Talk:Risk compensation

Initial comments
It seems to me that "The vastly greater amount of data and research now available leave no doubt that mandatory belt-wearing laws are among the most effective traffic safety measures. From 1979 to 2002 traffic deaths declined by 50% in Canada compared to only 16% in the USA. If the US declines had matched those in Canada, 200,000 fewer Americans would have been killed. A detailled, yet straightforward, calcualtion shows that 96,000 of these additional deaths were due to Canada having belt-wearing laws earlier than the US (pages 404-406 of "Traffic Safety" [2]." doesnt seem very neutral. If there really is "no doubt." better evidence than this must be produced.  If the US/Canada comparison is the strongest evidence, there is considerable doubt left in my mind. —The preceding unsigned comment was added by 68.33.169.239 (talk • contribs) 06:44, June 4, 2006. "The vastly greater amount of data and research now available leave no doubt that mandatory belt-wearing laws are among the most effective traffic safety measures. From 1979 to 2002 traffic deaths declined by 50% in Canada compared to only 16% in the USA. If the US declines had matched those in Canada, 200,000 fewer Americans would have been killed. A detailled, yet straightforward, calcualtion shows that 96,000 of these additional deaths were due to Canada having belt-wearing laws earlier than the US (pages 404-406 of 'Traffic Safety'." I do not have the source quoted, but I am familiar with the calculation.  It starts from the premise that seat belts save lives, and attributes to seat belts the differences in fatality rates between two countries.  In other words, it is an example of begging the question.  It is also stated in POV terms.  No country has, to my knowledge, yielded a provable reduction in fatalities which it provably attributable to a seat belt law. If the claims made by law proponents were accurate, seat belt laws should reduce the fatality rates dramatically over a very short period: this has never happened. The claim also misses the point - this is not about the efficacy or otherwise of seat belts, it's about their effect on driver behaviour. Just zis Guy you know? 08:26, 4 June 2006 (UTC)


 * Your comments suggest that you might not be familiar with the calculation – it incorporates the only really well established phenomenon to produce lower reductions than simple calculations from increased belt use – namely selective recruitment.


 * Belts reduce driver fatality risk in a crash by a firmly established 42%, as determined in a well-accepted study published in the peer-reviewed literature. The study is based on analysis of tens of thousands of fatalities.  The only way such effectiveness in a crash would not reduce the number driver deaths would be if wearing a belt generated a 72% increase in the risk of a severe crash.  Any such increase would be overwhelmingly apparent in any data set.  Insurance companies would have to charge those who started wearing belts an additional 72% collision premium to remain profitable!


 * In principle everything affects everything – so that wearing a belt must change driver behavior. The question “how much, and in what direction” can be determined only by evidence.  The evidence completely refutes any suggestion of a large behavioral response to being compelled to wear a belt.


 * If passing belt laws threatens pedestrians, this would show in data. In 1976 Canada’s two most populous provinces passed belt laws, and soon the whole country was covered.  From 1976 through 1983 pedestrian deaths declined by 29%.  During this period the US had no belt laws – pedestrian deaths declined by only 8%.  This enormous difference in large samples of publicly available data does not prove that drivers wearing belts dramatically reduces risks to pedestrians – but it does show how improbable the contrary claim is.


 * There is a science of traffic safety. Knowledge is generated by the methods of science – not by philosophical debate.  A few spurious effects reported more than a couple of decades ago are of little consequence – far less convincing than the large difference noted above.  Any claim that belt laws threaten pedestrians could be tested by a before to after law count of pedestrian fatalities in 49 US states, which introduced belt laws at different times.  No researcher with the skills to do this has bothered to do it because the claim that belt laws threaten pedestrians is not taken seriously.  An analysis of this type applied to the repeal of laws requiring motorcyclists to wear helmets showed that the repeal led to a 25% increase in motorcycle deaths – pretty much as estimated knowing the effectiveness of helmets in a crash (p. 299 of Traffic Safety (book)).

This article talks of "perceived risk". How does one measure risk? The perception of risk we have of cars is from things like what we hear in the newspapers and off friends about people dying in automobile accidents. Society ignores freak accidents because the percieved risk is negible. Surely drivers modify their behaviour until the perceived risk (e.g. number of deaths in the paper or heard from friends) becomes somehow acceptable. Can we add this somehow? - Chris Owen 14th June 2006

Seat belt statistics
I removed this chunk from the article because the data is disputed. See the unresolved discussions on the same data at Talk:Seat belt legislation/Archives/2023/November. -- de Facto (talk). 18:11, 9 August 2006 (UTC)

I added an article with 2007 data to give a bit of balance, but which also discusses the risk compensation effect (i.e. it tentatively suggests that knowledge of risk compensation may actually have an effect on drivers) a little.--AmericanEnglish (talk) 19:18, 9 June 2011 (UTC)

Ethology and traffic safety
The article begins by defining risk compensation as a phenomenon studied in ethology, but the bulk of the article concerns risk compensation in relation to traffic safety. Both are clearly worthy subjects as well as obviously related to each other, but perhaps they should have their own articles? At least there should be more on risk compensation in animals.--Rallette 08:57, 2 February 2007 (UTC)

Risk compensation, condoms and AIDS
I read that proponents and opponents of contraception were arguing about condom efficiency the field of AIDS prevention in terms of risk compensation. One noted critic on the issue is Edward C. Green, who argues that since people are willing take on more risk, they may disproportionally erase the benefits of condom use. This area of debate could be featured within the article along with precise references on the subject. ADM (talk) 20:48, 23 March 2009 (UTC)

Financial Crash of 2008
Financial institutions made many risky loans and investments in the years leading up to the near collapse of the global economy in 2008. Borrowers who could not afford to repay loans often had little to lose and risk. Risk compensation came in the form of huge public taxpayer bailouts of irresponsible lenders who had predicted such protections. --Mark Kaepplein (talk) 10:47, 17 September 2009 (UTC)


 * I don't believe that the phenomenon you are describing here is risk compensation. Surely it is more about situations where one person makes a gain but another person takes the loss when it all goes wrong leading to risky decisions and failuers? I assume this is a separate name in academic writing and may possibly have its own article on WP and be worth linking too from this article. PeterEastern (talk) 19:23, 31 August 2012 (UTC)


