Risky Business

Try a sports metaphor, Paul Slovic urges psychology graduate students learning about risk assessment at the University of Oregon in Eugene.

There are umpires who say, “I call them as I see them,” and others who say, “I call them as they are,” he tells the students.

In his classes, Slovic, who is president of the firm Decision Research in Eugene, as well as a psychology professor, has expanded the umping metaphor first suggested by late Stanford psychologist Amos Tversky. In their everyday decisions, people are most likely to reason in a third way, says Slovic: “They ain’t nothing ’til I call them.”

Welcome to the bold new subjectivism in risk-assessment theory, an interdisciplinary branch of decision-making research that draws on psychology, political science, and economics. The emerging direction of this field is less about the mathematical deduction of risk than it is about the perception of risk. Slovic put it another way in the August 1999 Risk Analysis: “Danger is real, but the concept of risk is socially constructed.”

The science of risk assessment—formerly characterized by actuarial tables that insurance companies use to calculate premiums—is getting a whiff of postmodernism. Studies are revealing differences in the way different groups of people look at danger, raising questions about the fixed and possibly biological nature of those perceptions.

For many, the idea of subjectivity in risk-perception research can be unsettling. Isn’t there a particular number that could be assigned to, say, the odds of dying from radon exposure or from having an infelicitous encounter with a semitrailer truck?

The problem with that view, Slovic argues, is that there are multiple ways to measure the costs involved. Consider the risk of death from radon. It could be expressed, for example, as deaths per million people exposed, as years of life expectancy lost due to exposure, as deaths as a function of the concentration of radon present, or in lots of other ways.

Moreover, the way risk is measured reveals the value system of the measurer, Slovic claims. Framing a risk in terms of reduction in life expectancy, for example, values the lives of the young over those of older adults, who have less of that resource to lose. Simply measuring deaths per million equates the suffering of those who expired quickly with those who lingered painfully.

Because the way risk is defined dictates the best course of risk reduction, any definition is fraught with value judgments. Says Slovic: “Defining risk is thus an exercise in power.” Since studies repeatedly show that definitions of risk depend on people’s racial group or their gender, this conclusion intensifies the stakes in assessing risk.

Group differences

The first evidence of group differences caught researchers by surprise, says Slovic. In the early 1990s, he and his colleagues were analyzing data from a survey of perceptions of environmental health risks in the United States. “We just happened to run the data by race and gender, and [the effect] kind of leapt out at us,” he says.

They called their discovery the “white male effect.” White men rated a variety of risks, from nuclear waste to street drugs, as significantly less threatening than did white females or men and women of other races. The white men who rated the risks the lowest also scored differently from the rest of the participants on several other factors. They put more trust in experts and resisted the idea that the public should give input on decisions about risk made by government institutions.

Melissa L. Finucane, a colleague of Slovic’s at Decision Research, recently tried to reproduce the white male effect, this time sampling more broadly from nonwhite populations. In the July Health, Risk & Society, she and her colleagues found the effect first reported in 1994 still to be valid.

Her team interviewed 1,204 U.S. adults who identified themselves as white, Hispanic, black, Asian, American Indian, or multiracial. The researchers asked participants for their views on the threat to themselves and their families of 13 activities and technologies. They also considered the risk level for 27 hazards to the U.S. public as a whole. Moreover, the team presented statements expressing various sociopolitical attitudes and asked participants whether they agreed or disagreed.

Women and nonwhites provided higher risk estimates for every question about risk to self and family as well as to nearly all questions about risk to the U.S. public. In addition to their lower risk estimates, white males reported different perceptions regarding other factors, Finucane says. They were significantly more likely to disagree with the statement that they had little control over risks to their health, for example.

From the survey responses, Finucane suggests that white males may have a lower risk perception in part because they view their own social power and control over risks as high. These attitudinal differences between the groups mean the white-male effect is probably based on sociopolitical factors and not biological differences, the research team asserts.

Differing perceptions

Margo Wilson, a psychologist at McMaster University in Hamilton, Ontario, bristles at the suggestion that the data from the University of Oregon researchers eliminate biology as an agent of the differing perceptions. “I think they’ve misrepresented what a biological model might be,” she says.

With psychologist Martin Daly, Wilson has argued that young, single males may have an adaptive advantage to being blind to dangers, at least for certain types of risks in certain types of circumstances. If derring-do proves irresistible to potential mates, the payoff in reproductive success may outweigh the decrease in overall life expectancy for this group.

A young-male effect that results from men’s and women’s different sexual strategies, rather than from culture, makes sense from an evolutionary perspective, Daly and Wilson claim. Many of the risk-perception questions posed in Slovic’s and Finucane’s work, such as those having to do with nuclear technology, are simply beside the point for any evolutionary model, Wilson says.

Men and women have faced mating dilemmas that have essentially remained unchanged as long as there have been people to mate, so successful strategies have had time to manifest themselves as sex-specific, biologically embedded psychologies. Nuclear technology, on the other hand, is simply too recent for any talk of a biologically adapted response to be meaningful.

