Document MMw2qqJREBMEn599Xov6xDdeL

Ript Anaiysis, Vol. 12, No. 2, 1992 btuitive Toxicology: Expert and Lay Judgments of Chemical Risks Nancy Kraus,' Torbjorn Malmfors,2 and Paul Slavic' Received March I , 1991; revised Augrcst 26, 1991 ~~ Human beings have always been intuitive toxicologists, relying on their senses of sight, taste, and smell to detect harmful or unsafe food, water, and air. As we have come to recognize that our senses are not adequate to assess the dangers inherent in exposure to a chemical substance, we have created the sciences of toxicology and risk assessment to perform this function. Yet despite this great effort to Overcome the limitations of intuitive toxicology, it has become evident that even our best scientific methods still depend heavily on extrapolations and judgments in order to infer human health risks from animal data. Many observers have acknowledged the inherent subjectivity in the assessment of chemical risks and have indicated a need to examine the subjective or intuitive elements of expert and lay risk judgments. We have begun such an examination by surveying members of the Society of Toxicology and the lay public about basic toxicological concepts, assumptions, and interpretations. Our results demonstrate large differences between toxicologists and laypeople, as well as differences between toxicologists working in industry, academia, and government. In addition, we find that toxicologists are sharply divided in their opinions about the ability to predict a chemical's effect on human health on the basis of animal studies. We argue that these results place the problems of risk communication in a new light. Although the survey identifies misconceptions that experts should clarify for the public, it also suggeststhat controversiesover chemicalrisks may be fueled as much by limitationsof the science of risk assessment and disagreements among experts as by public misconceptions. ~ ~~ KEY WORDS:Intuitive toxicology; risk perception; risk assessment; chemical risks; expert judgment. Human beings have always been "intuitive toxicologists," relying on their senses of sight, taste, and WIl to detect harmful or unsafe food, water, and air. b g the past several years, we have been studying perception of risks from chemicals, based on this -pt of intuitive toxicology. As we have come to recognize that our senses and %ions are not adequate to assess the dangers inherent -sure to a chemical substance, we have created : b n Research, 1201 Oak Street, Eugene, Oregon 97401. Consulting, AB, Vgstmannagatan 48, $113 25 Stock- .bar. Sweden. the sciences of toxicology and risk assessment to perform this function. Massive regulatory establishments have been formed to oversee the use of these sciences for standard-setting and policy-making in the interests of protecting public health. Yet despite this enormous effort, people in many industrialized nations feel increasingly vulnerable to the risks from technology and believe that the worst is yet to come.") Regulatory agencies have become embroiled in rancorous conflicts, caught between angry environmentalists on one side and frustrated technologists and industrialists on the other. The lay public is anxious and confused. Nuclear power has long been a focus of public concerns. Today,there is great concern and dissatisfac- tion with the production, use, transport, and disposal of 215 0272433-215SLX 50/1 1992 Socicry for Risk Analysis I 216 Kraus eta chemicals. The lay public has clearly not been satisfied with, or persuaded by, scientific assessments of risk for either nuclear power or chemical hazards. Industrialists and scientists are equally dissatisfied with the public, whose concerns they equate with ignorance or worse, irrationality. Industry leaders see an urgent need to educate and inform the public in order to erase the "fear and sensationalism" that has tarnished the image of chemicals and the chemical Many have turned to the young field of risk communication in search of guidance that might make conflicts over technological decisions easier to re~olve.(~M*u~c)h of the public unease and social conflict may be traceable to the inherent limitations of the science itself, which is short on knowledge of the mechanisms by which chemicals harm living organisms. In the absence of mechanistic knowledge, risk assessment must rely upon indirect methods, which infer human health effects from bacterial and animal responses to chemical exposures. These indirect methods have become the primary bases for government regulation and policy. Although the limitations and uncertainties underlying these methods are readily acknowledged, it is assumed that, used conservatively, they can identify chemical threats and estimate the upper bounds on risk with enough accuracy to adequately protect public health. Behind this orderly facade of what Efron@)has labeled "regulatoy" science, in the trenches of "basic" science, the field of risk assessment has been in a state of intense controversy for more than a decade.("") A number of investigators have argued that the ambiguities and uncertainties of risk assessment are fertile grounds for influence from political and ideological consideration^.(^^*^-^^) For example, Lynn(*6)found that attitudes of occupational health specialists toward fundamental issues in risk assessment were strongly related to whether they worked in industry, government, or academia and whether they had voted Democratic or Republican in the previous presidential election. Graham et u Z . ( ~ ) have illustrated these kinds of biases in their detailed description of the controversies surrounding the regulation of benzene and formaldehyde. A committee of the National Academy of Sciences has responded to this problem by calling for procedural changes designed to improve the scientific basis of risk assessment and insulate scientific analyses and interpretations from political influence.(4)It is significant that this committee highlighted the inherent subjectivity in toxicological risk assessment, noting @. 36) that the procedures scientists follow and the interpretations they make ". . . rest, to various degrees, on a mixture of scientific fact and consensus, on informed scientificjudgment, and on policy determinations (e.g., the appropriate degree of conserva- tism)." The Academy report further indicated the need Q examine the subjective and intuitive elements of expert apd lay risk assessments-both to shore up the foundationsqf scientificanalyses and to improve communication betwq, scientists and the lay public. 1.1. Objectives of this Research The National Academy of Sciences report acknowl, edged the subjective nature of risk assessment, b not fully come to grips with its implications. One implication is that different assumptions, concept and values, in addition to (or instead of) disagre about facts, might underlie much of the discrepa tween expert and lay views of chemical risks. The objective of the present study is to addre issue by exploring the cognitive models, assumpti and inference methods that comprise laypeoples' tuitive toxicological theories" and by comparin theories with cognitive models, assumptions, an ence methods of scientific toxicology and risk ment. Such comparisons should expose the similarities and differenceswithin the exper as well as the similarities and differences perceptions and expert views. We hope that the edge gained from these comparisons will provide able starting point around which to structure disc education, and communication about the assess risks from chemicals and the management of those 1.2. Questionnaire Development .:a. We began by identifying several fundamental ciples and jddgmental components within the scienceof risk assessment. Questions were developed based on t h q fundamentals in order to determine the extent to whidf laypeople and experts share the same beliefs and coo- ceptual framework. Our questions addressed the following four topics: Category 1: Dose-response sensitivity Categoly 2: Trust in animal and bacterial studies Category 3: Attitudes toward chemicals Categoly 4: Attitudes toward reducing chemical risks Questions on these topics were incorporated into a single questionnaire, designed for both experts and the pubk ".Respondents were instructed to think of the term "chem- icals', as including . .all chemical elements and cam- pounds, including pesticides, food additives, indushial chemicals, household cleaning agents, prescription aod `i I Intuitive Toxicology 217 nonprescription drugs, etc." Tables I through V present ognize the potential toxicity of natural substances as well the specific questions used, categorized according to four as synthetic ones.