Posted: 2001
This page is categorized as:    link to Outrage Management index
Hover here for
Article SummaryThis manual on how to use risk comparisons and risk statistics was commissioned to help chemical plant managers explain air emissions to their neighbors. Chapter III on risk comparisons, especially, is still relevant. Later research hasn’t borne out all its seat-of-the-pants conclusions, but the advice at the end of the chapter about the worst risk comparisons holds firm – in my terms these comparisons fail (especially when people are outraged) because they try to compare the hazard of high-outrage and low-outrage risks. The other chapters are also useful and not really outdated, I think. The appendices are both outdated and all too likely to be misused. They’re what the client originally wanted most. Vincent Covello, Paul Slovic, and I wrote the rest of the manual to soften them.

Risk Communication, Risk Statistics,
and Risk Comparisons:
A Manual for Plant Managers

Washington, DC: Chemical Manufacturers Association, 1988

Part III
Guidelines for Providing
and Explaining Risk Comparisons

Purposes for Providing Risk Comparisons

Assume that as part of your risk communication strategy, you have decided to explain a particular risk number. And assume that as part of that explanation, you want to compare the number to some other number. The first step is to be clear about why you want to make such a comparison.

Risk comparisons help put risks into perspective. Both research and experience show that people are typically less familiar with quantitative risk data than they are, for example, with quantitative length data. Most people have a solid grasp of how long 3 feet is. But they do not have a solid grasp of how risky a lifetime risk of 0.047 is. The job of risk comparisons is to make the original risk number more meaningful by comparing it to other risks.

Most risk comparisons provided by company spokespeople are intended to be reassuring. The typical risk comparison involves a risk that the company believes is small – often so small that it believes no action to reduce it further is appropriate. The purpose of the risk comparison is to help people understand why the company believes that the risk is small, so that people can make informed decisions about whether they want to accept the risk.

Less typical, but equally important, are cases involving risks that the company believes are large. The purpose of the risk comparison is to help people understand why the company believes that the risk is large. When faced with a serious risk the company should acknowledge that the risk is serious, offer comparisons that show it is serious, and focus on what is being done to reduce it.


Problems arise in presenting risk comparisons because people in the community often have good reasons to distrust industry’s efforts to minimize the risk. For example, most people recognize the company’s interest in minimizing public perceptions of chemical risks in general and of specific chemical risks in particular. Comparisons aimed at minimizing risk irritate the public because they are seen and resented as an effort to persuade people that the risk is smaller than they thought.


With this as a background, what then can be said about how best to provide and explain minimizing risk comparisons.

number 1
Acknowledge and state the company’s stake in the issue and why you are presenting comparisons. Since you hope to convince people in the community that the risk is tolerably small, you should be as explicit as possible about acknowledging this.
number 2
You should not ask, or expect, to be trusted. Instead, you should point out why people need not put all of their trust in you. You can cite various institutions and procedural mechanisms – including government regulations, government inspections, union safety committees, community oversight committees (including the new ones established under Title III), risk assessments by independent experts (even the assessments of opponents) – to make the point that the community is not, in fact, "at the mercy of the company." As the plant manager, you should urge the community to seek out various sources of information with differing biases. The less the community feels it has to trust you, the more the community will feel it can trust you.
number 3
The key to whether a "minimizing" risk comparison will be seen as useful or as misleading is whether you seem to be prejudging the acceptability of the risk to your audience. (See the earlier discussion on risk acceptability. [This section is not yet posted.]) People will be more willing to learn from your risk comparison – even a minimizing risk comparison – if they are confident on two counts: (1) that you accept the community’s right to participate fully in decisions that affect their lives, their property, and the things they value; and (2) that you understand the appropriateness of basing decisions about the acceptability of a risk on factors other than the size of the risk.

