Philip Morris
Communicating Risk Under Title III of Sara: Strategies for Explaining Very Small Risks in A Community Context
Fields
- Author
- Fisher, A.
- Mcclelland, G.H.
- Schulze, W.D.
- Mcclelland, G.H.
- Area
- LOGUE,MAYADA/OFFICE
- Type
- PSCI, PUBLICATION SCIENTIFIC
- BIBL, BIBLIOGRAPHY
- CHAR, CHART, GRAPH, TABLE, MAPS
- DRAW, DRAWING
- BIBL, BIBLIOGRAPHY
- Site
- N426
- Named Organization
- Apca 88 Meeting
- Epa, Environmental Protection Agency
- Request
- Stmn/R1-072
- Document File
- 2025545619/2025546382/Harvard University Office of
- Continuing Education Short Course Program Harvard School
- of Public Health
- Continuing Education Short Course Program Harvard School
- Master ID
- 2025545673/6381
Related Documents:- 2025545673-6381 Risk Analysis in Occupational and Environmental Health 910904 - 910906
- 2025545684 Telephone Locations and Protocol
- 2025545689-5696 Risk Assessment for Carcinogens: A Comparison of Approaches of the Acgih and the Epa
- 2025545697 Hps Newsletter Interview with A Risk Expert
- 2025545698-5711 Science and Its Limits: the Regulator's Dilemma
- 2025545713-5721 Risk / Benefit Analysis
- 2025545722-5725 Risk Management Commentary for Dr. D. Allan Bromley Assistant to the President for Science and Technology
- 2025545726-5729 Risk Assessment and Comparisons: An Introduction
- 2025545750-5792 Risk Assessment of Chemical Carcinogens: Is It Time for A Change?
- 2025545795-5799 Tools of Risk Analysis Applications of Epidemiology
- 2025545800-5810 Notice of Intended Changes - Benzene
- 2025545811-5822 Epidemiology in Risk Assessment for Regulatory Policy
- 2025545824-5850 Risk Analysis in Environmental and Occupational Health Use of Animal and Other Data As Predictors of Human Risk
- 2025545851-5871 Risk Analysis in Environmental and Occupational Health Uncertainties in Predicting Human Risks
- 2025545872-5881 How Do Cancer Risks Predicted From Animal Bioassays Compare with the Epidemiologic Evidence? the Case of Ethylene Dibromide
- 2025545882-5887 Use of Biological Assays in Short-Term Assessment of Inhaled Substances
- 2025545888
- 2025545889-5891 Risk Analysis in Environmental and Occupational Health Are Your Mushrooms Safe to Eat?
- 2025545892-5899 the Rat As An Experimental Animal
- 2025545901-5907 Non-Cancer Endpoints
- 2025545910-5939 Cancer Facts & Figures - 890000
- 2025545940-5941 Cancer Facts & Figures - 890000
- 2025545942-5944 Get - the - Lead - Out Guru Challenged A Decade-Old Scientific Argument Over the Effects of Low-Level Lead on Iq Turns Nasty Following Allegations of Misconduct
- 2025545945-5948
- 2025545949-5958 the Question of Thresholds for Radiation and Chemical Carcinogenesis
- 2025545959-5980 Are There Thresholds for Carcinogenesis? the Thorny Problem of Low-Level Exposure
- 2025545981-5990 Perspectives on Comparing Risks of Environmental Carcinogens
- 2025545991-5998 Acceptable Cancer Risks: Probabilities and Beyond
- 2025546000-6011 Ideas in Pathology Pivotal Role of Increased Cell Proliferation in Human Carcinogenesis
- 2025546012-6017 Cell Proliferation in Carcinogenesis
- 2025546019-6027 the Role of Expert Judgement in Risk Analysis
- 2025546029-6039 the Respiratory Tract As A Route of Exposure
- 2025546040-6045 the Respiratory Tract As A Portal of Entry for Toxic Particles
- 2025546047-6062 Limitations to the Use of Employee Exposure Data on Air Contaminants in Epidemiologic Studies
- 2025546063-6083 Benefit - Cost Analysis of Environmental Regulation: Case Studies of Hazardous Air Pollutants
- 2025546086-6089 Legislative and Regulatory Aspects of Risk
- 2025546090-6099 Connecticut's Dioxin Ambient Air Quality Standard
- 2025546100-6103
- 2025546105 Annals of Radiation Calamity on Meadow Street
- 2025546106 Caution Urged When Using Insect Repellents
- 2025546116 Volatile Organics and Inorganics Action Levels 900400
- 2025546134-6135 Summary of Radon Test Results of the Household Testing Program
- 2025546141-6145 Introduction to Discussion Sessions
- 2025546146-6149 Risk Assessment in Environmental and Occupational Health Risk of Alar (Daminozide)
- 2025546150-6160 Intolerable Risk: Pesticides in Our Children's Food
