physicians had no ratings, that one found 57 per cent had one to three ratings, that users did not trust ratings until a minimum number was available, that early negative reviews beget more negative reviews, and that doctors with good early reviews might continue to benefit even if clinical quality declined. https://pmc.ncbi.nlm.nih.gov/articles/PMC7551103/
4. Systematic review asking whether online patient reviews are associated with health care outcomes, published in a medical care research journal in 2022 and read as its abstract. Peer-reviewed review material. Source for the review of 32 studies finding consistent positive correlations between online patient reviews and patient experience at both organisation and individual provider levels, and mixed associations with quality measures. https://journals.sagepub.com/doi/abs/10.1177/10775587211014534
5. Systematic literature review of the credibility of physician rating websites, published in a health policy journal in 2023 and read as its abstract on a research-sharing platform. Peer-reviewed review material. Source for the inclusion of 28 studies; the finding that seven supported the credibility of rating websites, six found no correlation with alternative data sets and fifteen reported mixed results; and the conclusion that ratings seem credible when relying primarily on patients' perception but inadequate to represent values such as the medical quality of physicians. https://www.researchgate.net/publication/370038819_The_Credibility_of_Physician_Rating_Websites_A_Systematic_Literature_Review
6. Press release announcing the final rule banning fake reviews and testimonials, published by the American trade regulator in August 2024. Official government material. Source for the prohibition of reviews misrepresenting that they are by someone who does not exist, such as AI-generated fake reviews, or by someone without actual experience; the prohibition of compensation or incentives conditioned on reviews expressing a particular sentiment, conveyed expressly or implicitly; the rules on undisclosed insider reviews, including those by officers or managers and solicitation from relatives, employees or agents; and the regulator's statement that case-by-case enforcement without civil penalty authority might not deter clearly deceptive practices. https://www.ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commission-announces-final-rule-banning-fake-reviews-testimonials
7. Trade regulation rule on the use of consumer reviews and testimonials, published in the American federal register in August 2024. Primary regulatory document. Source for the list of prohibitions: selling or purchasing fake reviews, buying positive or negative reviews, certain undisclosed insider reviews, a company-controlled review website falsely purporting to be independent, certain review suppression practices, and fake indicators of social media influence; and for the effective date of 21 October 2024. https://www.federalregister.gov/documents/2024/08/22/2024-18519/trade-regulation-rule-on-the-use-of-consumer-reviews-and-testimonials
8. Advisory on the final rule published by a law firm in October 2024. Legal commentary material, flagged. Source for the statement that the rule allows the regulator to pursue penalties of 51,744 dollars per violation as well as consumer redress; for the prohibition of suppressing negative reviews through groundless legal threats, physical threats, intimidation or false public accusations; and for the prohibition of misrepresenting the review portion of a website when reviews are suppressed based on negative sentiment or low ratings. https://www.alston.com/en/insights/publications/2024/10/ftc-issues-final-rule-on-fake-reviews-testimonials
9. Article on the American rule and its Canadian implications, published by a law firm in September 2024. Legal commentary material, flagged. Source for the statements that Canada's competition regulator, in its deceptive marketing practices digest, described false or misleading online reviews as an international problem and takes note of American guidance in this area; and that the rule may apply directly to Canadian entities advertising and selling to American residents and is expected to have a significant influence on the Canadian regulator's view of compliance. https://gowlingwlg.com/en/insights-resources/articles/2024/ftc-announces-final-rule-to-combat-fake-reviews
10. Commentary on the final rule published by a law firm in August 2024. Legal commentary material, flagged. Source for the clarification that the prohibitions on purchasing fake reviews, disseminating fake testimonials and disseminating undisclosed employee testimonials do not apply to businesses that simply host consumer reviews, even if they prompt submissions or aggregate star ratings. https://www.freshfields.com/en/our-thinking/blogs/a-fresh-take/ftc-announces-final-rule-on-deceptive-reviews-102jh9p
11. Article on the Canadian competition regulator's warning about online reviews posted by employees, published by a law firm. Legal commentary material, flagged. Source for the statement that the civil false or misleading advertising provisions of the Competition Act allow administrative monetary penalties up to the greater of 10 million dollars, 15 million for each subsequent violation, or three times the benefit derived; and that the regulator recommends a compliance program to prevent misleading reviews by employees. https://www.nortonrosefulbright.com/en-ca/knowledge/publications/41049288/fake-reviews-competition-bureau-warns-businesses-of-stiff-penalties-associated-with-online-reviews
12. The same 2023 systematic review of physician rating website credibility, as published by the journal and read as an extract. Peer-reviewed review material. Source for the grouping of comparators into patient-reported experience measures, objective quality criteria and peer review; the statement that results seem to depend to a significant extent on the standard of comparison; and the advice that decisions concerning aspects beyond patient perceptions, such as long-term treatment outcomes, should not be made with the support of rating website data. https://www.sciencedirect.com/science/article/pii/S0168851023001069
13. Research letter on public awareness, perception and use of online physician rating sites, published in a general medical journal in 2014 and read from a copy hosted on a university server. Peer-reviewed material read from a copy, flagged. Source for awareness of physician rating sites at 65 per cent, lower than for cars; ratings considered very important by 19 per cent against 89 per cent for insurance acceptance; among those who sought ratings in the past year, 35 per cent selecting a physician on good ratings and 37 per cent avoiding one with bad ratings; 43 per cent of non-users reporting a lack of trust; and 34 per cent concerned about identity disclosure and 26 per cent about physician action if leaving negative comments. https://hai.ics.uci.edu/papers/jld130043.pdf
14. News report on the same survey. News material, flagged. Source for the survey of more than 2,000 American adults in 2012 and the finding that 23 per cent had used physician rating sites in the past year. https://www.nbcnews.com/news/amp/wbna54429549
15. Trade news report on the chronic care study, published in November 2020. Trade press material, flagged. Source for the study's use of ten years of admission and discharge data for patients with chronic obstructive pulmonary disease at hospitals in North Texas, tracking each patient across multiple physicians; its scoring of the sentiment of review text in addition to star ratings; and the statement that chronic disease patients may see several physicians over years and that social and economic factors influence outcomes, making accurate reviews more challenging. https://www.healthcarefinancenews.com/news/online-physician-reviews-are-met-distrust-among-medical-professionals
16. University news story on the data quality review, published in January 2021. Institutional publicity material. Source for a co-author's statement that the main hurdle affecting the accuracy of ratings was the glaring absence of negative ratings. https://www.buffalo.edu/ubnow/stories/2021/01/doctor-rating-websites.html
17. Analysis of 4,999 online physician ratings, published in a medical internet research journal and read through a public biomedical archive. Peer-reviewed material. Source for average ratings of 3.84 out of 5 on five-point sites and 77 out of 100 on hundred-point sites, the conclusion that most patients give physicians a favourable rating, and its citation of earlier survey work finding that 7 per cent of people who sought information about their provider posted a review. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3222200/
18. Nationwide cross-sectional analysis of online hospital patient reviews, read through a public biomedical archive. Peer-reviewed material. Source for the statement that only one earlier study of online hospital reviews had included Canadian hospitals. https://pmc.ncbi.nlm.nih.gov/articles/PMC13386053/
19. Study of the use of web-based patient reviews to assess medical oncologists, published in a medical internet research journal in 2023. Peer-reviewed material. Source for its summary of reference 4 as showing that web-based patient reviews can be considered a valid measure of patient experience, although not necessarily of care quality. https://doi.org/10.2196/39857