(21) topic areas and a miscellaneous set of items (Cate- With regard to Category 2, we hypothesized that, gory 9 2 All but one of the questions were posed as state- because the science of toxicology and the discipline of ' ments to be evaluated by an agree/disagree response.6 Each question was designed, whenever possible, according to a guiding hypothesis about how experts and risk assessment rely so heavily upon animal studies, experts would have a more favorable view than laypersons `regarding the value of such studies. The prediction that "lay toxicologists" might respond. For example, per- laypersons lack sensitivity to dose-response considera- haps the most important principle in toxicology is the tions and thus fear even small exposures to toxic or car- fad that "the dose makes the poison."' Any substance cinogenic substances led us to expect that they would cancause a toxic effect if the dose is great enough. Thus, exhibit far more negative attitudes toward chemicals than we expected experts to be quite sensitive to considera- experts when responding to the questions in Category 3. tions of exposure and dose when responding to the five Similarly, we expected that laypersons' concerns about itmu in Category 1. In contrast, the often observed con- small exposures would cause them, more than experts, ' cmm of the public regarding very small exposures or doses of chemicals led us to hypothesize that the public to want reduction and even elimination of chemical risks, regardless of cost (Category 4). would have more of an "all or none" view of toxicity Turning to Category 5, we expected that laypersons rad would be rather insensitive to concentration, dose, would perceive chemicals in prescription drugs as less `( rad exposure (thus, equating exposure with harm) when nsponding to items in Category 1. We also expected toxic than chemicals used in pesticides (question 5b). Further, we expected that laypersons' direct experiences I ns to view natural substances as less toxic than with medicines and their knowledge of the risk of ov- c substances (question 5a), and experts to rec- erdose would make them more likely to recognize that the dose makes the poison for prescription drugs than 'Tk actual survey contained several additional questions not dis- for chemicals in general (see the two forms of question i cI#d in this paper because they were essentially redundant with Se). Similarly, for question 5f, we expected that a 1 in povbas in Table 1. "Riore question, not shown in Tables I-V,was as follows: "In f#ral. humans are (more, equally, less) vulnerable to adverse efhof chemicals than are laboratory animals" (question 5 ) . `hireference to the observation by Paracelsus that "All substances En poisons; there is none that is not a poison. The risk dose differ- lO,OOO,OOO lifetime risk would be less worrisome if it came from taking a prescription drug than if it came from exposure to a chemical. Question 5h examined the interpretation of evi- dence pertaining to the cause of birth defects observed entiates a poison and a remedy" (as quoted in Ref. 20). in a region where pesticides had been used. We hypoth- Table I. Responses of Toxicologists and Laypersons to Questions About Dose-Response Relationships (Category 1)" optslions Strongly Strongly Don't know/ disagree Disagree Agree agree no opinion h. lfyou are exposed to a toxic chemical substance, then you are likely to suffer F pdverse health effects P lh you are exposed to a carcinogen, then you are likely to get cancer T 14.0 53.7 24.4 4.3 3.1 9.3 51.2 34.3 25.7 62.3 7.8 0.6 3.1 1.9 3.6 P 5.1 42.6 25.4 9.0 17.2 IC For pesticides, it's not how much of the chemical you are exposed to that T 61.5 33.1 1.8 2.4 1.2 Jbould wony you, but whether or not you are exposed to it at all P 11.9 47.3 29.2 6.9 4.6 .A chemical was found in a city's supply of drinking water in a concentration T Of30 parts per million. .,The water was filtered by a process that was able P lo reduce, but not eliminate, the chemical concentration in the water. Under 1.2 6.5 56.8 31.4 3.8 18.8 63.2 7.3 4.1 7.1 most circumstances, this means that the danger associated with drinking the Water has also been reduced There is no safe level of exposure to a cancer-causing agent T 27.7 47.0 13.9 4.8 6.6 *' P 6.6 28.1 35.5 18.4 11.3 entries are percentages. Unless otherwise noted, differences between groups in this and all subsequent tables were significant ai p < .01. `T = Toxicologists. *) = Public. - x ? i 218 Table Il. Responses of Toxicologistsand Laypersons to Questions about Trust in Animal Studies (Category 2) Questions Strongly I. Strongly Don't disagree Disagree Agree agree no opinh 2a. The way that an animal reacts to a chemical is a reliable predictor of how a P human would react to the same chemical P 2b. Laboratory studies of a chemical's harmful effects on animals will, if properly T done, identify all possible harmful effects of that chemical P 2c. Laboratory studies of a chemical's harmful effects on animals allow scientists T to accurately determine how much of the chemical it takes to cause similar P harm in humans 2d. If a scientific study produces evidence that a chemical causes cancer in ani- T mals, then we can be reasonably sure that the chemical will cause cancer in P humans 1.9 38.9 50.3 5.1 5.5 40.2 40.2 3.5 26.8 56.1 15.2 1.2 17.1 60.7 11.3 2.3 12.2 55.5 29.3 1.2 11.7 50.0 28.9 0.8 10.3 41.3 39.4 1.2 1.9 22.9 64.0 5.4 a 3.8 10.6 0.6 8.6 1.8 8.6 1.8 5.8 a T = Toxicologists. P = Public. f esized that experts would believe that the evidence presented was not sufficient to indict pesticides as the cause of the observed malformations, whereas laypersons would be more inclined to view the association between pesticides and birth defects as causal (see Jenkins and Ward,(") who demonstrated the difficulty people have in dissociating covariation from causality). We had no strong prior hypothesis about how either experts or laypersons would judge the relative vulnerability of animals and humans to adverse chemical effects. Differing views can be found in the scientific literature. Some argue that humans are less vulnerable than the particularly sensitive strains of animals used in many bioassays. Humans are also exposed to far smaller amounts of chemicals than the test animals. Others argue that humans are more vulnerable, because of their slower metabolic processes, their great genetic diversity, and their exposure to a greater number of carcinogenic substances than laboratory animals. Regulatory policies typically assume that humans are more sensitive to adverse effects than animals. 1.3. Administration of the Survey 1.3.1. Expert Sample The sample of experts was selected from the 1988 membership directory of the Societyof Toxicology (SOT). Only full members of the Society residing in the United States were considered for inclusion in the study. The names of these full SOT members were categorized into three subgroups based on the type of organization with which they were affiliated: academic, industrial, or regulatory. Random numbers were used to select 120 names from each of the three affiliational subgroups. The sur- vey was mailed to each of these persons, along with 8 cover letter from the past president of the SOT. In all, 360 questionnaires were mailed in October 1988, and 170 completed questionnaires were retumsd, for an overall response rate of 47%. Subgroup response rates were consistent with this overall rate; 44% of C~QSC from the academic subgroup returned their quest& naires, along with 49% of those affiliated with indus$y* and 48% of those affiliated with regulatory a Those who responded were highly educated (91 Ph.D. degree and 2.4% an M.D. degree); 84.6% of& respondents were male; 58.9% were between the agts of 33 and 48; 37.5% were ages 49-64; and 3.6% wac younger than 32 or older than 64 years. ad:+ 1.4. Public Sample 1 P, A sample of the general public was selected from the Portland, Oregon metropolitan area. A p r o f e s s i d listing organization provided a sampling frame organized by zip code. This population listing was screened for a minimum annual household income of $20,000, and 1100 households were randomly selected for the mailing, which took place on October 25, 1988. Those who had not replied by November 15, 1988 were sent another copy of the survey instrument. Of the approximately 975 deliverable question- naires, 262 usable questionnaires were returned, for a response rate of about 27%. These individualswere nonrepresentative of the Portland general popuiation in sev- era1ways. Specifically, respondents were more likely to be male (58.0% vs. 48.6% in the population); well ducated (95.7% high school graduates in the sample and I bi -the Toxicology 219 33.1% with graduate-school training vs. 78.7% and 9.9%, aq#dively, in the population); and high income [24.6% d b sample had household incomes greater than $50,000 vr;9.0% of the population in 1980 (corrected for 42.9% -dative inflation between 1980 and 1988). Most @6.5%) of the respondents were white (vs. 92.8% of *population); 24.1% were between the ages of 16- 46.4% were ages 33-48; 14.9% were ages 48-64; sda 14.6% were older than 64 years. rt vs. Layperson Comparisons expected that the toxicologists would differ .m the laypersons in their respo`nses to most However, there was striking agreement on 5g, which asked respondents whether or not with the statement: "I think I should know I can about the chemicals around me. The the more I can control the risks that these to my health and to the environment." of respondents who agreed or strongly s statement was 94.6% among the toxi93.4% among the public. Thus, though ptions and attitudes may differ, the stated moof experts and laypersons to understand and hemical risks appear to be remarkably similar. -- h `CategoryI : Dose-Response Sensitivity sents the responses for the five questions d to assess sensitivity to the view that the poison." For each question, memhibited much less appreciation of did the toxicologists. Specifically, the nts were more likely than the toxicolo- exposure to a toxic chemical makes one likely to Suffer adverse health effects exposure to a carcinogen makes one likely to get noted in the tables, all of the reported differences gists and laypersons or between subgroups wiihin are statistically significant at p < .OI by a chi-square test that included the nses. Significance levels are not given the fact of exposure to a pesticide is the critical concern, rather than the amount of exposure 0 reducing the concentration of a possibly harmful chemical in a city's drinking water would not reduce the danger associated with drinking that water there is no safe level of exposure to a cancercausing agent Note that these views, though much more common in the public sample, were not always held by a majority of the public. Note also the high percentage (17.2%) of the public who marked "don't know" or "no opinion" regarding whether exposure to a carcinogen implies that one is likely to contract cancer (question lb). 