A Categorization and Ranking System for Risk Comparison

Experience suggests that some kinds of risk comparisons are more likely than others to be perceived as an effort to preempt judgments about the acceptability of a risk. On the basis of this experience, it is possible to rank different kinds of risk comparisons in terms of their acceptability to people in the community. The highest-ranking comparisons are those that put the least strain on the trust relationship between a plant manager and the public. These comparisons tend to strike even skeptical listeners as relevant, appropriate, and helpful information. The lowest-ranking comparisons, on the other hand, are those that have no intuitively obvious claim to relevance, appropriateness, or helpfulness. Such comparisons are more likely to be seen as manipulative or misleading – that is, as efforts to preempt judgments about the acceptability of a risk. Thus, the lowest-ranking comparisons will be much more difficult to present and will require much more effort to communicate effectively. That difficulty, though, does not mean that you should never, under any circumstances, use one of the low-ranking risk comparisons. Sometimes the only way to make a point is to use one of the less credible sorts of comparisons.

The general rule of thumb is: Select from the highest-ranking risk comparisons whenever possible. When you have no choice but to use a low-ranking risk comparison, do so cautiously, being aware that the risk comparison could well backfire.

First-Rank Risk Comparisons (first choice – most acceptable)

a. Comparisons of the same risk at two different times.

  • The risk from air toxic X is 40 percent less than it was before we installed the scrubbers last October.
  • With our new procedures, by this time next year the risk will be cut in half.

b. Comparisons with a standard.

  • Exposure of plant workers to air toxic X is well below the level that the Occupational Safety and Health Administration considers safe.
  • Plant emissions of air toxic X are 10 percent of what is permitted under the old EPA standard, and slightly under the level established by the new, stricter EPA standard.

c. Comparisons with different estimates of the same risk.

  • Our best estimate of the risk is x, although you should be aware that the EPA has calculated an upper-bound or worst-case risk estimate of y.
  • Our best estimate of the risk is x on the basis of methodology Alpha and y on the basis of methodology Beta.
  • Our best estimate of the risk is x, whereas the government’s is y and the Sierra Club’s is z.

Second-Rank Risk Comparisons (second choice – less desirable)

a. Comparisons of the risk of doing something versus not doing it.

  • If we buy the newest and most advanced emission control equipment, the risk will be x, whereas if we don’t buy it, the risk will be y.

b. Comparisons of alternative solutions to the same problem.

  • The risk associated with incinerating our waste is x. The risk associated with using a landfill is y.

    Note: In using this type of comparison, be careful not to omit alternatives with lower risks than the one you are advocating.

c. Comparisons with the same risk as experienced in other places.

  • The most serious air toxic X problems have been encountered in the Denver area; our air toxic X problem is only one-fifth as serious as Denver’s.

Third-Rank Risk Comparisons (third choice – even less desirable)

a. Comparisons of average risk with peak risk at a particular time or location.

  • The risk posed by emissions of air toxic X on an average day is one-thousandth as great as the risk last Wednesday, when a valve malfunctioned.
  • The risk posed by emissions of air toxic X to the nearest home is 90 percent less than the risk at the plant gate, and the risk two miles from the plant gate is 90 percent less than the risk at the nearest home.
  • The risk posed by emissions of air toxic X to the average community resident is 90 percent less than the risk to plant workers.

b. Comparisons of the risk from one source of a particular adverse effect with the risk from all sources of that same adverse effect.

  • The risk of lung cancer posed by emissions of air toxic X is roughly three-hundredths of 1 percent of our total lung cancer risk.
  • The risk of lung cancer posed by emissions of air toxic X would increase the total number of cases of lung cancer expected in a community our size in a typical year from 500 to 500.15.

Fourth-Rank Risk Comparisons (fourth choice – marginally acceptable)

a. Comparisons of risk with cost, or of one cost/risk ratio with another cost/risk ratio.

  • To reduce the risk posed by air toxic X by half would cost y dollars.
  • Saving one life by controlling emissions of air toxic X would cost y dollars, whereas saving a life by improving particulate control would cost only z dollars.

b. Comparisons of risk with benefit.

  • The chemical product whose waste by-product is air toxic X is used by hospitals to sterilize surgical instruments and thus contributes to the saving of many lives.