 * Indeed. I think you may be looking for Moral hazard. Richard Keatinge (talk) 21:37, 31 August 2012 (UTC)


 * Of course. Thanks. I have now added a reference to moral hazard in the lead. PeterEastern (talk) 15:10, 1 September 2012 (UTC)

Neutrality (August 2012)
The overall tone of this article fails WP:NPOV and particularly WP:FRINGE. Risk compensation, and it's synonyms Risk homeostasis and the Peltzman effect are unproven hypotheses and repeated close examination by experts in the appropriate fields have shown that this effect is illusory or nonexistent. There have been some anecdotal observations, but the exact cause of the behavior is not fully explained. The overwhelming weight of the evidence is that ABS, condoms, seat belts, etc. increase safety. --Dennis Bratland (talk) 17:47, 30 August 2012 (UTC)
 * Note that after we merge the two forks (below), the resulting material can be reworked to present a more realistic and balanced approach.--Dennis Bratland (talk) 18:25, 30 August 2012 (UTC)
 * I look forward to it. But let us note that the science is very seldom absolutely definitive in this area, and, at least in the case of the UK literature on seat belts, I've seen some extraordinary wishful thinking. The relevant literature is that which tests the relevant hypothesis, and much of what's widely put forward as relevant or even definitive does not.
 * Separately, we should perhaps include a wider definition to allow this article to outline the effects of increasing apparent risk. The main example that springs to mind is the effect of changing to driving on the right in Sweden. If I remember rightly, about a 40% drop in the number of insurance claims, which returned gradually to normal over the next six weeks.
 * -- Richard Keatinge (talk) 21:55, 31 August 2012 (UTC)


 * I suggest that a single article will be better able to express the areas where there is agreement and not. Regarding your comments about wishful thinking of the UK literature, I suggest that you will need to come up with some good references to support that claim. Again, regarding the Sweden comment, a reference would be very useful. PeterEastern (talk) 15:05, 1 September 2012 (UTC)


 * Risk and Freedom: Record of Road Safety Regulation by John Adams is the seminal work in the field and a reference for the changes in Sweden - and quite a bit else. 1985 but very little affected by later work. Richard Keatinge (talk) 13:08, 13 September 2012 (UTC)

I have removed the neutrality banner because the article does now clearly state that initiatives like seat belts, speed limits and condom supply etc do reduce risk. It also provides good references to evidence that the benefits may however be reduced to a greater or lesser degree reduced for the reasons given. If I have missed the point then please lets continue the conversation here. PeterEastern (talk) 06:29, 27 September 2012 (UTC)

Merge Risk homeostasis and Peltzman effect here
Risk homeostasis and the Peltzman effect should be merged here. They describe the same thing by slightly different names. Treating them as separate subjects introduces unnecessary confusion. --Dennis Bratland (talk) 17:49, 30 August 2012 (UTC)
 * Good call. Agreed. EricEnfermero  Howdy! 01:54, 31 August 2012 (UTC)
 * Sounds good. I did some work on this article recently and I felt at that time that the difference between this article and Risk homeostasis were minor and a bit forced. I haven't looked at the Peltzman effect. Do either of you feel able to execute this merge? PeterEastern (talk) 08:14, 31 August 2012 (UTC)
 * Excellent, the differences between these concepts are ill-defined, vague, slight, and unimportant. Go for it if you have the time. Richard Keatinge (talk) 09:31, 31 August 2012 (UTC)
 * I disagree. Peltzman had a good point, but he talked about the accident risk per unit distance driven, while risk homeostasis theory focuses on the accident risk per unit of time of exposure to traffic.  These two denominators of risk should not be confused: just as the health risk per cigarette smoked should not be confused with health risk per smoker, because smokers who chose to smoke low tar/nicotine cigarettes may decide to smoke more cigarettes and thus fail to reduce the risk to their health, as emperical studies have shown.  — Preceding unsigned comment added by GeraldJSWilde (talk • contribs) 03:54, 22 September 2012 (UTC)
 * Professor Wilde (I assume you're the author of "Target Risk"?), many thanks for your comment and for joining us here. There are indeed several related, but different concepts here. If I could rephrase my clumsy comment: I suggest that the concepts would and probably should all fit into a single high-quality encyclopaedic article, suitable to give casual readers a clear introduction to the topic and to give them pointers for further reading. One obvious reference and external link should be to "Target Risk" itself. Richard Keatinge (talk) 10:37, 22 September 2012 (UTC)
 * Would you like me to have a go at a merge. I now no sort of expert on this subject, but am reasonably experienced at WP editing and doing these sorts of merges while and trying to keep the essence of the message. Possibly I have a stab at it and then people who are more knowledgeable than me on the subject do a series of refinements to get it right. Would that be useful? PeterEastern (talk) 19:49, 23 September 2012 (UTC)
 * Go, baby, go. Regarding the Pelzman effect, I think there is some truth to it being different than risk homeostasis and risk compensation, yet the article's citations, and its claim to notability, rest primarily on sources who don't understand the distinction at all, and often say "Pelzman effect" when they really mean risk compensation or risk homeostasis. The differences are subtle and confusing, and that's all the more reason to cover all in one article, so we can clearly lay out the differences. Moreover, evidence found to support risk compensation doesn't necessarily lend support to the Pelzman effect, and this needs to be made explicit. Put another way, we redirect Non-flammable to Flammability because you can't really explore one concept without the other. --Dennis Bratland (talk) 20:12, 23 September 2012 (UTC)
 * Support per Dennis. Thanks for offering to do the work, Peter. Please carry on. Richard Keatinge (talk) 06:48, 24 September 2012 (UTC)


 * I have now completed the mechanical merge itself. The next stage is to rework the new text into a more integrated and concise article. I will contribute to this process, but would very much appreciate input from other people with specialist knowledge. PeterEastern (talk)

Integration of new content

 * Is the 'Peltzman effect' just another name for the same effect? If so, then I suggest we add a brief reference to Sam Peltzman in the lead and remove the para. PeterEastern (talk) 06:33, 27 September 2012 (UTC)