2.1.2. Categoly 2: Trust in Animal Studies Comparisons with regard to trust in animal studies are shown in Table 11. There was relatively little difference in responses of the public and toxicologists to questions 2a, Zb, and 2c (see Table II). Both groups of respondents were greatly divided in their opinions about whether animal reactions to a chemical are reliable predictors of human reactions to the same chemical (question 2a). About 41% of the experts and 46% of the public disagreedwith that assertion. Therewas very little agreement, in either group, with the propositions that proper animal studies could identify all possible harmful effects of a chemical (2b) and that laboratory studies could al- low scientists to make accurate predictions about the amount of chemical exposure needed to cause human harm (k). The contrast between questions 2a and 2d is instructive. The confidence of toxicologists in predicting human harm from animal studies decreased when a chemical was found to be a carcinogen in animals (agreement decreased from 55.4% in 2a to 40.6% in 2d). However, the public respondents became much more confident in the prediction of human harm from an animal test when the test was said to produce evidence of carcinogenicity in animals (agreement rose from 43.7-69.4%). As a result, the two groups differed considerably on question 2d. 2.1.3. Categories 3 and 4: Attitudes Toward Chemicals and Risk Reduction Many of the public respondents in this study appeared to believe that if large exposures to a chemical are harmful, then small exposures are also harmful, and 220 b u s et al. that animal tests can give little guidance about riskexcept when they imply harm. Given these views, we would expect these individuals to be consistently negative in their general attitudes toward chemical risks (Category 3) and to be strongly in favor of extreme measures to reduce such risks, regardless of cost. The data in Tables I11 and IV support this expectation. The majority of the public respondents 0 agreed that we have perceived only the tip of the chemical risk iceberg (83.9%agree or strongly agree) 0 agreed that contamination is greater now than ever before (88.4%) 0 disagreed that people are unnecessarily frightened about small amounts of pesticides in groundwater and food (68.7%) disagreed that people worry unnecessarily about chemicals (87.1%) agreed that chemical risks are too scary and they don't like to think about them (84.0%) 0 agreed that it can never be too expensive to reduce the risks associatedwith chemicals (61.6%) Although not a majority view, a surprisingly high percentage of the public sample agreed that they do everything they can to avoid contact with chemicals and chemical products (40.0%) and agreed that all use of prescription drugs (17.2%)and chemicals (29.3%)must be risk-free. In comparison with the public sample, the toxicologists conveyed more favorable attitudes toward chemicals on every question in Category 3. In most cases these differenceswere substantial. The story was similar with respect to questions in Category 4, with toxicolo. gists showing much less agreement with the assertions that use of chemicals must be risk-free regardlessof cost. 2.1.4. Category 5: Additional Comparisons The results for question 5a in Table 5 confirm commonly held belief that the public has much more confidence in the safety of natural (as opposed to synthetic) chemicals than do the experts. Answers to questions %, 5e, and 5f show that the public respondents had a mu& more favorable view of chemicals and were more tolerant of chemical risks if the chemicals were in the form of prescription drugs. Similarly, people were more likely to recognize that the risk from a prescription drug is dependent on dose than they were to recognize dose dependence of chemicals in general (question 5e). Question 4c, shown in Table IV, also found a more tolerant public attitude toward risks from prescription drugs than for risks from chemicals in general. The toxicologists' versions of these questions used only the term prescription drugs, due to an error in printing the questionnaire. Compared to the public, toxicologists were not as likely to see prescription as less toxic than pesticides (question 5b), more likely to 1 1 t Questions Table III. Responses of Toxicologists and Laypersons to Questions about Attitudes Toward Chemicals (Category 3) ~ ~~~ . Strongly Strongly Don't know/ disagree Disagree Agree agree no opinion 3a. Our society has perceived only the tip of the iceberg with regard to the risks T" associated with chemicals Pb 3b. The land, air, and water around us are, in general, more contaminated now T than ever before P 3c. Use of chemicals has improved our health more than it has harmed it T P 3d. People are unnecessarily frightened about very small amounts of pesticides T found in groundwater and on fresh food P 3e. People wony unnecessarily about what chemicals can do to their health T P 3f. Chemicals are a major force behind technological advancement T P 3g. Chemical risks are too scary. I don't even like to think about them T P 3h. I do everything I can to avoid contact with chemicals and chemical products T in my daily life P 10.9 36.4 37.6 8.5 1.6 9.0 50.2 33.7 3.6 24.8 53.3 13.9 1.5 8.1 45.2 43.2 0.0 3.0 59.4 33.3 7.9 26.0 41.7 5.5 1.8 26.8 53.0 14.0 13.1 55.6 20.1 3.9 3.6 53.3 33.9 4.2 27.1 60.1 8.1 1.6 0.0 1.8 61.1 35.9 1.9 12.4 55.4 13.2 74.6 24.9 0.6 0.0 37.5 46.5 11.7 3.6 44.0 46.4 6.0 3.6 11.7 47.5 31.1 8.9 6.7 5.5 4.2 1.9 4.2 18.9 4.3 7.3 4.8 3.1 1.2 17.1 0.0 0.8 0.0 0.8 a T = Toxicologists. P = Public. ! htuitive Toxicology 221 Table Tv. Responses of Toxicologists and Laypersons to Questions about Attitudes Toward Risk Reduction (Category 4) O#stions Strongly Strongly Don't know/ disagree Disagree Agree agree no opinion 4a. While we should always try to minimize the risks we take by using chemicals, it is unrealistic to expect that we can completely elimipate those risks 4b. It can never be too expensive to reduce the risks associated with chemicals 4c. AH use of prescription drugs must be risk free. (Alternative form: All use of chemicals must be risk free) T" P T P Tb Pdd PC ''T = Toxicologists. P = Public. ` -Td = Toxicologists, prescription drugs. Pd = Public, prescription drugs. `PC Public, chemicals. 0.6 1.2 45.8 52.4 2.8 14.2 65.8 15.0 27.7 54.2 12.7 3.0 2.7 28.7 43.4 18.2 54.8 39.9 3.6 0.6 13.4 64.2 12.7 4.5 8.9 52.8 22.8 6.5 0.0 2.3 2.4 7.0 1.2 5.2 8.9 , -ri",.. Table V. Responses of Toxicologistsand Laypersons to the Miscellaneous Questions in Category 5 & c b lchemicals, as a rule. are not as harmful as man-made chemicals T" ps % Ingeneral, chemicals used in prescription drugs are less toxic than chemicals T used in pesticides P !k In general, chemicals used as pesticides are potentially more harmful than are T chemicals used as fertilizers P a.How microorganisms(e.g., bacteria) react to a chemical is a reliable indicator T ofhow humans would react to the same chemical P k. Some prescription drugs are harmful if they are taken in large doses, but arc T& not harmful if taken in small doses. (Alternate form: Some chemicals are Pdd harmful if people are exposed to them in large amounts, but are not harmful P d ifpeople are exposed to them in small amounts) 5f. A 1 in lO,OOO,OOO lifetime risk of cancer from taking a particular prescription T b drug is tw small a risk to worry about. (For perspective the lifetime risk of Pff dying in a car accident in 1 in 100.) [Alternate form: A 1 in 10,OoO.OOO P f lifetime risk of cancer from exposure to a particular chemical is too small a risk to worry about. (For perspective, the lifetime risk of dying in a car accident is 1 in loo)] 5g. I think that I should know as much as I can about the chemicals around me. T The more I know,the more I can control the risks that those chemicals pose P to my health and to the environment 51 Residents of a small community (30,000people) observed that several mal- T formed children had been born there during each of the past few years. The P town is in a region where agricultural pesticides have been used during the past decade. It is very likely that these pesticides were the cause of the dformations T = Toxicologists. `p = Public. `Td = Toxicologists, prescription drugs. Pd = Public, prescription drugs. 