    Note: Risk/benefit comparisons tend to be more acceptable when the benefits accrue to the same people. However, even that type of comparison may strike people as bribery (the community will receive x amount of tax revenue from the facility) or blackmail (the community will lose x number of jobs if you shut down our facility). The general rule of thumb is: Explain the benefits of the chemical product separately from its risks; people generally prefer to make their own risk-benefit comparisons.

c. Comparisons of occupational risks with environmental risks.

  • The community is exposed to far less air toxic X than our plant workers, and medical tests at the plant show no evidence of adverse health effects.

d. Comparisons with other risks from the same source.

  • Our problem with air toxic X is no more serious than our problem with air toxic Y (which the community has long found acceptable).

e. Comparisons with other specific causes of the same disease, illness, or injury.

  • Air toxic X produces far less lung cancer than exposure to natural background levels of geological radon.

    Note: This last example violates at least one of the distinctions that the public considers important in evaluating risks: man-made versus natural. For that reason and others, such comparisons fall into the fourth rank.

Fifth-Rank Risk Comparisons (last choice – rarely acceptable
– use with extreme caution!)

All of the types of comparisons listed above have some claim to relevance and legitimacy – a strong claim in the top ranks, a much weaker claim in the bottom ranks. In the fifth rank are all risk comparisons that have little or no claim to relevance or legitimacy. Central among these are comparisons of two or more completely unrelated risks.

Even in the fifth rank, however, distinctions can be made. For example, the more a risk comparison disregards factors that people consider important in evaluating risks, the more likely it is to be ineffective. (For a list of such factors, see Table C. 1 in Appendix C.) The most important concerns people have are: (1) catastrophic potential, (2) dread, (3) voluntariness, (4) newness, (5) familiarity, and (6) controllability.

The classic example of a comparison that violates these distinctions is to tell people at a public meeting that their risk from air toxic X is lower than the risk they took when they drove their cars to the meeting or when they enjoyed a cigarette during a break. Unless there is already a high level of trust between the speaker and the audience, this sort of comparison is almost guaranteed to provoke outrage because it seems to make the following claim:

Since the risk of emissions of air toxic X is less than that of driving or smoking, two conclusions follow: (a) the risk of emissions of air toxic X must logically be more acceptable, and (b) people who drive or smoke have surrendered their right to object to the plant’s emission of air toxic X.

This is a false argument, based on a flawed premise. To seem to be advancing such a claim is to invite resentment from your audience.

Comparing the risk from air toxic X to such things as the risk of food additives is also far from ideal. There is no special reason to believe that the risks of food additives are relevant to the risks of air toxic X. But at least the comparison does not appear to do major violence to any of the most important risk distinctions. By contrast, comparing the risk of air toxic X to the risk from driving without a seatbelt violates most of the major risk distinctions. The latter risk is voluntary, familiar, and controlled by the individual. Comparing this risk to that of emissions of air toxic X – which is likely to be perceived as involuntary, unfamiliar, and beyond the citizen’s control – is bound to infuriate the audience.

A comparison that is just as bad (if not worse) is to relate the risk of air toxic X to an unfamiliar risk from a familiar activity, such as the risk from ingesting the aflatoxin in peanut butter. At least people recognize that driving without a seatbelt is genuinely risky, though they see little analogy between that and plant emissions. People do not widely recognize eating peanut butter as risky, and they see even less analogy with plant emissions. A plant manager who tells an audience of mothers that air toxic X is less dangerous than the peanut butter they feed their children is likely to be seen as patronizing and not believable.

The conclusion to draw from this discussion of risk comparisons is not that risk comparisons are impossible or useless. It is just that you must take responsibility for the risk comparison, just as you must take responsibility for the plant risk data.

Warning. Throughout this discussion, it has been implied that you can count on the quality of the comparison data. Unfortunately, that is not so. A list of some of the most important data problems is provided in the appendices.

Copyright © 1988 by Chemical Manufacturers Association

For more on outrage management:    link to Outrage Management index
      Comment or Ask      Read the comments
Contact information page:    Peter M. Sandman

Website design and management provided by SnowTao Editing Services.