- 2025546161-6162 Pesticides, Risk, and Applesauce
- 2025546163-6168 Daminozide Special Review Technical Support Document - Preliminary Determination to Cancel the Food Uses of Daminozide
- 2025546169 Daminozide / Udmh
- 2025546170-6172 the Relative Risk of Daminozide (Alar / Kylar) Use
- 2025546173 Be Most Wary of Nature's Own Pesticides
- 2025546174-6175 A Movie Star Pares the Apple Industry
- 2025546176-6183 Summary of Toxicology Data on Daminozide and Udmh
- 2025546184-6194 Attachment I Graphs of Data From NCI / Ntp 83 Daminozide
- 2025546195-6196
- 2025546197-6202 Daminozide Special Review Technical Support Document - Preliminary Determination to Cancel the Food Uses of Daminozide
- 2025546203-6224 Regulatory Decision - Making Under Uncertainty: the Case of Alar
- 2025546226 Epa Moves to Reassess the Risk of Dioxin Urged on by the Scientific Community, Epa Is Developing A New Model for Estimating Dioxin's Risk
- 2025546227 US Government Orders New Look at Dioxin the Environmental Protection Agency Is Evaluating Data From the Past Decade That Suggest Dioxin's Toxicity May Be Overestimated. A Risk Assessment Model Based on Biological Mechanism Is Being Drawn Up.
- 2025546228-6235 Dioxin Toxicity: New Studies Prompt Debate, Regulatory Action New Data on Dioxin's Effect on Humans, A Clearer Picture of the Cellular Events It Precipitates, and New Animal Toxicity Studies May Provide Epa with A Firm Basis for Regulation
- 2025546236-6250 the Regulation of Gene Expression by 2,3,7, 8-Tetrachlorodibenzo-P-Dioxin
- 2025546251-6253 Dioxin Risks Revisited Armed with A New Understanding of How Dioxin Works on the Molecular Level, A Number of Scientists Are Challenging Epa to Change the Way It Does Risk Assessment
- 2025546255-6258 Lead Toxicity Case Study for Short Course on Risk Analysis in Occupational and Environmental Health 910904 - 910906
- 2025546259-6267 Lead
- 2025546268-6275 Lead in Bone: Implications for Toxicology During Pregnancy and Lactation
- 2025546276-6281 the Long-Term Effects of Exposure to Low Doses of Lead in Childhood An 11 - Year Follow-Up Report
- 2025546282-6285
- 2025546298-6321 Review 890000 Alice Hamilton Lecture Lead and Human Health:Background and Recent Findings
- 2025546323-6348 Traps and Errors in Risk Analysis
- 2025546349-6356 Health Risks the Perception of Reality and the Realty of Perception
- 2025546363-6368 Industrial Risk Perceptions
- 2025546369-6370 Too Many Rodent Carcinogens: Mitogenesis Increases Mutagenesis
- 2025546371-6373 Has Risk Assessment Become Too 'conservative'?
- 2025546374-6378 Health and Safety Risk Analyses: Information for Better Decisions
- 2025546379-6381 Telling Reporters About Risk Dealing with Reporters Needn't Be the Least Agreeable Part of the Job.
- Author (Organization)
- Epa, Environmental Protection Agency
- Journal of the Air + Waste Management As
- Office of Policy Analysis
- Univ of Co Boulder
- Journal of the Air + Waste Management As
- Litigation
- Stmn/Produced
- Characteristic
- EXTR, EXTRA
- Date Loaded
- 24 May 1999
- UCSF Legacy ID
- nmp02a00
Document Images
Comimunicating Risk Under Title 111 of SARA:
atrategi:es for Explaining Very Small Risks in
a Community Context
Ann Fisher
U.S. Environmental Protection Agency
office of Policy Analysis
Washington, D.C.
Gary H. AAcCleiIand and Wilfiam D.SchuIze
University of Colorado
Boulder, Colorado
Under Title ][II of SARA, companies must provide
information about chemicals that they manufacture,
store, or process. Communities will use data about
potential accidental releases to develop local emergen-
cy plans. Dataa about routine chemical releases will be
made available to the' public on a computer data base.
Simply having such data available does not ensure
consensus about reducing potential chemical risks.
Laboratory and field research are summarized, indi-
cating that people tend to edit small risks to zero as
being too small to worry about, or to adjust them
imperfectly Zrom an anchor equal to the potential loss.
These results suggest recommendations for communi-
cating about the risks posed by accidental or routine
releases of chemicals.