 * How should we deal with Risk Homeostasis? I am not qualified to do much with this content, but it does seem to warrant it's own section. Do others agree? If so, then what changes should be made to the content? PeterEastern (talk) 06:33, 27 September 2012 (UTC)

Neutrality (February 2013)
It appears that an IP user has restored the NPOV tag without comment. Is the POV of this article still in dispute? If so, please explain. Absent further discussion, I propose that we remove the NPOV tag. Mr. Swordfish (talk) 20:30, 13 February 2013 (UTC)
 * I have removed the banner with a comment requesting that an explanation of the perceived issues should be put on the talk page before reinstating it. PeterEastern (talk) 11:38, 14 February 2013 (UTC)


 * I'm going to restore the Neutrality tag. My issue is that the article fails to clearly cite the countervailing evidence, and it does not cite the alternative explanations for the risk compensation hypothesis. The fact that auto fatality rates per mile are 1/20 what they were 80 years ago. The numerous studies that show helmets save lives are not cited; only the few studies that show some odd or unexpected results are cited. Many activists oppose mandatory helmet laws on libertarian grounds; they cherry pick evidence against helmet safety to bolster their case. With subjects like ABS, there are several possible explanations why there were counterintuitive results in some ABS research: :
 * Brakes are not a factor in many accidents
 * ABS does not help if the driver does not brake hard enough
 * They don't prevent cars from being hit from the rear or side
 * Some drivers during the years when ABS was new to most people were not using it effectively, such as attempting to manually pump the pedal instead of holding steady pressure.
 * I turned up lots of counter evidence easily with Google. Consumer Reports and all major government safety agencies strongly support universal ABS on cars and motorcycles. Hence, risk compensation is a fringe theory lacking in widespread support.I wish I had time to fix this article section by section. Until then, it needs to be tagged as violating NPOV. --Dennis Bratland (talk) 20:22, 15 May 2013 (UTC)


 * I have adjusted the lead to give greater weight to the cautionary note regarding the size of this effect in comparison to the overall benefit of the interventions. I note that both the seat-belt section and the speed limit sections start with clear statements relating to the evident benefits of the changes. Something similar in relation to the ABS and cycle helmets sections may also be appropriate. Overall I feel that a banner is an over-reaction and would ask you to make enough changes to the article so that you feel able to remove it. It is my understanding that such a banner is normally only used when agreement on the text cannot be reached after efforts to do so. PeterEastern (talk) 03:11, 16 May 2013 (UTC)
 * I have made further changes to the article, notably to the Cycle helmet and ABS sections and have now removed the banner. Personally I do agree with your earlier concerns but feel that they have been addressed now. If you feel that it is still unbalanced then please suggest what changes are still required. PeterEastern (talk) 03:32, 16 May 2013 (UTC)


 * >Consumer Reports and all major government safety agencies strongly support universal ABS on cars and motorcycles. Hence, risk compensation is a fringe theory lacking in widespread support


 * I strongly disagree that this is a fringe theory. It is well supported by published sources, for instance the Smithsonian magazine (http://www.smithsonianmag.com/science-nature/Presence-of-Mind-Buckle-Up-And-Behave.html):


 * ...consider the possibility that some drivers have caused accidents precisely because they were wearing seat belts. This counterintuitive idea was introduced in academic circles several years ago and is broadly accepted today. The concept is that humans have an inborn tolerance for risk—meaning that as safety features are added to vehicles and roads, drivers feel less vulnerable and tend to take more chances. The feeling of greater security tempts us to be more reckless. Behavioral scientists call it "risk compensation." (emphasis mine)


 * Risk compensation theory does not say that behavioral changes entirely negate the positive effects of safety features like ABS brakes. Instead, it predicts that the net benefit is somewhat less than it would be without risk compensation.  So even in the presence of risk compensating behavior a net benefit (usually) remains. Safety experts look at this total net benefit and make there recommendation accordingly - the fact that they recommend ABS brakes does not disprove risk compensation theory.  I support removal of the banner - the article accurately reflects the published literature.  Mr. Swordfish (talk) 13:47, 17 May 2013 (UTC)
 * Agreed, although Shared space is based on the view that changing behaviour following the removal of 'safety' equipment results in less injuries, explained by Risk compensation. What is clear is that no one is claiming that the effect is simple or overwhelming or completely counterbalancing. PeterEastern (talk) 07:50, 18 May 2013 (UTC)

Disputed (again)
The last vestige of counter-point has been scrubbed form the lead. The lead and the whole article is written as if risk compensation is real and is universally accepted. The glaringly obvious fact that death rates and accident rates have fallen is the elephant in the room. This Smithsonian opinion piece cites George Will and other conservative/libertarian talking heads as evidence that it is accepted. Will is not a safety expert, he is a right wing pundit. My concerns, and those of Mr Swordfish above, have been ignored.The book Motorcycling and Leisure ISBN 0754675017 devotes a chapter to detailing what is wrong with risk compensation theory, and would be a good start.

-- Dennis Bratland (talk) 15:55, 12 August 2013 (UTC)


 * The article needs a lot of work and I'm not volunteering to do it. But the structure isn't bad. We correctly start with a definition, "effect in ethology whereby people tend to adjust their behavior in response to the perceived level of risk". Now, nobody would seriously suggest that this accounts for every phenomenon in every field to which it might be applied. And I hope that nobody is going to suggest that it accounts for no phenomena at all at any point in human history - the idea that no human being would ever adjust behaviour in response to perceived risk is simply silly, imagine stroking your pussy cat and then imagine stroking an equally-furry but hungry tiger... Going on from there, the article is structured by fields in which risk compensation has been supposed to operate, and then it has two out-of-place sections, both on road traffic, on two leading academic proponents of versions of risk compensation as a major (not all-encompassing) phenomenon in explaining some aspects of behaviour on the roads.