'k = Public, chemicals. Strongly Strongly Don't know/ disagree Disagree Agree agrw no opinion 45.6 40.2 11.2 2.4 10.8 34.0 37.8 7.3 22.8 41.9 21.6 6.6 11.5 32.1 31.1 6.9 6.0 30.5 52.1 6.6 5.4 30.9 32.8 3.5 38.4 51.2 5.5 1.2 12.4 43.4 7.8 1.6 1.2 11.4 50.0 34.9 2.2 12.7 65.7 14.2 2.4 28.0 50.4 7.2 0.6 10.0 1.2 11.2 4.8 27.4 3.7 34.9 2.4 5.2 12.0 1.8 4.2 53.0 40.0 1.5 14.3 65.4 14.3 3.3 26.8 52.8 6.5 1.2 4.5 10.6 0.6 4.1 47.9 46.7 0.8 4.7 53.9 39.5 22.2 59.3 4.3 1.2 3.9 23.4 39.5 9.0 0.6 1.2 13.0 24.2 I 222 recognize the dependence of drug risk on dose (5e), and more likely to agree that a 1in lO,OOo,OOo risk of cancer from taking a prescription drug is too small a risk to wony about (5f). There was considerable disagreement among both toxicologists and members of the public about whether pesticides are generally more harmful than fertilizers (question 5c). The toxicologists agreed with this statement more than the public did, the latter producing many more "don't know or no opinion" responses (27.4%). Question 5d is similar to the questions about trust in animal studies (Category 2; Table 11). Toxicologists almost unanimously disagreed (89.6%) that microorganisms are reliable indicators of human reactions to a chemical. Members of the public sample also tended to disagree with this statement but exhibited an extremely high percentage of "don't knowlno opinion" responses (34.9%). Results for question 5h confirmed the hypothesis that the public would be much more likely than the toxicologists to view an association between pesticide use and birth defects as a causal relationship (48.5% agreement among the public vs. 5.5% among the toxicologists). The percentage of "don't knowho opinion'' responses to this question was high in both groups. Results from question 5i (not shown in Table v) indicated that toxicologists were about equally split between the views that humans are more, equally, or less vulnerable to adverse effects of chemicals than animals. The dominant response in the public sample was equal vulnerability (35.0%). More than 20% in both samples marked the "don't knowlno opinion" response. 2.2. Predicting Perceptions and Attitudes Comparisons between toxicologists and the public sample showed that these two groups differed greatly with regard to appreciation of dose-response relation- ships and attitudes toward chemical risks and risk re- duction, but did not differ much regarding trust in animal testing. From these data, it appears that dose-response views and attitudes may be closely related but that neither type of response is related to trust in animal testing. A separate analysis was done to test these relationships at the level of the individual respondent in the public sample. The items within each of the first four categories (Tables I-IV) were assigned scores on a fourpoint scale of agreement corresponding to the four responses (strongly disagree = 1, disagree = 2, agree = 3, strongly agree = 4). Responses of each person in the public sample were scored, and the scores were su across all items in the category to produce four scores per person: 0 dose-response sensitivity (highscore indicates sensitivity to dose) trust in animal studies (high score indicates hi& trust) 0 attitude toward chemicals and their risks (hi@ score indicates a pro-chemical view) attitude toward risk-reduction (high score in& cates acceptance of some degree of risk chemicals) The correlations between scales were computed across members of the public sample. The pattern correlations indicated that laypersons who believed that exposure and dose mediate chemical risks were likely be more favorable toward chemicals and relatively im concerned about their risks (r = .68) and less concerned about the necessity of reducing chemical risks (r = .47j, Attitudes toward chemical risks and attitudes toward reduction were also positively correlated (r = -55). Trust in animal studies did not correlate significantlywith any of the other scales. Multiple regression analyses were performed to pndict scale scores from demographic characteristicsof public sample. Examination of variables making tistically significant contribution to the regression tion showed that scores on the attitude toward chemic4 risks scale could be predicted moderatelywell (R = .49) on the basis of education, gender, and scientifi Scientific training, more education, better he male gender were all predictive of more f tudes. Dose-response sensitivity was pred -40)on the basis of health status, education, and Being older; well-educated, and in good health was padictive of greater dose-response sensitivity. Trust in animal testing was not very predictable (R = -22)on tbt basis of demographic variables. 2.2.1. Education It is instructive to assess the influence of the im- portant demographic variables by examining data from specific questions. Table VI presents items on which tbc college educated and noncollege educated respondents showed statistically significant differences. In general, respondents with a college degree had more favorable attitudes toward chemicals, greater appreciation of the mediating role of dose and exposure, and less c O n m Intuitive Toxicology 223 Table VI. Comparisons Between Responses of College-Educated and Non-College Educated Persons in the Public Sample" oaestions Strongly Strong-ly. Don't know/ disagree Disagree Agree agree no opinion 1b. If you are exposed to a carcinogen, then you are likely to get cancer NC IC For pesticides, it's not how much of the chemical you are exposed to that should worry you, but whether or not you are exposed to it at all .Id. A chemical was found in a city's supply of drinking water in a concentration of30 parts per million. . . the water was filtered by a process that was able to reduce, but not eliminate, the chemical concentration in the water. Under most circumstances this means that the danger associated with drinking the water has also been reduced le. There is no safe level of exposure to a cancer-causing agent 3c -Use of chemicals has improved our health more than it has harmed it X chemicals are a major force behind technological advancement 4. B can never be IOO expensive to reduce the risks associated with chemicals NC C NC C NC C NC C NC C NC C 3.3 38.5 23.0 9.0 7.0 46.9 28.1 9.4 8.9 40.3 39.5 5.6 15.4 53.8 19.2 7.7 4.8 23.2 60.8 0.8 3.1 13.8 65.4 13.8 2.5 23.8 38.5 22.1 10.2 31.3 32.8 15.6 9.2 34.2 32.5 3.3 7.0 18.8 49.2 7.8 3.3 15.4 51.2 5.7 0.8 10.1 58.9 20.2 0.0 23.6 52.0 16.3 5.4 33.3 34.9 20.2 26.2 8.6 5.6 3.8 10.4 3.8 13.1 10.2 20.8 17.2 24.4 10.1 8.1 6.2 h dfirences between college and non-college educated respondents on question l e were significant at p c .OS. All other comparisons were sipifkant atp c .01. = Non-College q$=cotkge dput risks. In these respects, the college-educated retipdents appeared to be somewhat more similar to the &kcdogists than were the noncollege educated persons. Examination of Table VI indicates that college-educated nspondents were less likely to agree that worries about pesticides should depend only on whether or not one was exposed at all (IC) 0 more likely to see reduction in the concentration of a possibly harmful chemical in water as reducing risk (Id) 0 less likely to agree that there is no safe level of exposure to a cancer-causing agent (2e) more likely to agree that use of chemicals has improved our health more than it has harmed it (3c) 0 more likely to agree that chemicals are a major force behind technological advancement (30 0 less likely to agree that it can never be too ex- pensive to decrease the risk associated with chemicals (4b) Table VI also illustrates that respondents who did Mt have a college degree were far more likely to answer 6` don't knowlno opinion" on certain of these items (lb, Id, 30. 2.2.2. Gender Within the public sample, women were consistently more concerned about chemical risks than men, and they had less favorable attitudes regarding the benefits of chemicals. Questions for which gender differences were largest are shown in Table VII. Note that differences were particularly large for the attitudinal questions-3c, 3f, and 4b. Women were much less likely to agree that use of chemicals has improved our health more than it has harmed it (3c) and less likely to agree that chemicals are a major force behind technological advancement (3f). Note also the high percentage of "don't knowino opinion" responses among women for these questions. On question 4b, women were more likely than men to agree that it can never be too expensive to reduce chemical risks. On the other questions, women were more likely than men to: 0 agree that, for pesticides, one should worry about whether or not they were exposed at all (IC) 0 agree that they do everything they can to avoid contact with chemicals and chemical products in their daily life (3h) 0 agree that pesticides caused the malformations in the children (5h) I 224 Table VII. Responses of Men and Women in the Public Sample to Selected Question? Questions IC. For pesticides, it's not how much of the chemical you are exposed to that Mb should worry you, but whether or not you are exposed to it at all W` 3c. Use of chemicals has improved our health more than it has harmed it M W 3f. Chemicals are a major force behind technological advancement M W 3h. I do everything I can to avoid contact with chemicals and chemical products M in my daily life W 4b. It can never be too expensive to reduce the risks associated with chemicals M W 5h. Residents of a small community (30,000 people) observed that several mal- M formed children had been born there during each of the past few years. The W town is in a region where agricultural pesticides have been used during the past decade. It is very IikeIy that these pesticides were the cause of the malformations Strongly Strongly Don't disagree Disagree Agree agree 14.4 54.1 20.5 7.5 no o p m j .