Title III of the Superfund Amendments and Reauthoriza-
tion Act of 19$6 (SARA) is also called the Emergency Plan-
ning and Community Right-to-Know Act.' Its purpose is to
facilitate informed public participation in decisions about
chemical risks--at the community level where these risks
occur. Companies must provide information about hazard-
ous and toxic chemicals that are present in their facilities as
part of their manufacturing, storage, or processing activities.
Title III covers both accidental and routine releases.
The availabil:ity of data per se does not ensure that every
community will reach an easy consensus regarding what-if
anything-should be done about the potential risks posed by
these chemicals, Rather, different groups can be expected to
have opposing reactions to this information. Apathy and/or
denial may characterize one group's responses to this ava-
lanche of information about the presence of chemicals and
their routine releases. A;typical resident might say: "This is
information about chemicals that have been present in my
community for years. Besides, safety practices and regula-
tions now in place have reduced the amounts of these chemi-
cals that get into the environment. I don't know anyone who
got sick from these chemicals, and the companies using them
provide lots of jobs here. I don't need this information,
especially because it might lower the value of my property."
Public officials can be discouraged by this type of response
because the Title III data can be used to protect communi-
ties from substEmtial risk in specific situations. For example,
From: The Journal
of the Air & Waste
Management Association
(JAPCA), Vol. 39,
No. 3 (Mar. 1989)
the data could be used in a Local Emergency Plan to indicate
that the appropriate evacuation routes depend on wind di-
rection in case of an accidental release into the air; if resi-
dents are not aware of this, they may use evacuation routes
that carry them into a pollution plume, rather than away
from it. Another example is the case of communities where
all companies are complying with their emission permits,
but where the combined effect of these emissions may create
potential "hot spots" in terms of annual emissions. If the
residents ignore the Title III data, they may lose an opportu-
nity to negotiate for changes that could reduce their poten-
tial exposure.
Public officials and business firms also are concerned
about potential misinterpretation or even deliberate misrep-
resentation of chemical data. For example, the routine re-
lease data is reported in pounds per year, comparing 35,000
pounds of chemical X to 10,000 pounds of chemical Y may
give the impression to the community and the media that
chemical X represents a larger problem than chemical Y.
However, this impression ignores important factors such as
the comparative toxicities of X and Y and whether the re-
lease is likely to result in exposure. Special interest groups
also could play on the apprehension that might be created by
the sheer size of the numbers associated with the units in
which the data must be reported (e.g., reporting 10,000
pounds may be far more frightening than the same informa-
tion expressed as 5 tons). Public concern about the large
number of pounds could lead to pressurk~ for reducing emis-
sions of substances much less likely to harm the community
than smaller quantities of chemicals that are more toxic or
more likely to result in exposure. Such considerations may
result in a secondgroup of citizens becoming overly con-
cerned about some chemicals. The behavior of this latter
group contrasts dramatically with that of the apathetic
group.
When both the apathetic and concern reactions occur in
the same community, there is likely to be conflict about
interpreting risks revealed under Title III as well as about
other risks. In this work, we use relevant research results to
derive policy recommendations for communicating risks
posed by either accidental or routine releases of chemicals in
a community. The main objective of these recommendations
is to assist government officials and members of Local Emer-
gency Planning Committees (LEPCs) as they help citizens
put the risks in context, that is, to raise community aware-
ness of the larger risks without causing undue concern about
the smaller ones.
Copyright 19fq-Air & Wdte Mnnasement Aa®ocintion
March 1989 Volume 39, No. 3 271

Fiyi"1. A maJOf; of rWc ludgoment cons}stent whh birtadal distributions oi
percefved risk.
Sara Title 111--IBackgroaend
Title III relies primarily on local planning and action.
State coai!si:ons have been appointed to establish LEPCs,
which use dat& relevant to potential accidental releases for
preparing local emergency plans. The LEPCs must include
elected state srid local off icials; policy, fire, civil defense, and
public health professionals; haspital and transportation offi-
cials; as well as representatives from industry, community
and environmental groups, and the media.2
Companies must participate in emergency planning if
they have more than published threshold quantities of 366
substances lusta:d as extremely hazardous.3 They must notify
the LEPCs Fmd their state commissions about releases of
these chemicgls that are above specified quantities. Compa-
nies must also submit information regarding inventories of
hazardous chemicals to the state commissions, LEPCs, and
local fire departments. They are not required to submit this
information to the U.S. Environmental Protection Agency
(EPA). The law became effective on October 17, 1987.
Prior to indtiation of Title III, EPA had a Community
Emergency Prisparedness Program, which relied on volun-
tary submission of information so that communities could
plan for chemiiVal emergencies 4 Elements of this EPA pro-
gram were incorporated into the Title III legislation. EPA
and other federal agencies provide guidance and training for
helping the l:.lTCs and state commissions cope with the
deluge of information from companies that are required to
report under Title III. For example, the National Response
Team, which ifs composed of 14 federal agencies, has pub-
lished guideline:s for emergency response planning.5 The law
required that the emergency plans be ready by October 17,
1988; they must be reviewed at least annually.