Furthermore, just what participants are responding to when they answer Finucane’s questions isn’t exactly clear, Wilson continues. For example, men and women might—for reasons that are biologically based—react differently to questions involving risk to the family. White and nonwhite males may answer the questions differently because of sociologically based disparities, such as those in education or wealth.

The real comparison, Wilson says, shouldn’t be across race and sex, but within groups closely matched in cultural factors. For example, data from Daly and Wilson’s book Homicide (1988, Walter De Gruyter) indicate that in each ethnic group and culture they studied, males kill each other at a significantly greater rate compared with females killing females. And yet, she says, women in Chicago kill other women more than men kill other men in England.

Does that say there isn’t a sex difference? Wilson asks. She contends that it merely shows that cultural variables can obscure a noncultural difference.

Immediate concern

The question of group differences in risk perception isn’t just academic. It’s also of immediate concern to policy analysts. If risk isn’t an objectively measurable quantity, and if assessments vary systematically by sex and race, whose standard should prevail when governments and industries must determine an acceptable risk level?

John D. Graham, director of the Harvard Center for Risk Analysis in Boston, says that researchers at his center have found that female scientists perceive higher risk from a number of potential hazards than male scientists do. That result confounds any attempt to reframe the debate as one pitting educated opinion against lay beliefs.

In Graham’s view, the problems raised by the white male effect can be avoided as long as the public has sufficient input into risk assessment.

In practice, says Nils-Eric Sahlin, a soft-spoken professor of philosophy at Lund University in Sweden, risk experts don’t often indulge the judgments of the public. Experts, says Sahlin, are quick to characterize nonspecialists’ risk judgments as naive. That’s wrong, he says.

This opinion—that views differ not because of naïveté but because each group accurately reports its own, very different life experiences of risk—is gaining popularity as part of the political movement known as environmental justice, says Robin Collin, a law professor at the University of Oregon in Eugene.

Collin is a supporter of the movement, which advocates an equal distribution among people of benefits and burdens from decisions affecting the environment and the use of natural resources. She claims that in any government decision about risk, the most precautionary standard should be embraced.

“If we are concerned about protecting future generations, we ought to be following the risk perceptions and judgments of women and people of color,” she says.

For Sahlin, attempting to solve policy difficulties by favoring one group—any group—isn’t the answer. The issue goes deeper than differences in gauging risk levels. Even if all groups assessed risks equally, opinions could diverge. “You and I might agree the probability of a fatal accident is .9,” he says, “but you say it’s worth taking it, and I say it’s not. Then, we have a problem.”

It’s a problem, Sahlin says, that can only be solved by providing full information about what experts know and don’t know about particular dangers. The white male effect reflects a gap in trust between people with power and those without, between the sexes, and among the races, he says. The effect can be erased only by full disclosure and information sharing, a suggestion he acknowledges is not mainstream. “Paul [Slovic] says this is a crazy idea,” Sahlin adds with a laugh.

Indeed, the dogma that the public will settle for nothing less than a risk-free society is well rooted in the risk-perception field. As early as 1981, Daniel Kahneman of Princeton University and Stanford’s Tversky demonstrated that people value a risk reduction from 1 percent to zero more highly than the equivalent reduction from 2 percent to 1 percent. The general public, risk researchers have assumed, would not take kindly to the news that risk elimination may be impossible to achieve.

In an actual test of this assumption, however, Kazuya Nakayachi of the University of Shizuoka in Yada, Japan, reported in 1998 that people’s trust in a fictitious risk-management agency wasn’t diminished when the agency stated that risk elimination is impossible, compared with when it claimed that all risk indeed could be eliminated.

Furthermore, Nakayachi reports in a paper scheduled for publication in the October Risk Analysis, although people highly valued a total removal of risk, as Tversky and Kahneman found, they put an even greater premium on a risk reduction that took the first step in combating a hazard. His results suggest that, contrary to researchers’ assumptions, people don’t irrationally respond to their fears about risk and may be amenable to honest, trust-restoring news from the agencies charged with the scientific management of risk.

Risk assessment

The question about biology’s role in the white male effect and in risk assessment in general remains open, and it will stay open for a long time, Sahlin says. In 100 years, he points out, a demographic group other than white males may have the greatest control of society’s risk factors and therefore will perceive less risk than other groups do. If the sociologists are right, he says, the white male effect is not static.

In the past, theories about risk have been prescriptive. They have assumed that people ought to behave in certain ways based on certain objective calculations made by experts. The study of risk perception, however, is descriptive. Under its framework, says Rajeev Gowda, a political scientist at the University of Oklahoma in Norman, some of what has previously been termed error in assessing risk or as differing perceptions accompanying race and sex may simply reflect people’s values in a way that hasn’t been recognized before.

From the perspective of risk science’s mathematical roots, attempting to cater to a multitude of viewpoints may be an inefficient way to set risk-based policies. But, Gowda says, “if people’s values say it’s OK to live with some inefficiency, then in a democratic setting we say ‘OK,’ and get on with it.”