--- 3.4 9.3 37.4 41.1 6.5 5.6 6.9 21.4 51.7 8.3 11.7 8.8 33.3 27.5 1.0 29.4 1.4 8.2 61.6 19.2 9.6 2.9 19.0 46.7 3.8 27.6 15.8 50.7 27.4 6.2 6.7 44.8 34.3 12.4 0.0 1.9 4.1 34.7 40.8 14.3 6.1 1.0 19.2 46.2 25.0 8.7 4.9 29.9 36.1 5.6 23.6 2.9 14.3 43.8 14.3 24.8 Gender differences were significant at p < .05 for questions 3h and 4b, and at p < .01 for questions IC,3c, 3f, and Sh. M = Men W = Women Gender differences were less evident in the sample of toxicologists, perhaps because the number of women was too small (n = 26)to measure them reliably. However, female toxicologists appeared to be more concerned about chemical risks and less favorably impressed with the benefits of chemicals than were male toxicologists, though the differences between men and women toxicologists were smaller than the differences between men and women in the public sample. In other words, men and women exhibited different response patterns in both samples, although women toxicologists were far more similar to men toxicologists than they were to women members of the public. 2.3. Affiliation Effects The three subgroups of toxicologists responded similarly on many of the questions. However, there were also important differences between the groups on a number of questions, as shown in Table VIII. Question l e demonstrates that toxicologists from industry were much less likely than others to agree with the assertion that there is no safe level of exposure to a cancer-causing agent. With regard to the reliability and accuracy of animal tests, industrial toxicologists were somewhat more favorably inclined toward the tests (questions 2b and 2c) than other toxicologists, though the differences were not statistically significant and are not shown in the table. k However, if a study produced evidence that a c caused cancer in animals (question 2d), the toxicologistswere much less confident tha regulatory scientists in the ability to extr mans (22.0% agreement vs. 48.9% and other two groups). Similarly, industri much more likely than others to see nerable than animals to adverse (question 5). Industrial toxicologists also had the most f attitudes toward chemicals and their risks (questio and 3d), and they were the group of tox most often dihagreed with the statement were the likely cause of the observ question 5h. 2.4. Are Public/Toxicologist Differences Merely Demographic? The toxicologists in our survey tended, for the most part, to be highly educated, white males. We have see0 that, within the general public, attitudes toward chemical risks are influenced by education and gender. Is it possible that these factors, and not expertise in toxicology, account for the observed differences between toxicolo- gists and the public? Two kinds of analyses were pa- formed to address these questions. First, multipk regression analyses were conducted to predict answers Intuitive Toxicology 225 Table VIII. Responses of Toxicologists According to Affiliation" QKstions Strongly Strongly Don't know/ disagree Disagree Agree agree no opinion IC. There is no safe level of exposure to a cancer-causing agent Ab 18.0 IC 45.8 Rd 17.5 a. If a scientific study produces evidence that a chemical causes cancer in ani- A 4.1 mals, then we can be reasonably sure that the chemical will cause cancer in I 18.6 humans R 7.0 k. Our society has perceived only the tip of the iceberg with regard to the risks A 8.0 associated with chemicals I 19.0 R 5.3 36 People are unnecessarily frightened about very small amounts of pesticides A 2.1 found in groundwater and on fresh food I 0.0 R 3.5 9. Rcsidcnts of a small community (30,000 people) observed that several mal- A 20.8 termed children had been born there during each of the past few years. The I 29.3 b w is in a region where agricultural pesticides have been used during the R 16.1 past decade. It is very likely that these pesticides were the cause of the nulformatiOnS 46.0 22.0 12.0 42.4 5.1 0.0 52.6 15.8 3.5 44.9 46.9 2.0 57.6 22.0 0.0 38.6 50.9 1.8 26.0 46.0 12.0 44.8 24.1 8.6 36.8 43.9 5.3 35.4 45.8 14.6 16.9 61.0 20.3 29.8 50.9 7.0 58.3 2.1 4.2 63.8 1.7 0.0 55.4 8.9 0.0 2.0 6.8 10.5 2.0 1.7 1.8 8.0 3.4 8.8 2.1 1.7 8.8 14.6 5.2 19.6 More EsuauY Less Don't know/ vulnerable vulnerable vulnerable no opinion humans are (more, equally, less) vulnerable to adverse effects of A 15.2 21.7 23.9 than are laboratory animals I 19.6 21.4 44.6 R 26.3 33.3 17.5 ~~ ~~ ~~~ ~~~ ~~ ' ~ c m c e asmong groups were significant et p < .05 for questions 2d, 3a, 3d, and Sh, and at p < .01 for questions l e and Si. 'A = Academic. r3pI-sndt8dustry. "Ir= Regulatory. 39.1 14.3 22.8 b questions on which the two groups disagreed, using @der, race, education, and group status (member of tbe public vs. toxicologist) as predictors. These regres- & showed that group status was by far the most im- portant predictor, with education, race, and gender lecounting for significant but small amounts of the varh c e in responses to the questions. This result can be illustrated concretely by comparing white male toxicolest with college-educated white male laypersons on tw0typical questions-lb and 2d. The results, presented kt Table IX,show that the public sample still responds Saite differently from each of the three groups of toxi, ~ g i s t s e, ven when education, race, and gender difftnnces are minimized. Results from other questions similar. 2.5. Toxicologists' Attitudes Toward Public Concerns We examined relationships between toxicologists' U e f s about the reliability of animal tests (question 2a) and their beliefs about whether people worry unnecessarily about chemicals (questions 3d and 3e). Comparing questions 2a and 3d, the proportion of toxicologists who agreed that people are unnecessarily frightened about small pesticide residues was about the same (70.0%) within the subsample who agreed that animal tests are reliable predictors as among those who disagreed with the assertion that animal tests are reliable (69.3% of whom agreed that people are unnecessarily frightened). Comparing questions 2a and 3e, those who distrusted animal tests were somewhat more likely (45.3%) to agree that people worry unnecessarily about what chemicals can do to their health than were those who trusted animal studies (33.3% of whom agreed that people worry unnecessarily). These results raise a question. If distrust in animal studies is independent of or even positively related to toxicologists' beliefs that people wony unnecessarily about chemicals, what is the evidence, experience, or belief on which such views are based? At least a partial answer to this question can be found by examining responses to the questions about I 226 IC 33.3 R` 28.2 2d. If a scientific study produces evidence that a chemical causes cancer in ani- P 0.0 mals, then we can be reasonably sure that the chemical will cause cancer in A 4.8 humans I 20.0 R 10.0 57.8 4.4 66.7 2.6 28.6 63.6 45.2 45.2 55.6 22.2 47.5 37.5 0.0 0.0 3.9 2.4 0.0 2.5 4.4 2.6 3.9 2.4 P = Public. A = Academic toxicologists. `I = Industry toxicologists. R = Regulatory toxicologists. reliability (2a) and unnecessary fear (3e) in light of responses to a third question (59 asking about the relative vulnerability of humans and laboratory animals. The re- sults of this three-way comparison implicated percep- tions about vulnerability as a key factor in mediating attitudes toward public fears and distrust in animal tests. Toxicologists who disagreed with the statementthat animal tests are reliable predictors of human reactions and who also believed that humans are less vulnerable to adverse effects of chemicals than are laboratory animals were the persons most likely to believe that people are unnecessarily frightened about very small amounts.of pesticides. 2.6. Additional Data: The EPA Baseline Survey Concern about the low response rate (27%) and the limited geographic representation of our public sample caused us to collect additional data. We were fortunate to be able to add four questions to a telephone survey conducted by McCallum et uZ.(=) for the Environmental Protection Agency (EPA) in six communities across the United States: Albuquerque, New Mexico; Cincinnati, Ohio; Durham, North Carolina; Middlesex County, New Jersey; Racine, Wisconsin; and Richmond, Virginia. Several criteria guided the selection of these communities: the presence of major industries using, storing, processing, or releasing chemicals; location of a superfund or other hazardous waste site nearby; existence of an active local emergency planning or environmental group; and previous experience with chemical-emissions problems. A total of 3129 completed interviews were obtained from the six communities (sample sizes in each community ranged from a low of 503 to a high of 604). -- The overall response rate was 59.1%, ranging from 1 high in Racine of 62.9% to a low in Middlesex of 52.1%. Men and women were equally represe by design. Comparison of the sample wit demographics indicated that the respondents were what younger, better educated, and more affluen the general public as a whole. We selected two questions pertaining to sponse sensitivity (questions l b and IC),one question (3c), and one from Category 5 (5f). Th ing was changed slightly in each case to imp (e.g., the term "carcinogen" was replaced by the p "chemical that can cause cancer") or to in erality (e.g., "pesticide" was replaced by "ch in question IC). The results from these new samples, shown i ble X, differed even more from the respon