Data about routine releases of over 300 listed chemicals
must be submitted annually both to the states and to EPA,
beginning July 1, 1988. The threshold quantity of release for
reporting purp>ses is 10,000 pounds per year for facilities
using a listed chemical. Facilities manuf acturing or process-
ing a listed chemical have annual reporting thresholds that
decrease to 25,000 pounds per year by July 1, 1990. These
data will be made available to the public by EPA through a
computerized '.Coxics Release Inventory. States also will
make the routine release data available.
The Toxics Release Inventory will provide information
regarding how much of each listed chemical is released from
the facility into the air, water, and land. The quality of the
data is expected to be variable, because there are no moni-
toring requirements in the legislation. A significant concern
is that this data base will not have information that relates
the emissions information to the likelihood and potential
consequences of exposure. There is little overlap between
the chemicals that must be reported for emergency planning
and those that must be reported for routine release. This is
mostly because concerns about emergency releases (e.g., ex-
plosions, fires, acute health effects) often accompany differ-
ent chemicals than the chemicals associated with chronic
health and environmental effects that are of concern from
routine releases. EPA is in the process of developing fact
sheets that describe what is known about the consequences
of exposure. The agency also is preparing a personal comput-
er version of the Graphical Exposure Modeling System
(GEMS), which is a model for combining routine emissions
information with geographic characteristics to predict po-
tential exposures s
The legislation does not specify that LEPCs are responsi-
ble for interpreting the routine release data. However, the
public may turn to the LEPCs, local and state health and
environmental offices, state commissions, and even EPA
officials, to help them understand the implications of Title
III information. EPA recognizes that the LEPCs have the
potential to be a community focus for managing both emer-
gency and routine release risks under Title III.
DffiflcuRles In Understanding Community Chemical Risks
To make recommendations regarding communication
about chemical risks in a community, it is necessary to un-
derstand how people form beliefs about risks associated with
chemicals and how these beliefs change. Figure 1 shows our
model of risk judgement as a first step for explaining how the
same risk information can lead some people to dismiss a risk
as too small to worry about while others view the risk as a
threat to themselves, their family, or their property. Several
factors may affect whether a person worries about a particu-
lar risk. The first part of this section describes some of the
empirical evidence supporting the model in Figure 1. The
second part describes how various factors mayy influence
possible outcomes under the proposed model.
A AAodel of Risk Judgement
Substantial empirical evidence indicates that people have
difficulties evaluating small probabilities. McClelland et al.
used laboratory experiments to demonstrate that subjects'
bids (i.e., the amount they were willing to pay) for insurance
against a loss were approximately equal to the expected
value of the loss-as predicted by economic theory7,8-for
probabilities of loss greater than approximately 0.1.9 Howev-
er, subjects consistently bid more than the expected value of
insurance for smaller probabilities of loss.
A more detailed examination of the results from the
McClelland, et al. low-probability risk experiments is shown
in Figure 2a.9 Economic theory predicts that people will bid
the expected value of insurance for a particular risk, so that
the ratio of bid to expected value of the insurance would be 1.
However, in this experiment there were more bids at twice
that ratio, and a substantial number at four times the ex-
pected value. In addition, many of the subjects bid zero for
the insurance against a small probability loss. The results
indicate that the distribution of the ratio of bids to expected
value is bimodal.
A similar pattern can be seen in Figure 2b, which repre-
sents community beliefs about the risks associated with a
Superfund site located in Monterey Park, California.9 As in
the laboratory experiments, a substantial share of residents
in the community judged the risk to be zero, while approxi-
mately 30 percent perceive the risk to be as high as one in one
hundred. This is much higher than the scientists' estimates
of potential risk from the Superfund site. For example,
EPA's risk estimates imply an upper bound on nearby resi-
dents' risk of cancer from vinyl chloride of 1.67 X 10'4.
Although the results of these two case studies need further
confirmation, they do suggest that the factors resulting in
272
JapcA

I
bimodal distributions of perceived risk may be the same in
the laboratory ex periments and in an actual situation.