 * I suggest that we should get rid of the headings by academic proponent, integrating their useful material into the section on road transport. Under that heading, in particular, we need to be very clear about what claims are and are not being made and refuted. A word to the wise: the various arguments aren't always founded on a crystal-clear and accurate idea of what they're trying to (dis)prove. In fact much of the argumentation in the literature is based on a failure to address exactly what the other side is actually saying. The article you quote shows this: Person 1 finds A and thinks it's an example of risk compensation while Person 2 uses a different approach to a different dataset, finds B and claims that the whole idea of risk compensation has been disproved. We should elucidate the various confusions found in reliable sources, leaving the reader clear about the facts and the arguments, not further confused by them.
 * -- Richard Keatinge (talk) 16:57, 12 August 2013 (UTC)
 * The book I cited, Motorcycling and Leisure, demolishes risk compensation, and in doing so, the authors cite a number of quality studies which agree that risk compensation is a failed theory.The problem is that the theory is much more than the banality that behavior can be affected by perceived risk. Risk compensation makes the extraordinary claim that actual risk tends towards homeostasis in spite of safety measures. That safety devices and procedures literally don't work. Extraordinary claims require extraordinary evidence. This article plainly states that is is a fact that helmets, seatbelts, anti-lock brakes, etc, don't help at all. This extremist anti-regulation view has essentially been rejected by all but fringe groups and this article makes no note of these facts.Another glaring example: Grim Findings in Latin Crash Tests at the NYT shows the direct connection between traffic deaths and a lack of mandatory safety engineering in cars. There is as much or more traffic density in Latin America as there is in Europe and North America, so Smeed's law (another questionable theory) doesn't apply. Cars in these countries lack the numerous safety devices on cars in the Europe and the US, and we see, as expected, a horrific fatality rate. Brazil has the second highest fatality rate per mile in the world. You don't need to go through mountains of data with a fine comb doing subtle analysis to see from these facts that risk compensation theory is grossly in error.
 * --Dennis Bratland (talk) 18:43, 12 August 2013 (UTC)


 * I'm sorry not to make myself clear. The version of risk compensation that you quote is not the one given by serious academics, nor is it the one given in the article. It is a straw man set up by people who don't like the idea. And while I'm at it, the article does not say that it is a fact that helmets, seatbelts, anti-lock brakes, etc, don't help at all. If we are to achieve a reasonable article we will need to start with the banality that behaviour can be affected by perceived risk, and then to give a good account of the extent to which this is shown in reliable sources to affect measurable aspects of behaviour, or not as individual cases may demonstrate. We could include a section on rejection of the entire idea, there would be no shortage of reliable sources, and the heading could perhaps be "Risk compensation defined as the idea that all engineering improvements to road safety are useless." But the main body of the article needs to show a careful and conceptually clear approach to the academic literature. Richard Keatinge (talk) 07:26, 13 August 2013 (UTC)


 * I agree with Richard here. At the risk of repeating myself, risk compensation theory does not say that behavioral changes entirely negate the positive effects of safety features.  I've also read as much of the Motorcycling and Leisure chapter on the subject as is available at http://books.google.com/books?id=hp2kMO8YYKEC&printsec=frontcover#v=onepage&q&f=false and I don't find anything that refutes the theory.


 * I do think that the article should be clear that risk compensation theory does not assert this, and to this end I have restored the sentence in the lead deleted yesterday. It is sourced by citations later in the article. Asserting that  "risk compensation theory claims safety devices and procedures literally don't work" is a gross distortion of the theory.


 * All this being said, our job as editors is to accurately describe the theory and provide a fair assessment of how widely accepted it is, with citations to back it up. While I believe the article could stand some improvements, the "disputed" tag is overkill and I support it's speedy removal.
 * --Mr. Swordfish (talk) 19:28, 13 August 2013 (UTC)
 * I would begin by replacing the lead: "Risk compensation (also Peltzman effect or risk homeostasis) is an observed effect."


 * It has been observed (in a minority of studies) that sometimes a safety measure has less effect than predicted, but it has not been proven that risk compensation is the cause. Alternate explanations are possible. For example, it is known that ABS is only effective if the driver uses it correctly. Drivers have to be trained to press and hold the pedal firmly. An Albany, New York police department began having problems in 1992 when some of their newer patrol cars had ABS and others did not. The police Assistant Chief said, "Officers in pre-crash situations were confused about whether a car was ABS or not. They thought they had brake failure and then tried to pump them." ABS cars had a red warning sticker added to the dashboard, which led to a reduction in crashes by 1995. . For civilian drivers, far less skilled than trained police officers, you can't expect the full benefits of ABS until it is universal and you don't have drivers frequently getting into a car with ABS coming straight from driving a car without ABS and not adjusting accordingly.
 * Another explanation is the relationship between driving and social class. ABS was introduced on high-end cars, and slowly trickled down to cheaper cars. BMW was first. The studies saying "drivers with ABS tailgated more, drove faster" etc. doesn't correct for the fact that during this period ABS was only on expensive cars. Yet we have seen time and again that drivers of the most expensive cars are more aggressive and rude. See UK US (recently) and (2010) and Traffic: How We Drive by Tom Vanderbilt for even older research showing drivers of fancy cars are measurably more inconsiderate  and dangerous.
 * Again and again, we have correlation (safety feature + risky behavior) but we don't have causation. There are alternate reasons why there might sometimes be a correlation between a safety measure and behavior.So first, risk compensation should be called an "explanation" or a "hypothesis" not an "observed effect". Second, the actual refutations of the hypothesis and other counter-evidence needs to be enumerated in each section of the article. I don't know if the Google books preview of Motorcycling and Leisure is complete enough; I have a copy of the book and will use it to expand the article as soon as I possibly can. Until then, readers must at least see a warning that this article is misleading.
 * --Dennis Bratland (talk) 20:10, 13 August 2013 (UTC)
 * I don't have the book you recommend, but I will leave it to you to chase the references it cites and check that they actually substantiate relevant points. We will need something more than ill-thought-through rejection to make a good article. Risk compensation is an observed effect, though maybe not to the extent suggested by its supporters and certainly not to the extent suggested by its opponents. We don't need a misleading article, we need one that is well-researched and well-written. Please bear in mind that this is an article about a theory of how humans work in the mass. For almost any such theory, it's trivially easy to show that the theory does not explain everything and that the evidence falls short of mathematical proof. For an encyclopaedia the interesting questions are: how far and in what circumstances is risk compensation a useful concept, and what is the quality of the evidence for specific formulations? Some people may choose to fire ill-aimed broadsides at ill-understood versions of someone else's ideas, but that process is tedious and best given only passing remark by an encyclopaedia. Richard Keatinge (talk) 21:04, 13 August 2013 (UTC)

We need to be careful with our terminology. "Behavioural Adaptation" is the term long used to describe how humans change their behavior in response to changing conditions, in particular to changes in the road safety measures. That people change their behavior in response to changes in their surroundings is unremarkable and uncontroversial and there is a large body of academic works addressing behavioral adaptation.