cologists than did the responses of the Port1 sample. For example, 73.3% either agreed so agreed strongly that "if a person is expose ical that can cause cancer, then the person is likely get cancer later in life," in contrast to 34.4% of tbe Portland sample and 8.4% of the toxicologists. The new version of question ICalso produced a much higher rate of agreement with an all-or-none view of toxicity ("It's not how much of a chemical you are exposed to that matters to your health . . ."). Almost 63% agreed with this statement compared to 36.1% in the Portland sample and 4.2% of the toxicologists (on a version of the question that substituted pesticide for chemical). About 56% disagreed that some chemical risks are too small to W O about (compared to 30.1% in the Portland sample wbo responded to a 1 in 10,000,000 risk). There was about an even split in the percentage of persons who agreed I f f i 1 ~ '' lntuitive Toxicology 227 Table X. Responses from the EPA Baseline Survey Community Education Middle- Gender Some Age Rich- Dur- AIbu- Cincin- sex Racine H.S. college <30 30-50 >50 Total mond ham querque nati County County Male Female or less or more yr yr yr lb. Ifa person is exposed to a chem- ical that can cause cancer, then that person is likely to get cancer later in life Strongly disagree 6.5 6.7 5.7 8.1 6.9 5.0 6.3 7.9 5.0 5.7 7.0 4.6 5.9 8.8 Somewhat disagree 18.3 19.0 20.0 21.1 17.4 15.3 17.2 20.4 16.2 11.7 22.8 17.7 19.0 17.6 Somewhat agree 35.8 34.8 38.4 34.6 34.1 35.6 37.3 35.2 36.5 31.3 39.0 43.5 37.0 27.6 Strongly agree 37.5 36.4 33.3 34.0 39.8 42.9 38.4 34.2 40.8 50.2 28.8 33.7 36.0 43.2 Don't know/no opinion 1.6 2.8 2.2 2.0 1.4 0.8 0.8 1.8 1.4 1.1 2.0 0.4 1.6 2.6 IC. it's not how much of a chemical p u are exposed to that matters toyour health, it's whether or not you're exposed at all StrongIy disagree 18.6 16.6 q . 0 17.4 20.6 19.7 16.1 18.7 18.5 19.8 17.8 19.2 18.7 17.7 Somewhat disagree 17.0 15.0 17.8 19.4 17.0 14.7 18.0 19.4 14.7 13.3 19.6 21.6 17.7 12.4 Somewhat agree 23.5 25.1 20.6 21.9 24.2 23.1 25.0 24.9 22.0 22.0 24.4 22.6 24.1 23.3 Strongly agree 39.2 40.9 38.2 39.7 36.8 40.2 39.6 35.7 42.8 43.3 36.5 36.2 38.2 43.3 Don't knowho opinion 1.7 2.4 1.4 1.6 1.4 1.7 1.3 1.3 2.0 1.6 1.7 0.4 1.3 3.3 k.Chemicals have improved our health more than they have harmed it Strongly disagree 24.3 22.5 25.1 25.7 24.8 27.6 20.7 20.2 28.4 32.8 18.6 22.6 24.9 24.4 Somewhat disagree 22.9 23.5 23.6 19.4 20.0 25.6 24.8 19.9 25.9 22.9 22.9 26.9 24.0 18.0 Somewhat agree 34.4 36.2 31.7 35.0 35.0 30.0 37.7 37.3 31.5 28.3 38.5 37.2 34.0 32.8 StrongIy agree 15.5 15.0 15.6 17.2 17.2 13.1 14.9 20.5 10.4 13.1 17.1 12.0 14.9 19.2 Don't knowho opinion 2.6 2.6 3.2 2.8 2.8 3.0 1.7 1.7 3.5 2.8 2.5 1.4 1.7 5.2 sf- There are some chemical risksthat arc too small to worry about Strongly disagree 35.3 36.0 36.2 31.6 33.5 38.6 35.8 30.7 39.9 41.0 31.5 37.0 34.5 34.9 Somewhat disagree 21.1 19.6 19.6 20.9 23.0 23.7 20.0 20.0 22.2 20.6 21.4 24.8 21.1 18.2 Somewhat agree 26.4 28.3 26.1 28.9 25.0 22.3 27.6 28.0 24.8 22.1 29.4 25.2 28.4 24.1 Strongly agree 16.8 16.2 16.6 18.0 18.2 15.5 16.6 21.1 12.5 15.6 17.6 13.0 15.6 22.0 Don't knowho opinion 0.4 0.0 1.5 0.6 0.3 0.0 0.0 0.2 0.6 0.7 0.1 0.0 0.4 0.8 Sample size 3129 506 505 506 505 503 604 1574 1555 1283 1846 720 1502 907 * !kce: McCallum et and the percentage who disagreed that "chemicals have improved our health more than they have harmed it." About 47% disagreed, compared with 33.9% in the Portland sample and 3.0% of the toxicologists. The "don't know/no opinion" response category Was prominently displayed as an option in the written mwey administered to the Portland survey but was not Specifically mentioned as a response option in the EPA kkphone survey. As a result, the frequency of "don't how" responses was smaller in the EPA survey than the Portland survey. Further analysis of Table X shows that the overall response patterns or these questions varied rather little across communities, gender, education, and age. As in the Portland sample, women's attitudes were somewhat less favorable toward chemicals than were men's and respondents with more formal education were more favorable toward chemicals than were less educated persons. Older respondents were more likely than younger persons to agree that it's not how much of a chemical that matters and to agree that some risks are too small to worry about. However, they were slightly less likely than younger persons to agree that exposure to a cancercausing chemical will cause cancer later in life. I 228 3. DISCUSSION 3.1. Expert vs. Public Views Our primary objectives were to describe and compare responses of laypeople and toxicologists and to test a number of hypotheses about the differences between the public's views as "intuitive toxicologists" and the views of experts. In general, we believe that this aspect of the study was successful. Toxicologists and laypeople were found to differ greatly, documenting some common assumptions (e.g., that the public believes natural chemicals to be safer than synthetic ones) and verifying many of the other hypotheses that motivated the questionnaire. Of particular importance is the pattern of responses to questions in Category 1, which indicates that the public is much less sensitive than the experts to fundamental considerations of dose and exposure. Although the public recognizes the importance of these factors in some domains (e.g., prescription drugs), they generally tend to view chemicals as either safe or dangerous and they appear to equate even small exposures to toxic or carcinogenic chemicals with almost certain harm. This ori- entation was found to be associated with high levels of concern regarding chemicals, including very small residues of chemicals on food, and a desire to reduce chem- ical risks at any cost. It is sobering to find that 40% of our public respondents said that they do everything they can to avoid contact with chemicals and chemical products in their daily lives. It is also remarkable to find that fewer than 50%of the public respondents in the Portland sample and fewer than 25% in the EPA sample recognized that exposure to a chemical that is carcinogenic does not make one likely to get cancer later in life. Several other findings were noteworthy: (a) the public's tendency to attribute causality to a temporal association between pesticide use and birth defects; @) the strong negative attitudes of the public toward chemicals and their risks; (c) the relatively favorable perceptions of prescription drugs9; (d) the finding that 30% of the public respondents did not agree that a 1in lO,O00,000 lifetime risk of cancer from exposure to a chemical was We attribute this to the fact that prescription drugs are familiar, high in perceived benefit, and administered by highly trusted professionals. The more favorable view of prescription drugs relative to industrial and other chemicals is analogous to the more favorable perceptions of medical x-rays in comparison to industrial radiation technologies such as nuclear power and food irradiation (see Ref. 24) too small to worry aboutlo; and (e) the finding that and more highly educated persons were somew concerned about chemical risks-in general, re of college-educated persons were slightly more sim to responses of toxicologists than were responses of sons with less education. One of the major surprises in the data was the of difference between the public and toxicologists regard to their confidence in extrapolation fr studies. Both groups were divided in their opi the high percentage of experts who lacked co animal studies is particularly noteworthy in light of extensive reliance on such studies in risk managem The public's trust in extrapolation from animal stincreased greatly when these studies were said to duce evidence of carcinogenicity. It is well k people (scientistsincluded) trust information it is congruent with their a priori expectations theories.(26*2"The reaction of our lay respo evidence of carcinogenicity is consistent with if we can assume that laypeople expect most c to be carcinogenic in humans. The credibili public respondents gave to evidence of carc is also consistent with a basic tenet of "regul ence which gives much greater weight to posi than to negative findings.(=) The confidence icologists in extrapolation from animal stud appear to be influenced by a positive finding ogenicity). 3.2. Additional Data: The New Jersey Study ": Further evidence that the public perceptions titudes observed in the present study are widespre from a telephone survey of 500 residents in Jersey communities conducted in November 19 Wein~tein.`~T~h)is survey produced strong evide dose-response insensitivity and an "all-or-none' ception of toxicity. Specifically, 78.4% of the reSpae dents agreed strongly (and 6.6% agreed moderately) &I the statement: "If even a tiny amount of a cancerducing substance were found in my water, I woul&'t drink it." The statement: "When some chemical is dis- covered in food, I don't want to hear statistics, I US want to know if its dangerous or not" elicited shnbg agreement from 62.0% and moderate agreement fros 21.6%of the respondents. The New Jersey responddS lo Many policy-makers have argued that lifetime risks equal to 0 r - b than 1 in 1 million are de minimus-too small to worry about ffr regulate(u) The present results suggest that many laypersons w d d object to this view. Intuitive Toxicology 229 also recognized limitations in the knowledge of experts. The statement, "Although experts are willing to make estimatesof the risk from hazardous waste, no one really hows how big the risks really are," elicited 84.9% agreement. About 78% agreed with the statement: "For a lot of chemicals, we don't know enough to make de- cisions." As in the other studies described earlier, women held more negative attitudes than men about chemical risks and more highly educated individuals were less likely to see chemical risks as "all-or-none" phenomena. 