Bimodal distributions of risk perceptions may be ex-
plained by two cognitive processes: 1) dismissalto11 and 2)
anchoring and adjustment.12_l4 An intuitive explanation for
these processes is that individuals confront so many low
; probability risks that it is impossible to develop an appropri-
ate response for each one on the basis of analytical evalua-
tion. One coping strategy is to dismiss those risks that are
perceived to be below some threshold (i.e., the left side of the
risk judgement model presented in Figure 1). In the McClel-
land et al. insurance experiments previously cited, fewer
people bid for insurance as the probability of loss falls, so the
amount of dismissal increases. For those who do think the
risk is large enough to evaluate (i.e., the right side of Figure
1), the problem is how to decide on an appropriate level of
concern (i.e., how much to bid to protect against a loss in the
insurance experiments). The model indicates that people
first anchor on the loss. That is, they focus on the magnitude
of the potential loss. Then they adjust their concern (or bid
for insurance) downward to_reflect the fact that the loss will
occur only some of the time. The cognitive psychology litera-
ture indicates that such adjustments nearly always are in-
complete.13.1s.1s In the context of the insurance experiments,
the concept of incomplete adjustment can be used to explain
why the bids for nisurance end up being larger than expected
value (for respondents in the upper mode of the bimodal
distribution).
For the previously cited insurance experiments, these cog-
nitive nitive processes fi)r forming a risk perception can be shown
; in a simple equatilon:
B=L--i;1-e)(L-pL)=pL+e(L-pL) (1)
where: B = the bid for protection against loss
L = the loss if the hazard occurs
p = the probability of loss, and
e = the adjustment factor
The equation in d.icates that people anchor on the potential
loss L and adjust this amount toward the expected value of
the loss, pL. An expected value model would predict the
adjustment to be i;L - pL), with e = 1. However, the term (L
- pL) is modified by (1 - e)'because the adjustment is
incomplete. Us'vn@; the data from the insurance experiments,
the model predicts the underadjustment factor to be only 2-
3 percent 9 This error still distorts responses significantly for
low probabilities because the difference between the anchor,
L, and the expected value of the loss, pL, is very large for low
probabilities. For example, if L = 100 and p = .01 then (L -
pL) = 99. If the underadjustment factor is 2 percent, then B
= 2.98. Compared with the expected value pL = 1, this
implies an adjustment error of 1.98. As the size of the adjust-
ment needed becomes larger, so does the adjustment error
(e.g., if L = 1000, (L - pL) = 990, and e(L - pL) = 19.8). The
adjustment error seems especially large compared with the
expected value pL, which will be small for low probabilities.
Given the bimaiality that is likely to occur in a communi-
jty's perceptions of low-probability chemical risks, the best
strategy for the LEIPC (or other responsible group) may be to
help people approach the more appropriate mode of either
"dismissal" or "concern," while recognizing that neither
[mode may be accurate. In order to select the most appropri-
ate mode, the LEPC could use data provided under Title III,
additional information about whether those releases might
lead to exposure, aand dose-response data. The risk commu-
nication strategies needed to help people get into a concern
mode may differ f rom those needed to help them get into a
dismissal mode.
The judgemen t of whether the concern mode is more ap-
propriate than the dismissal mode is not a trivial issue. The
true size of the risk cannot be determined because of the
uncertainties associated with various steps of the risk esti-
March March 1989 Volume 39, No. 3
mation process. Kisk assessments typically yieia estimates
of individual risk and estimates of the total number of peo-
ple affected. However, other characteristics of risk also are
important to the public. This makes it likely that there will
be an element of value judgement in the LEPC's (or other
responsible group's) decision about which mode is more ap-
propriate.
Detertninants of Dismissal versus Concern
Several factors may influence whether people dismiss a
risk or evaluate it. Some of these factors are discussed below.
Framing of gains and losses. In their description of pros-
pect theory, Kahneman and Tversky indicated that it is
important to determine whether the risk being communicat-
ed will be viewed by community residents as an increase or a
decrease in their level of risk.lo,17 People are more concerned
about losses than about gains relative to the status quo. This
means that a perceived increase in risk (a loss) will have a
greater psychological impact than the same size reduction in
risk (a gain). In common sense terms, going from thinking
one is "safe" to believing one is "unsafe" makes an individ-
ual comparatively unhappier than going from thinking he is
"unsafe" to believing he is "safe" makes him happier.
Because most community members probably are unaware
of potential risk that must be reported under Title IIl, the
data are likely to be viewed as a new risk and a loss in well-
being. Thus, the risk is more likely to be evaluated than
dismissed, and it is likely to be weighted more heavily be-
cause it is viewed as a los& If the community is judged to
need help getting into the dismissal mode, these consider-
ations suggest that expressing risks in terms of the probabili-
ty that "there will not be an accident" or that "there will not
be adverse health effects" may generate less concern than
expressing the risks in terms of the probability that "there
will be an accident" or that "there will be adverse health
effects." The reverse would be true if the community is
judged to need more concern.
Another framing issue is the quantitative expression of
risk. Although people have difficulty understanding low-
probability risk, some results of the insurance experiments
indicated that bids converged toward expected value when
the risk was expressed as an aggregate across several time
periods.ls Subjects were told both that the probability of loss
on any given round was 0.01, and that this meant the proba-
bility of at least one loss across 25 rounds was about 0.25.