The expression "risk compensation" is used by many academic researchers to indicate that behavioral adaptation relates to behavior change in response to changes in perceived risk. This is also a non-controversial term, although some of the theories that attempt to explain it are.

Then there is "risk homeostasis" which "...set the debate alight, rather like petrol on flames" in 1982 by asserting that "road users did not just adapt to perceptions of changing risk through compensatory behaviors, but that the process was a homeostatic one, producing overall equilibrium in safety-related outcomes." This is a controversial proposition and not universally accepted. (quotes from http://books.google.com/books?id=F8HfuHnFbCMC page 28)

The article is not really clear about these distinctions, perhaps as an artifact of the merger last fall. I'll go through the article and attempt to make these distinctions more clear; hopefully this will address the concerns raised in this thread. -- Mr. Swordfish (talk) 19:12, 14 August 2013 (UTC)


 * Brilliant, go for it. Do bear in mind that usage does not always make these distinctions clear; in due course we will need to reference definitions as well as everything else. Richard Keatinge (talk) 20:13, 14 August 2013 (UTC)

Is this article still disputed? If so, please provide concrete examples and reliable sources. If not, let's remove the banner. I'll give it a couple of days and if no response I'll remove it. Mr. Swordfish (talk) 19:03, 19 August 2013 (UTC)


 * I'm sorry I don't have more time to do this myself, but I'll do what I can. Note that if the corrections take a long time, then the article should remain tagged for maintenance. The policy WP:NPOV doesn't make exceptions for impatience. It takes as long as it takes to get it right.Note that the chapter is Motorcycling and Leisure has copious criticisms, such as "Risk homeostasis/compensation theory is not widely accepted in the road safety fraternity", and it details not just field data that contradict it, but conceptual problems, such as the failure to take into account that people are motivated not just by personal physical risk, but also financial risk, or other losses due to accidents, and that there is no identified mechanism for people to be aware of deviations from the base level risk in order to adjust the risk "thermostat". And on and on. One expert derides proponents as "abandoning the rigors of Aristotelian logic and the multiplication table." It's a fringe theory that has not gained acceptance or been adopted in public policy. --Dennis Bratland (talk) 16:22, 20 August 2013 (UTC)


 * It seems that we may be talking past one another. The article as written deals mainly with the observed effect of people modifying their behavior in response to perceived changes in risk, not the theory of how it happens.


 * The article starts off stating "Risk compensation is an observed effect..." For decades, road safety engineers have seen this effect (although not all the time in every study); the term and the idea appears in numerous studies.  This should not be controversial.


 * Some academics have tried to build models (theory) to explain and predict how risk compensation works; there are numerous "risk compensation theories" that have been proposed, the most famous of which is Wilde's "risk homeostastis" theory. From my reading, I have the impression that there is no broadly accepted theory or model, although there is a general agreement in the academic literature that road users do adapt their behavior to risk-reducing measures,  that road safety engineers try to account for this when designing safety equipment, and that this is a mainstream as opposed to fringe practice.   Thus, except for Wilde's theory, the article mainly deals with observations and examples rather than the theories. The article basically ignores the theories (except for Wilde's), and catalogs the observed effects as well as studies where the risk compensation effect was not observed.


 * Perhaps the article should more clearly delineate the difference between the observed effect and the theories. Doing so would require us to present the various competing theories and my sense is that other than "risk homeostasis" the other theories are not sufficiently notable to warrant inclusion.   As it is, except for Wilde's theory, the article mostly focuses on evidence of the effect, not its underlying causes, and attempts to present the evidence in an NOOV manner. So I don't really understand what's disputed.  Is there some particular sentence or paragraph whose "factual accuracy" is problematic? Mr. Swordfish (talk) 19:46, 20 August 2013 (UTC)

Proposed improvements
As I said above, it shouldn't be called an observed effect. It should be called a theory. Or a proposal to explain some apparent anomalies. Since 95% of the data has no anomalies, it's not even certain that there is any need for explanation. A small minority of studies have weird results. So what? It's either a flaw in the study, or a ransom sampling error, to be expected when you have chosen a 95% confidence interval.Even if there is a real phenomenon -- that sometimes a safety measure has less than expected benefit -- risk compensation has not been observed. The phenomenon could have some other explanation. I gave a detailed example with ABS. Nobody has proven that people make the choice to take bigger risks when they have a new safety device. They have merely proposed that maybe that's the explanation, and the vast majority of experts are not convinced.The article must: In lieu of the above 4 changes, a more ambitious, long term, alternative would be to merge up into a broader article about theories of risk, giving due weight to the accepted theory that try to explain behavior. In practice, it is broadly accepted that condoms, helmets, bullet-proof vests, etc, work. The theory is simply that if you mitigate risk, injuries and mishaps decrease. No need for counterintuitive explanations. --Dennis Bratland (talk) 21:09, 20 August 2013 (UTC)  I should emphasize that I appreciate the improvements Mr swordfish has made. The article is much better than it was. I just think that the specific changes above are necessary.
 * 1) State this is a theory, not an observed effect. Nobody knows for a fact that people act dangerously to maintain a constant risk.
 * 2) State at the outset it is not broadly accepted. No academic, business, or governemnt group supports it.
 * 3) State at the outset that nobody has applied risk compensation/homeostasis, by, for example, banning airbags or helmets. The application of shared space doesn't rest on changing behavior by "scaring" drivers. It works largely by forcing drivers to make independent decisions. Note the glaring Citation needed under Shared_space where some Wikipedian tried to credit risk compensation.
 * 4) The cherry-picked evidence in each sub-section, ABS, seat belts, Swedish driving, driving on the right, speed limits, bike helmets, ski helmets, children, and HIV should give full weight to the vast majority of studies that show these things are beneficial. It needs to say these are anomalous results. Alternatively, each of these example sections must be deleted until they can be fixed.