3.3. Understanding the Origins of Public Attitudes siod Concerns What accounts for the public's insensitivity to dose and strong concerns about even minute exposures to toxic or carcinogenic chemicals? Psychological and anthropologiarl research on the concept of "contagion" may provide one possible explanation. FrazeP) and dffcn'bed a belief, widespread in many cultures, that things l&%ave been in contact with each other may influence eat$ other through transfer of some of their properties via essence." Thus,"once in contact, always in contact," wen if that amtact (exposure) is brief. Rozin et Sbm that this belief system, which they refer to as a "law d amtagion," is common in our present culture. The implieation of this work is that even a minute amount of a bxic substance in one's food will be- seen as imparting taxiCity to the food; any amount of a carcinogenic substance will impart carcinogenicity, etc. The "essence of harm" that is contagious is typically referred to as contamhlion. Being contaminatedclearly has an all-or-nonequal~ty it-like being alive, or pregnant. When a young child drops a sucker on the floor, the brief cantact with "dirt" my be seen as contaminating the candy, causing the par- ent tothrow it away rather than washing it off and returning a to the child's mouth. This all-or-nonequality irrespective Of the degree of exposure is evident in the observation by - ..Erikson(33)that "To be exposed to radiation or other toxins is to be contaminated in some deep and lasting way, to feel dirtied, tainted, corrupted" @. 122). A contagion contamination model is obviously very different fiom the scientist's model of how contact with a chemical in- dUces carcinogenesis or other adverse effects. Further exakation of the concepts of contagion and contamination may help us better understand the origins of the public's m m s about very small exposures to chemicals. . Alternatively, the intuitive toxicologist may conceive of a toxic chemical as equivalent to a highly dangeiouS virus or bacteria such as botulism, salmonella, Or hepatitis. Early and influential writers such as Rachel Carson(34)may have spawned this perception by drawing analogies between carcinogens and infectious diseases and referring to epidemics of cancer." Exposure to the regulatory philosophy that there is no safe dose of a carcinogen may also have contributed to the public attitudes we observed here. In any event, it is clear that we need to probe further to understand the determiners of public attitudes and the mechanisms by which laypeople believe chemicals induce various kinds of harm. 3.4. Disagreement Among Toxicologists Among the most important findings in this study was the great divergence of opinion among the toxicologists themselves about fundamental issues in risk assessment and, in particular, the high percentage of toxicologists who doubted the validity of the animal and bacterial studies that form the backbone of their science. These results provide a quantitative description of the criticisms and disagreements that are clearly evident in statements of individual scientists(*.") and in detailed case studies of specific chemical controver~ies.(T~h)ese results also clash with the messages given the public. For example, the American Chemical Society's carefully written public information pamphlet on "Chemical Risk: Personal Decisions" informs people that "toxic effects found in rodents usually occur in humans exposed to the same chemical." Although acknowledging the fact that prediction to humans is not certain, the pamphlet concludes that "they are the best predictors we have."12 The impression given the public by this authoritative source is, we submit, quite different from the impression gleaned from questions in Category 2 of our survey or from the many strong criticisms of animal testing in the scientific literature. The affiliation bias we observed is particularly noteworthy, indicating that toxicologists working for industry see chemicals as more benign than do their counterparts in academia and government. Industrial toxicologists were somewhat more confident than other experts in the gen- l1 Carson may have been a major influence behind another popular view of our public respondents-that natural chemicals are not as harmful as synthetic chemicals. In Silent Spring, she observed that although ". . . natural cancer-causing agents are still a factor in producing malignancy, . . ., they are few in number and they belong to that ancient array of forces to which life has become accustomed from the beginning" (p. 195). She contrasts these natural chemicals with synthetic agents of the industrial era against which "man had no protection" (p. 196). `* A survey of other presentations of toxicology designed to educate the public shows that this relatively favorable view of the validity of animal tests is typical. I 7 230 era1 validity of animal tests-except when those tests derstanding of these concepts, and their relationship provided evidence for carcinogenicity-in which case exposure, significant progress most certainly will hw many of the industrial experts changed their opinions. been achieved. An earlier study of 136 occupational health spe- However, the present results also suggest that cialists by Lynn('@ obtained remarkably similar data demonstrating the effect of affiliation on opinions about the value of animal data and the existence of thresholds acontroversies over chemical risks in our society may k fueled as much by weaknesses in the science of assessment as by misconceptions of the public. Our 60 for carcinogens. She found that only 27% of respondents (and those of Lynn) indicate that scientists cannot agrcx working for industry agreed that "a substance which is whether or not their main method of testing is r e l i a shown conclusively to cause tumors in animals should or accurate in predicting human health effects and be considered a carcinogen, thereby posing a risk to humans," whereas 69% and 52% of government and scientists who work for industry see fundamental i m aquite differently from their colleagues in academia university employees agreed. Eighty percent of the in- government. dustry respondents agreed that thresholds exist, in con- These results place the problems of risk commp trast to 61%of the academics and 37% of the government nication in a new light. Our risk management procesopr employees. She also found that these responses were are open and adversarial-we battle in courtrooms correlated with political orientation (1iberaVconserva- community halls, in view of the media, with tive) and with voting in the 1980 presidential election. each side of the issue attacking the other's Finally, Lynn showed that these differences were con- models, and data. The young science of risk assess- gruent with the public stances of industrial and regula- is too fragile, too indirect, to prevail in an a d v e w tory representatives in a number of major risk-management atmosphere. As shown in surveys by Weinstein(* controversies. Data from the present study, and the data Mitchell,(") the public is well aware of the l i m i t a w presented by Lynn, contribute to a substantial literature of expertise in risk assessment. Risk assessment, documenting the strong influence of the social and po- invaluable to regulators in the design of mana litical context on risk a s s e ~ s r n e n t . ( ~ . l ~ - ' ~ . ~ ~ ~ ~ . ~ )strategies, is not at all convincing to the publi Some reviewers of this manuscript have argued that this should not surprise us, given the many cri the affiliation differences we and others have observed risk assessment in the literature. Our survey reflect the fact that toxicologistsin industry tend to work that these criticisms are not a minority view. with less harmful chemicals than toxicologists in regu- iation bias we and others have observed is a natu latory or academic settings and thus have a lower base- growth of the scientific ambiguity-but a disturb rate expectancy for chemical risks. We cannot rule this nonetheless. It feeds the public sense of distrust. interpretation out, although, in response to the statement Risk communication cannot take place wi "Most chemicals cause cancer," industrial toxicologists propriate messages to communicate. The mess in our study did not disagree more than other toxicolo- anating from the risk-assessment commun gists (96.4% of all toxicologists disagreed or strongly contradictory and confusing. Do animal studies disagreed with this statement). The weight of evidence timate cancer risks to humans because of the use seems to us to favor a sociopolitical interpretation of the ticularly sensitive strains of animals or because o affiliation effect. effects of cell proliferation caused by extreme do do they underestimate the risks because of the genetic diversity in the human population, the 3.5. Limitations of Risk Assessment rate of metabolism in humans, or the exposure of hu- mans to a greater