The resulting bids (to protect against any loss for the block
of 25 rounds) showed less of a bimodal distribution, and they
were closer to the expected value.
These results suggest that it may be effective to express
risks in terms of a longer time frame, such as a lifetime, at
least for annual risks in the range of 10-2 to 10-3. This
strategy is less likely to succeed for smaller risks because the
risk aggregated over an individual's lifetime still is smaller
than the range of probabilities that most people understand.
However, expressing aggregate (lifetime) risk to the neigh-
borhood or community might have large enough probabili-
ties to accomplish better understanding. For example, an
individual lifetime risk estimate of 10-; could be explained
as one expected death over 70 years in a community of 10,000
people.
Experience. The amount and nature of prior experience is
an important determinant of how much concern individuals
will have about a risk. Risks that are familiar, for which the
science is understood, and with which they have had prior
benign experiences are more likely to be dismissed. Risks
that are unfamiliar, not well understood, and for which there
are no perceived benign experiences are more likely to gener-
ate high levels of concern.19 For example, across 50 rounds in
the insurance experiments using a probability of 0.01, the
share of people in the concern mode dropped steadily with
273

benign egp,erience until the adverse event actually occurred
on the 33rd round 9 Then there was a sharp drop in the
fraction of subjects in the concern mode, reflecting the
gambler's fall'.acy that a low-probability event is less likely on
the next round because it occurred on the previous round.
During succeeding rounds, the share of subjects in the con-
cern mode grew as fewer and fewer people felt comfortable
dismissing tbe risk.
Many communities will recall only benign experience rele-
vant to Title III, and tend to be in the dismissal mode. But in
communities where there has been a chemical accident or an
emergency release, a high share of the population may be in
the concern mode.
Characteriatics of the risk. Technical risk assessment
identifies which adverse effect could occur and estimates its
probability of occurrence and the number of people expect-
ed to be affeci`ed. These are the only parameters included in
a risk assessrr ent. However, individual's beliefs about other
factors may nifluence whether they dismiss or express con-
cern about a particular risk. There are several important
characteristics of risk that cause people to have more con-
cern?°,21 The :more serious and dramatic the consequences of
a risk, the higher will be the anchor in the anchoring and
adjustment process, so the final level of concern will be
higher. Risks that are dreaded, that can affect many people
at one time, and that are considered to be unfair or morally
wrong tend to result in higher concern.19
Personal characteristics. There is some evidence to indi-
cate that personal characteristics affect risk perceptions. For
example, people with'more education, who are white, and
who tend to ask a doctor a lot of questions or read regularly
about health have less concern about radon.22 Families with
children, relatively young people, and women all tend to be
more fearful of Superfund sites (and Superfund sites contain
many Title III chemicals).9 For the Superfund sites included
in this study, education, income level and occupation were
not found to have an impact on risk beliefs of people living
nearby, however.
Media attention. The need to maintain ratings or circula-
tion gives the media an incentive for sensational coverage,
especially when there is public controversy. Media coverage
is likely to focus on those factors that encourage evaluation
and lead to concern (e.g., a story reporting higher cancer
rates in the area). The McClelland et al. research showed
that frequent exposure to media reports about a Superfund
site was significantly correlated with being in the concern
mode.
Physical reminders. Risk judgements are influenced by
perceptual cues. The more people are reminded of a risk, the
more likely they are to be in a concern mode. Responses from
45 percent of the residents living near a Superfund site
revealed that many of them perceived a dramatic decline in
risk after the site was closed 9 No special closure activities
had. been undertaken to safeguard the community from the
wastes already at the site, but the disappearance of physical
reminders such as trucks and workers on the site may have
been enough to change the community's risk beliefs. In-
creased perceptual cues regarding Title III could come from
sirens and fire trucks signalling the emergency release of a
chemical, or from odors that accompany routine releases.
Chemical releases that are odorless and colorless are less
likely to result in people being in the concern mode.
Reaommendatiosa
1_ 2 4 9 18
Bid for insurance
Expected value of the bss
Figtwo 2a. Diaributlan of subjects' concern obtained from a laboratory
exper;mettt.9
a
n
r
v
Perceived risk of death
b f j
Figure 2b. Distribution of nearby residents' concems about a Superfund
site 9 Annual risk of death: a = no risk; b= one in 9 million; f = one in 100
thousand; 1= one In 10 thousand; n= one in one thousand: r = one in one
txmdred; v= one In ten.
274
The following are recommendations for LEPCs and other
groups that may be asked to interpret Title III data. Some of
them are consistent with risk communication guidelines al-
ready available, but others are new.23
o Identify and address community concerns. Effective
risk communication is crucial if Title III is to lead to
informed local decision making about chemical risks in a
community contezt. Effectiveness requires recognition
that community concerns may not be addressed by the
usual components of risk assessment (e.g., residents may
be worried by odors from a local chemical plant while
experts may know that the odors are harmless, and not
think it important to address the issue in discussions of
Toxics Release Inventory data).