--Dennis Bratland (talk) 21:56, 20 August 2013 (UTC)


 * 1) I have no objection to calling it a "theory" and will edit the intro accordingly.  Most of the material published since 2000 is atheoretical, so this is probably the motivation behind the current wording - current researchers are more interested in describing and documenting what as opposed to why.   I'd prefer not to get into the weeds of the specific theories: DiC (Hollnagel),  qualitative model of behavorial adaptation (Brown & Noy), situational control framework (Ljung et al), unified model of driver behavior (Oppenheim et al), Task Difficulty Homeostasis TDH (Fuller et al), Risk Allostasis Theory RAT (Fuller), Risk monitor Model RMM (Vaa et al), etc.  This would be too much for an introductory article.
 * 2) I think it is fair to say that it has not been universally accepted, but one can say that about every theory in social science so I don't think it needs to be stated in the article. The claim that "it has not been broadly accepted" would need a cite to back it up.  "No academic, business, or governement group supports it" is an extraordinary claim.  Do you have any evidence to support this claim?
 * 3) Risk compensation predicts that in some cases airbags or helmets will result in a net safety benefit, just one not as large as one would predict in the absence of risk compensation.  It hardly predicts that these things are actually dangerous and should be banned. So, I fail to understand your logic here.
 * 4) Agree that many (most?) studies show that the net effect of seat belts, ABS, etc is positive.  And many of these same studies show evidence of risk compensation.  Many others do not find evidence of risk compensation and explicitly say so. As for the charge of "cherry-picking" you'll have to back that up if you want it to be taken seriously.


 * The purpose of this article is to provide a neutral synopsis of this field of study, and fairly represent the published literature. Since this is social science, one cannot expect "proof" at the mathematical level of certainty nor universal consensus among the researchers. There are a couple of survey works that give a neutral overview of the subject, and this material is what is (or should be) reflected in the article:


 * https://doclib.uhasselt.be/dspace/bitstream/1942/4002/1/behavioraladaptation.pdf
 * http://injuryprevention.bmj.com/content/6/2/82.long
 * http://books.google.com/books?id=F8HfuHnFbCMC&printsec=frontcover#v=onepage&q&f=false (the first four chapters are a review and survey of the subject - chapters 5 & 6 are not a neutral presentation)


 * I would suggest reading this material; if what is in the article misrepresents what's in the source, please call attention to it here on the talk page or edit the main article to correct it. In particular, I do not find your claim that "a small minority of studies have weird results" is reflected in the source material, nor do I find any support for your "95%" claim or that "the vast majority of experts are not convinced."
 * -- Mr. Swordfish (talk) 16:05, 21 August 2013 (UTC)


 * Again, I ask: what specific facts stated in the article are in dispute? And can you provide reliable sources that supports the opposite of those facts? Mr. Swordfish (talk) 14:50, 26 August 2013 (UTC)


 * Seeing no response after a week, I'm going to remove the banner. If we're going to restore it, please cite specific instances of facts that are in diapute.  Mr. Swordfish (talk) 15:11, 3 September 2013 (UTC)

Another note
Another note: "But as a more substantive claim that accident investigations are meaningless 'rituals of reassurance' with no effect on safety, or that people have a 'fundamental tendency to compensate for lower risks in one area by taking greater risks in another,' it is demonstrably false." Steven Pinker on Malcolm Gladwell's touting of risk compensation, in the New York Times.

This is one example of the tendency of those with the Gladwell/Freakonomics/Nurture Shock mindset to accept risk compensation uncritically out of an unreasonable love of counterintuitive results. This helps to explain why risk compensation is so popular, in spite of its near total lack of acceptance or practical application. It's fun, and talking about it makes you seem more clever than the herd, even if it's not true.

--Dennis Bratland (talk) 19:00, 30 September 2013 (UTC)


 * Which is why this is an important Wikipedia article, and one that deserves more work to get it up to the highest standard to act a rebuttal to people who use it to support fringe theories. Let's look to weave more of this into it. PeterEastern (talk) 15:37, 1 January 2014 (UTC)
 * Yes, you're right. Thanks for working on this article.Yesterday the NYT had an article on the increase in ski injuries despite more skiers wearing helmets. Several explanations were offered, such as a culture of taking greater risks: "a push toward faster, higher, pushing the limits being the norm, not the exception". Another said "because those injuries typically involve a rotational component that today's helmets cannot mitigate." And finally, at the end of the article, we have risk compensation offered as a theory, that the helmets themselves increase risk taking behavior: "But some medical professionals say that wearing a helmet could give skiers and snowboarders a false sense of security... 'There’s no 100 percent prevention of brain injury,' said Alan Weintraub, the medical director of the brain injury program at Craig Hospital in Englewood, Colo. 'Because the more the head and brain are protected, the more risks people take, the more velocities happen with those risks and the more velocities are transmitted to the skull and brain.'"This is what I think is a reasonable way to present the theory: it's not the leading theory to explain ironic injury/counter-intuitive results, but it is a theory. If leading experts actually believed that wearing helmets led to more injuries, they'd be forced to say "Stop wearing helmets!" But nobody goes that far, especially since there are 3 or 4 other good explanations of why ski injuries are up. Risk compensation is not the only, or even the most accepted, explanation for these outcomes, but it is a significant point of view.Also, I'll admit not that it looks like risk compensation is explicitly a part of the rationale for shared space -- in fact, shared space is probably the best example of risk compensation being put in to practice in the real world, and so should get more attention given that in most other cases risk compensation is merely talked about but never applied.