number of carcinogenic substances One of the major motivations behind this study was than laboratory animals are exposed to? Are there, ff to develop an understanding of the differences between are there not thresholds for carcinogens?Regulators brnr the ways that professional toxicologists and laypersons sought to justify their actions through reliance on assess chemical risks, in order to facilitate communica- entific consensus, insulated from considerations of @- tion and perhaps reduce the "gap" between expert and lay views. To a certain extent, some of the present re- ". ..icy and politics. Said William Ruckelshaus, at lbc beginning of his second term as EPA administrator: sults do point toward concepts that experts should clarify risk assessment at EPA must be based only on scient& for the public. It is obvious that the words "toxic" and evidence and scientific consensus. Nothing will erdt "carcinogenic" mean very different things to experts public confidence faster than the suspicion that pdis and laypersons. If scientists could impart a better un- considerations have been allowed to influence the as- Intuitive Toxicology 231 m m e n t of risk" (Ref. 37, pp. 1027-1028). Outside of the regulatory arena, many of the scientific foundations and findings of risk assessment are under dispute and appear to be shaped by political factors. What, then, is the message that should be communicated? Risk assessment has been oversold because of the need to rationalize decisions about chemicals. Even Ruckelshaus came to admit this toward the end of his second term. The attempt to quantify risks to human health and the environment from industrial chemicals is, he said, "Essentially . . . a kind of pretense; to avoid the paralysis that would result from waiting for `definitive' data, we assume that we have greater knowledge than scientists actually possess and make decisions based OR those assumptions" (Ref. 38, p. 26). The challenge to toxicologists and risk assessors is clear. These scientists must look anew at the strengths and limitations of their craft. They should work to develop stronger, more definitive ways to assess chemical risks. They should play a greater role in interpreting the *,health implications of their data for the public. In doing they should acknowledge the subjective elements, jadgments, and assumptions inherent in their analyses, aswe11 as the degree of uncertainty in their conclusions. Abwe all, they must protect the young science of risk iimessment from being misrepresented, misused, and ;Ibused in the regulatory process. We concur with Gra- b et a1.(9) in the conclusions they drew from their analysis of the intense controversies surrounding efforts to regulate benzene and formaldehyde: We believe that a precise and honest view about the role of .science in chemical regulation will strengthen both science and democracy. Modest expeaations .. can be realized, and that mans strengthened public confidence in science in the long run. Modest expectations for science also foster and legitimize cxplicit political discussion in our democracy about how to cope with chemical hazards. @. 218) ACKNOWLEDGMENTS Primary support for this research came from grant sm-8722109 from the National Science Foundation to &&ion Research. Any opinions, findings, and conclusions or recommendations expressed in this report are those of the authors and do not necessarily reflect the views of the National Science Foundation. We are grateful for additional funding from the Environmental Prot-on Agency, the Risk Science Institute, the Dow Chemical Company, and the Monsanto Chemical Com- pany. The students and faculty of the Risk Assessment Summer School (RASS) of the International Union of Toxicology patiently helped us pretest our questionnaire. Many persons have helped us by providing critiques of our work and other valuable services. Without implying that they concur in our findings and conclusions, we wish to thank Frederick Allen, Patricia Beattie, Vincent Covello, Michael Davidson, Anthony Dayan, Doug Easterling, Ann Fisher, Gary Flamm, William Freudenberg, James Gibson, Bernard Goldstein, John Graham, P.J. (Bert) Hakkinen, Cecelia Hagen, Carol Henry, Fred Hoerger, Daniel Krewski, Mark Layman, Birgitta Lewander, Sarah Lichtenstein, Frances Lynn, David McCallum, Leisha Mullican, A. John Newman, Tom Osimitz, Emil Pfitzer, Peter Preuss, Sarah Spedden, and James D. Wilson. REFERENCES 1. L.Harris, Risk in a ~~~ Socie!~(public opinion survey, Marsh & McLennan, Inc., New York, 1983). 2. C. F. Callis, "Improving the Public Understanding of Science," Environmental Science and Technology 24, 410-411 (1990). 3. G. C. Pirnentel, "Chemistry at a Crossroad," Chemical and En- gineering News 67(18), 53-55 (1989). 4. National Academy of Sciences. Risk Assessment in the Federal Government: Managing the Process (National Academy Press, Washington, D.C., 1983). 5. W. D. Ruckelshaus, "Risk in a Free Society," Risk Analysis 4, 157-162 (1984). 6. E. Efron, TheApocalyptics (Simon & Schuster, New York, 1984). 7. B. N. Ames and L. S. Gold "Too Many Rodent Carcinogens: Mitogenesis Increases Mutagenesis," Science 249,970-971 (1990). 8. D. A. Freedman and H.Zeisel, "From Mouse-to-Man: The Quan- titative Assessment of Cancer Risks," Statirtical Science 3, 3-56 (1988). 9. J. D. Graham, L. C. Green, and M. J. Roberts, In Search of Safefy: Chemicals and Cancer Risk (Harvard University Press, Cambridge, Massachusetts, 1988). 10. T. Malmfors, "Toxicology as a Science," Trendr in Pharmaco- logical Sciences 2 (l), I-IV (1981). 11. J. Marx, "Animal Testing Challenged," Science 250, 743-745 (199o). 12. L. Clarke, "Politics and Bias in Risk Assessment," The Social Science Journal 25, 155-165 (1988). 13. T. M. Dietz and R. W.Rycroft, The Risk Professionals (Russell Sage Foundation, New York, 1987). 14. W. T. Gormley, Professionalism within Environmental Bureau- cracies: ThePolicy Implications of Personnel Choices (Occasional Paper No. 1, University of Wisconsin, Roben M. Lafollette In- stitute of Public Affairs, Madison, Wisconsin, 1986). 15. B. B. Johnson and V . T. Covello, The Social and Cultural Con- srructwn of Risk (Reidel, Dordrecht, The Netherlands, 1987). 16. F. M. Lynn, "OSHA's Carcinogens Standard: Round One on Risk Assessment Models and Assumptions," in B. B. Johnson and V. T. Covello (eds.), The Social and Culrural Construction ofRisk (Reidel, Dordrecht, The Netherlands, 1987), pp. 345-358. 17. A. Mazur, The L$namics of Technical Controversy (Communi- cations Press, Washington, D.C., 1981). 18. G. M. H. Swaen and J. M. M. Meijers, "Influence of Design Characteristics on the Outcome of Retrospective Cohort Studies," British Journal of Industrial Medicine 45, 624-629 (1988). 19. A. S. Whittemore, "Facts and Values for Environmental Toxi- cants," Risk Analysis 3,23-33 (1983). 232 20. L. J. Casarett and J. Doull. Faxicolog: The Bask Science of Poisons (MacMillan. New York. 1975). 21. B. Ames, "Dietary Carcinogens and hticarcinogens." Science 221, 1256-1264 (1983). 22. H. M. Jenkins and W. C. Ward, "Judgment of Contingency Between Responses and Outcomes," Psychofogical Monographs 79 (I, Whole No. 594) (19aS). 23. D. B. McCallum, S. L. Hammond, L. A. Morris, and V . T. Cwello, Public Knowledge and Perceptions of Chemical Risks in Six Comrmrniticr (Report No. 230-01-90-074,U. S. Environmental Protection Agency, Washington, D.C., 1990). 24. P. Slwic, "Perception of Risk from Radiation," in W. K.Sinclair (ed.), Proceedings of the Twenty-Fifrh Annual Meeting of the SwNational Council on Radiation Protection and Measumnents, Vol 11: Radiation Protection F+: lh NCRPar Yms, OJCRP, Bethesda, Maryland, 1990). pp. 73-97. 25. C. Whipple, De Minimus Risk (Plenum, New York, 1987). 26. J. J. Koehler, Judgments of Evidence Quafiq Among Scientists as a Function of Prior BelkJs and Commitments (unpublished doctoral dissertation, University of Chicago, Chicago, Illinois, 1989). 27. L. Ross and M. R. Lepper, "The Persewerance of Beliefs: Empirical and Nonnative Considerations," New Directionsfor Methodology of Social and Behavwml Science 4, 17-36 (1980). 28. U. S. Environmental Protection Agency, "Guidelines for Carcin- Kraus et d. t t ogen Risk Assessment," Federal Register 51, 33,992-34.005 (1986). 29. N. D. Weinstein, Attituaks of rhe Public and the Depamnenr of Environmental Protection Toward Environmental Hazards (Fm Report, New Jersey Departmentof Emironmental Protection, 1988). 30. J. G. Frazer, The New Golden Bough: A Study in Magic and Religion (MacMillan, New York, 1959;original work publis& in 1890). 31. M. Mauss,A Geneml Theoryofhfagic (Norton, New York, 19n; original work published in 1902). 32. P. Rozin, L. Millman, and L. Nemeroff, "OpeTation of the Laws of Sympathetic Magic in Disgust and Other Domains," J m of Personuldy and Social Psychologv 50, 703-712 (1986). 33. K. Erikson, "Toxic Reckoning: Business Faces A New Kind of Fear." Harvard Business Review 118-126 (1990). 34. R. Carson, Silent Spring (Houghton Mifflin,New York, 1962; quotationsherein taken from the paperback version, Fawcett, 1969). 35. S. Jasanoff, Risk Mangemenr and Political Culture (New Yo&, Russell Sage Foundation, 1987). 36. J. V. Mitchell, "Perception of Risk and Credibilityat Toxic Sires," Risk Anabsis 12,19-26 (1992). 37. W.D.Ruckelshaus, "Science, Risk, and Public Policy," Sciem 221, 1026-1028 (1983). 38. W. D. Ruckelshaus, "Risk, Science, and Democracy," Isnrcsin Science and Technology 1 (3), 19-38 (1985). 1 E t