Establish and protect credibility. Individuals communi-
cating risk must be viewed as credible by the community.
The diverse composition of LEPCs should demonstrate
absence of bias toward any particular interest group.
Care should be taken, however, because there are only
limited public resources to support LEPCs' activities.
The interests and available expertise of industry repre-
sentatives on LEPCs may result in their having a large
share of the committee's work. This could be perceived
as self-serving. However, informed review by other com-
mittee members and the LEPC's state commission
should ameliorate such concerns. Other neutral experts
(from local colleges, laboratories, etc.) also may be called
upon to reinforce the risk communication messages.
. Account for typical reactions to low-leuel risks. Because
we observe fairly few fatal chemical accidents and chem-
ical-related illnesses, nearly all of the Title III risks will
have annual odds smaller than one in one hundred.
Therefore, people will have difficulty understanding
these risks and will tend either to dismiss them or to have
a high level of concern about them, potentially resulting
.IAPC:A

in cornrnunity conflict. The LEPC (or other responsible
group) iwill have to decide whether the larger problem is
raising awareness of those who tend to dismiss (so that
they would become less likely to ignore warnings about
actions to take in a chemical emergency) or reassuring
those who believe routine release risks to be larger than
the scientific evidence indicates. It is unlikely that all of
the divergence between those in the dismiss mode and
those in the concern mode can be resolved. This may be
appropriate, though, because some of the community
residenis may be at higher actual risk.
Recognize that characteristics of risk matter. A famil-
iar, well. known, and undramatic risk generates a lower
level of concern than one with the same probability and
consequ ence that is new, poorly understood, and dra-
matic. Even if people are convinced that the probability
DEATH 1/1
Z
Y
X
w
V
U
T
S
R
9
Yearly mortality-p P
rate for smokers 0
Average deaths -~~
due to motor vehicles
N
M
L
K
J
I
H
G
Yearly mortality -: F
rate due to average E
exDosureto X=ravc D
Average chance
of dying in a plane crash
(both public and private)
C
B
A
1/10
1/100
1/1000
1/10,000
1/100,000
1/1A00,000
0
cause some differences in characteristics may cause peo-
ple to reject the validity of the comparisons. For exam-
ple, several voluntary risks are included in Figure 3,
while people in a community may feel that Title III risks
are imposed on them involuntarily. An alternative risk
ladder could be developed, however, with better match-
ing of characteristics of the comparison risks and the
Title III risk.
Treat the media as a legitimate partner. Providing com-
plete and consistent information to the media will mini-
mize the likelihood that they will become catalysts for
inadvertently high levels of concern. In addition, access
to experts can make it easier for reporters to develop an
accurate but interesting story about a risk that the
LEPC views as potentially large but that people are
dismissing.
4 Qimbing Mt. Everest Once
4 Stunt person
~ Sport parachuter
~ Firefighter
4 Mortality rate due to
drinking one saccharine
diet drink a day
4 Average chance of being fatally
struck by lightning
Figttra 3. Example of a risk ladder, indicating risk for one year of expostre unless oftwmse
specified9
and consequences are the same, they still often object
more to a risk that is imposed upon them than to one
voluntarily sought, one that affects many people at once
rather t'!han one;at a time, or one that involves dread.
This ntay indicate that the community really wants more
of its resources devoted to reducing some risks compared
with others that may have a higher probability or affect
more people. Such preferences should be acknowledged
when communicating about risk.
Use comparable ri:sks. Risks should be expressed in con-
crete terms and put in perspective. Several suggestions
are provided in a manual the Chemical Manufacturers
Association has ' developed for plant managers.24 One
approac;h is to match characteristics of the risks posed by
Title III: chemicals with characteristics of other risks
with which people have more familiarity. An example of
using comparable risks is presented in Figure 3, which
shows actual risks associated with various activities.9 A
Title III chemical could be placed on this risk ladder,
next to the corresponding scientific estimate of risk. If
the situation where the risk is being explained cannot
accommodate the time needed to read and understand a
risk ladder, one or two comparable risks can be de-
scribed. Comparisons need to be used with caution, be-
e Account for individuals' characteristics. For example,
people with young families are likely to have higher
concern about risks, especially compared with the elder-
ly. Communications need to be targeted to subgroups,
accounting for ways to reach them as well as making the
message personalized to help them shift into the appro-
priate mode of dismissal or concern.
Disclalmer
This work was presented at the APCA'88 Meeting, Dallas
Texas, June 20-24, 1988. The research was partially funded
by the U.S. Environmental Protection Agency. The views
expressed are the authors' and should not be attributed to
the funding agency or their employers.