 * --Dennis Bratland (talk) 22:31, 1 January 2014 (UTC)
 * Interesting. Personally I am wondering if extreme sports is one area where people to compensate more than in other situations. Sky diving seems to fit that model according the Booth. Possibly skiing does as well, although it is not for us to do primary research! I have added a clear reference and quote to the shared space article linking the rational to Risk compensation. PeterEastern (talk)

Risk compensation is not Risk Homeostasis
I would like to propose that we resurrect the risk homeostasis page as it is not the same as risk compensation. Can we please start a dialogue about this? — Preceding unsigned comment added by Annettejswilde (talk • contribs) 18:55, 30 December 2013 (UTC)
 * The two were merged about 16 months ago - see discussion above. I agree that Risk Compensation and Risk Homeostasis are not synonymous and that the comments in the discussion above stating that they are synonymous were misguided.  However, I do think that Risk Homeostasis is a sub-topic of Risk Compensation and should be mentioned in this article.  I don't have strong feelings about whether it should have its own article - good arguments could be made either way.  Let's hear them.  Mr. Swordfish (talk) 21:21, 31 December 2013 (UTC)
 * Personally I would prefer to see both Risk Homeostasis and Peltzman effect being properly integrated into this article - currently they are placed as isolated sections in at the end where few people will spot them and are not even mentioned in the lead. I think that would be much more useful to the general reader than splitting them out again. PeterEastern (talk) 09:19, 1 January 2014 (UTC)
 * Very happy to discuss it, however I am increasingly of a view that these should be covered in a single comprehensive article. I have done a lot of work today on the citations to get the article up to higher general standard and feel that it could be a very good article indeed with a bit more work. PeterEastern (talk) 15:25, 1 January 2014 (UTC)
 * I agree that risk homeostasis is not the same as Risk compensation - it might be described as a stronger form of risk compensation with additional sub-hypotheses - but I do suggest that, for an encyclopedic audience, they are best presented as part of one article. Richard Keatinge (talk) 00:50, 2 January 2014 (UTC)

Speed limits
Sorry to be persistent, but the first paragraph in the speed limits section was misleading and biased. There is evidence too that speed limit changes are ineffective - this was not mention. Rather than re-hash the whole argument in this article, let's defer to the speed limit article where it is rightly discussed in detail. LatchWits (talk) 23:37, 4 January 2014 (UTC)
 * You're advocating for a WP:POVFORK. You can't designate Article A for pro-A arguments and Article B for anti-A arguments. Every article must be neutral, not just the sum of several articles. That doesn't mean all debates about speed limits belong here, but it must be stated clearly that these small numbers of counterintuitive studies that support raising speed limits to lower accident rates are overwhelmed by the vast majority of studies which tell us, quite simply, that speed kills.Similarly, you can't just have a hatnote saying, "go read this other article to see the citations". That's citing Wikipedia as a source, violating WP:CIRCULAR. Any assertion made on this page requires a citation here on this page, not in some other Wikipedia article.--Dennis Bratland (talk) 23:45, 4 January 2014 (UTC)
 * This article is dedicated to risk compensation, why complicate and confuse it with unrelated content? There is not a clear consensus amongst evidence-based studies that a change of unenforced speed limits has a corresponding effect on traffic speeds, why try to give the impression that there is? There is however, a consensus that reduced speeds tend to result in a reduced accident and casualty rates - but that is, of course, entirely independent of the prevailing speed limit. LatchWits (talk) 23:55, 4 January 2014 (UTC)
 * I have tried to clarify the distinction between traffic speeds, posted speed limits, and speed limit enforcement and the role of controlling traffic speeds in reducing road casualties in a short opening para - I hope we can agree on something like that. I would strongly advise against us opening up the full-blown folk of the 'are speed limits effective argument' here, this has been covered in exhaustive detail in the speed limits article in the past! We should simple reflect the consensus view from that article. I have removed the citation to the old 1992 research, if we need anything, lets use the World Bank report referenced from the lead of the speed limits article. PeterEastern (talk) 08:02, 5 January 2014 (UTC)
 * I whole-heartedly agree, and with the minor tweaks I've made, I think the current paragraph is sufficient introduction. LatchWits (talk) 11:28, 5 January 2014 (UTC)
 * Happy with you adjustments. Thanks. PeterEastern (talk) 15:59, 5 January 2014 (UTC)
 * I think it would be very valuable for us to read and integrate more from Vrolix (2006) - already cited - as there is a lot of very relevant material relating to road safety which is not covered appropriately in this article, including a whole section of cruise control; also on the issue of risk displacement and moral hazard which we don't mention other than as an entry in 'see also'. PeterEastern (talk) 09:09, 5 January 2014 (UTC)
 * The point, with respect to speed limits specifically, that needs to be clearly brought out is that drivers adapt their behaviour (including choice of speed), based on perceived risk (including the perceived risk of being caught breaking any speed limit) and hence a speed limit sign, in isolation, does not have much impact on perceived risk, other than the weight given by their internal moral compass to the act of breaking the law and the weight given by the speed limit sign to their perceived likelihood of causing an accident because they are exceeding that limit. Vrolix may help with this. LatchWits (talk) 11:41, 5 January 2014 (UTC)
 * Agreed, and as far as I can see from a quick skim of Vrolix the whole subject of road safety and risk compensation is more complex than this article currently conveys. I don't think what we have a wrong or bad, it just isn't definitive and top grade yet. I will see what I can do, but I am pretty busy and it would be great if others could get involved with reviewing the literature and adding content as well! PeterEastern (talk) 15:59, 5 January 2014 (UTC)

Recent improvements and suggestions for further improvement
There has been a recent flurry of edits to this page, and overall I think they have improved the article. But I also agree with Peter that "there is a lot of very relevant material... which is not covered appropriately in this article". In the spirit of improvement, here's my take on the main shortcomings of the article in its present state:

The idea of Risk Compensation has been around since at least the 1930's but didn't attract much in the way of serious study until the 1970's. In 1975, Sam Peltzman published "The Effects of Automobile Safety Regulation" (http://www.jstor.org/stable/1830396) where he stated that "... offsets (due to risk compensation) are virtually complete, so that regulation has not decreased highway deaths." According to Peltzman, regulation was at best useless and even counter-productive. This was an extraordinary claim and like a man-bites-dog story it attracted a lot of attention.