References
1. Superfund Amendments and Reauthorization Act of 1986, PL
99-499, or 100 Stat. 1623.
2. Superfund Amendments and Reauthorization Act of 1986, PL
99-499, Section 301(c).
3. "Risk screening guide-interim final, Appendix F," U.S. Envi-
ronmental Protection Agency Office of Toxic Substances,
Washington, DC, 1988.
March 1989 Volume 39, No. 3
275

4. "The Emergency Planning and Community Right-to-Know Act
of 1986: questions and answers," U.S. Environmental Protec-
tioa Agency Emerv
e~acy Planning and Community Right to-
Know Information Hotline, Washingtoa, DC,1988.
5. "The hrm3rdous materials emergency;>iiinning guide," National
Response Team, NRT-1, Washington,1YC,1987.
6. "Risk scr,-ening guide-interim final, Appendix G," U.S. Envi-
ronmental Protection Agency Office of Toxic Substances,
Washington, DC, 1988.
7. J. von NeumAnn, 0. Morgenstern, Theory of Games and Eco-
nomic Be,'tavior, 3rd ed., Princeton University Press, Princeton,
NJ, 1953.
8. T. Tietenberg, Environmental and Natural Resource Econom-
ics, 2nd ed., Soott, Foresman and Company, Boston, 1988.
9. G. H. M:cClelland, W. D. Schulze, D. L. Coursey, B. Hurd, J. R.
Irwin, B;. ]Et. Boyce, "Risk Communication for Superfund Sites:
An Analyris of Problems and Objectives," Draft report to U.S.
Environm.ental Protection Agency, Office of Policy Analysis,
Washin;gGln, DC,1967.
10. D. Kahvensan, A. Tversky, "Prospect theory: an analysis of
decisions ~:mder risk," Econometrica 47: 263 (1979).
11. P. Slovi,c, B. Fischhoff, S. Lichtenstein, B. Corrigan, B. Combs,
"Preference for insuring against probable small losses: Insur-
ance implications," Journal of Risk and Insurance 44: 237
(1977).
12. P. Slovic, "'Influence of the Response Mode Upon Relative Im-
portana,- of Probabilities and Payoffs in Risk Taking," in Pro-
ceedings of the 75th Annual Convention of the American Psy-
chological Association, 1967.
13. A. Tversky, D. Kahneman, "Judgment under uncertainty: heur-
istics aad biases," Science 185: 1124 (1974).
14. H. Einhorn, R. M. Hogarth, "Ambiguity and uncertainty in
pmbabilis :ic inference," Psychological Review 92: 433 (1985).
15. D. C. Poulton, "The new psycbophysica: six models for magni-
tude estimation," Psychological Bulletin 69:1(1968).
16. S. Lichten stein, P. Slovic, B. Fischhoff, M. Layman, B. Combs,
"Judged fiequency of lethal events," Journal of Experimental
Psychology 4: 551 (1978).
17. A. Tversky, D. Kahneman, "The framing of decisions and the
psychology of choice," Science 211: 453 (1981).
18. J. Doyle, G. McClelland, W. Schulze, "Response Variation to
Alternate Framings of Identical Low-probability Risk," Univer-
sity of Colorado draft report, Boulder, Co., 1988.
19. P. Slovic, "Informing and educating the public about risk," Risk
Analysis 6: 403 (1986).
20. D. Litai, "A Risk Comparison Methodology for the Assessment
of Acceptable Risk," doctoral dissertation, Massachusetts Insti-
tute of Technology, Cambridge, Mass., 1980.
21. P. M. Sandman, "Explaining Environmental Risk," U.S. Envi-
ronmental Protection Agency, Washington, D.C., 1986.
22. F. R. Johnson, A. Fisher, "Conventional wisdom on risk commu-
nication and evidence from a field experiment," Risk Analysis,
in press.
23. V. T. Covello, F. W. Allen, "Seven Cardinal Rules of Risk Com-
munication," OPA-87-020, U.S. Environmental Protection
Agency, Office of Policy Analysis, Washington, DC, 1988.
24. V. T. Covello, P. M. Sandman, P. Slovic, Risk Communication,
Risk Statistics, and Risk Comparisons, Chemical Manufactur-
ers Association, Washington, DC,1988.
Ann Fisher is affiliated with the Office of Policy Analysis,
PM-221, at the U.S. Environmental Protection Agency,
Washington, D.C. 20460. Gary H. McClelland is with the
Department of Psychology, University of Colorado, Boulder,
CO 80309-4305. William D. Schulze is with the Department
of Economics at the University of Colorado. This paper was
submitted for peer review on April 6, 1988. The revised
manuscript was received December 27, 1988.
276 JAPCA