About a decade later, Gerald Wilde introduced the theory of Risk Homeostasis, which proposed that humans seek to maintain a constant level of risk, i.e. risk is adjusted homeostatically much like body temperature. This was another extraordinary man-bites-dog story and attracted a lot of attention in the popular press.

Reading the literature published in the 30 to 40 years hence, my view is that the extreme claims of Wilde and Peltzman are not in the mainstream or representative of the field as a whole. The recent literature is much less outre - a dog-bites-man story or more aptly, a dog-barks-at-man story. Consequently, it has not attracted the kind of attention in the popular press that Peltzman and Homeostasis has. I think the current article over-emphasizes these two extraordinary claims and might give the reader the wrong impression. In particular, Peltzman and Wilde's assertion that risk compensation is complete ( i.e. that it totally negates the benefits of safety measures) has not been supported by empirical studies or widely accepted by the mainstream. I don't think there is anything factually incorrect in the article right now, just that except for the examples section there's little in the article that goes beyond Wilde and Peltzman.

So, I propose we expand the Overview to represent the current mainstream expert opinion, and seek to place Peltzman's paper and Wilde's theory in the proper historical context. This requires work, so it may not happen immediately. I think the sub-sections on the Peltzman Effect and Risk Homeostasis are fine as they are, they just need a better framework. Of course, there are a bunch of CN tags that need to be resolved in the Peltzman Effect section.

I agree that Vrolix (https://doclib.uhasselt.be/dspace/bitstream/1942/4002/1/behavioraladaptation.pdf) is an excellent RS to base the expansion on. In addition, the first four chapters of Rudin-Brown & Jameson (http://books.google.com/books?id=F8HfuHnFbCMC) and Hedlund's article (http://injuryprevention.bmj.com/content/6/2/82.long) provide the kind of literature survey that is the gold standard for wikipedia reliable sources.

A second, much more minor quibble is the recently added name-dropping. "Hedlund claimed X, Williams said Y, Carson noted Z" presupposes that our readers know who Hedlund, Williams, and Carson are and I don't think we can assume that. This form of writing is fine for journals where the readership is expected to know (or take the time to find out) whom is being cited, but for a general interest article it's overkill - having the name in the cite is sufficient for wikipedia.

-- Mr. Swordfish (talk) 21:21, 7 January 2014 (UTC)


 * I agree with your analysis completely. Re Williams and Carson, I named them because I was quoting them in order to get the point across that others in the sector though that Wilde was going well over the top. However, I don't however know these people, have no reason to believe that they are worth naming and agree that they should probably not be named unless they are notable as individuals to the story in which case we need to take the time to introduce them properly. Peltman and Wilde are clearly central characters in the story and should be named and introduced - possibly in a new history section? Are you able to have a go at some of this? I think it would be really good to get more energy from more sources building this article. PeterEastern (talk) 05:00, 8 January 2014 (UTC)


 * Support your analysis Mr swordfish and Peter Eastern, go for it. I might suggest John Adams (geographer) as another central figure in this story and one who is more definitively part of mainstream thought. Richard Keatinge (talk) 08:24, 8 January 2014 (UTC)

Peltzman Effect Citation
Thanks to Peter for digging up a cite for the Peltzman Effect in the Journal of S H & E Research. Unfortunately, that paper cites this article for the definition of the Peltzman Effect, resulting in undesired circularity. And this has been true of many potential citations that I've tried to chase down - there's a blog article discussing it, and it refers the reader back here. I have found this to be most frustrating, and I'm wondering if the "sources" may be relying on Wikipedia instead of the other way around.

Certainly, there is no shortage of reliably sourced citations of Peltzman and his publications, but the term "Peltzman Effect" is conspicuously absent from the academic literature. Should we consider removing the term from the article? Mr. Swordfish (talk) 19:37, 9 January 2014 (UTC)


 * I found what may be a better reference : Safety Externalities of SUVs on Passenger Cars: An Analysis of the Peltzman Effect Using FARS Data by Maria-Paulina Diosdado-De-La-Pena http://books.google.com/books?id=lFNvi7Cj7HMC&pg=PA4#v=onepage&q&f=false


 * I'm not exactly sure how to evaluate this book's reliability (published by ProQuest in 2008), and while it contains the term "Peltzman Effect" in the title, the author concentrates on the externalities of the riskier behavior rather than the behavior itself and seems to present the "Peltzman Effect" as the embodiment of the externalities:


 * "This offsetting driver behavior is known as the Peltzman Effect, where consumers of a good (in this case SUVs) pose an externality on non-users..."


 * Not sure what to make of all this - do we have the definition wrong?
 * -- Mr. Swordfish (talk) 20:08, 9 January 2014 (UTC)


 * Apologies for the soliloquy here - going to Google Scholar instead of google provided many good cites for a definition of the Peltzman Effect, and I've used one as a cite. I think the section is on solid ground.  Mr. Swordfish (talk) 21:32, 9 January 2014 (UTC)
 * No problem at all. Personally I am much much more concerned by the articles where people (including me) can make major changes with virtually no debate about content and edits, also by ones where very few people seem to dare to touch the content in meaningful ways any more! PeterEastern (talk) 03:12, 10 January 2014 (UTC)

Vandalism
This article seems to have been vandalized yesterday, Feb 22. I'm showing in the history that a bot tried to correct it, but refreshing the page is still showing the vandalized content for me. I did not find an option to confirm the bot's work, and it doesn't seem to have gone through, so I'm bringing it to y'alls more experienced hands here. 98.209.228.228 (talk) 04:15, 24 February 2014 (UTC)
 * It was fixed. You need to force a page reload with F5 or CTRL+F5 or Command+R. If that doesn't work, empty your cache. --Dennis Bratland (talk) 04:29, 24 February 2014 (UTC)
 * Those steps are not working for me, but the problem is likely on my end. Thanks for the clarification. 98.209.228.228 (talk) 16:57, 24 February 2014 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 1 one external link on Risk compensation. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Corrected formatting/usage for http://www.skydivenet.com/fatalities/fatalities_history.html

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.—cyberbot II  Talk to my owner :Online 06:55, 1 April 2016 (UTC)