The Interplay of Science and Values in Assessing and Regulating Environmental Risks
Author(s): Frances M. Lynn
Source: Science, Technology, & Human Values, Vol. 11, No. 2 (Spring, 1986), pp. 40-50
Published by: Sage Publications, Inc.
Stable URL: http://www.jstor.org/stable/689231
Accessed: 14-12-2017 09:12 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/689231?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to
Science, Technology, & Human Values
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
The Interplay of Science and Values in
Assessing and Regulating Environmental
Risks
Frances M. Lynn
In the late 1970s, the U.S. Occupational Safety
and Health Administration (OSHA) proposed a
standard for identifying, classifying, and regulating
carcinogens.' The hearings for the standard
attracted the largest number of participants in
OSHA's rulemaking history and lasted for two
months (May and June 1978). Although a standard
was never implemented, OSHA's efforts stimulated
a debate over the assumptions and decisionmaking
process for assessing and regulating cancer
risks which continues today. This debate
about the methods for identifying and estimating
occupational cancer risks appears at first to be
predominantly scientific, but an analysis of the
responses to OSHA's proposal led me to hypothesize
that regulatory values and other social and
political values influenced the selection among
scientific assumptions and, furthermore, that
these regulatory values seemed linked to place of
employment.
To test these hypotheses, I designed an empirical
study that explored the impact of regulatory
values, institutional affiliation, and other social
and political attributes on occupational health
scientists' selection among assumptions used in
the OSHA standard. I conducted 136 interviews
with occupational physicians and industrial hygienists
working for industry, academia, and gov-
ernment.
This article describes some of the results of that
study and focuses on the role of science and scientists
in the decisionmaking process for determinFrances
Lynn is Research Assistant Professor, Institute
for Environmental Studies, The University of
North Carolina at Chapel Hill, Chapel Hill, NC 27514.
ing whether a product or technology is acceptably
safe. It addresses the question of how to incorporate
expertise into the formation of regulatory
policy, while at the same time protecting democratic
control of the final decision. The article is
divided into three sections: A detailed look at the
policy context surrounding discussions of risk,
analysis of the interview findings, and a discussion
of the study's conclusions and policy implications.
In an appendix, I describe, in more detail,
the choice, construction, and evaluation of the research
instrument, the sampling process and data
analysis techniques.
Policy Context
In the last ten years, a quiet but persistent effort
has developed to incorporate different types of
formal analytic methods such as cost-benefit
analysis and risk analysis into environmental decisionmaking.
Supporters of these techniques
view them as a means to make environmental
decisionmaking more rational and less highly
charged. Critics challenge their use both on the
grounds of methodological flaws and also because
they believe that embedded within the techniques
are value-laden assumptions often missed
in the false precision suggested by the use of numbers
and statistics.
Risk analysis, particularly as applied to the risk
of contracting cancer from exposure to chemicals,
is currently receiving widespread attention. Following
OSHA's efforts to adopt its carcinogens
standard, other governmental bodies and private
groups-including the U.S. Environmental Pro?
1986 by the Massachusetts Institute of Technology and the President and Fellows of Harvard College. Published by John Wiley & Sons
Science, Technology, & Human Values, Volume 11, Issue 2, pp. 40-50 (Spring 1986) CCC /0162-2439/86/020040-11$04.00
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
Lynn: Environmental Risk 41
tection Agency (1985), the White House Office of
Science and Technology Policy (1985), the State
of California (1982), the National Academy of Sciences
(1983), and the U.S. Department of Health
and Human Services (1985)-have issued documents
with the hope of affecting the way cancer
risks are assessed and regulated.
Most of these reports have adopted William
Lowrance's division of the field of risk into two
phases: risk assessment and risk management.2
Lowrance characterized the first phase, risk assessment,
where one estimates the health effects
that varying doses of a substance pose, as an
objective pursuit with decisions and recommendations
most appropriately made by scientists.
Lowrance viewed the second phase, risk management,
as being more value-laden, involving tradeoffs
between health risks and ethical, economic
and other social considerations. In risk management,
Lowrance saw a role for participation in
decisionmaking by the non-expert.
In 1981, the National Academy of Sciences
(NAS), in response to a directive from the Congress,
sponsored another study of risk, convening
the Committee on the Institutional Means for the
Assessment of Risk to Public Health. This Committee
had three primary objectives: to assess the
merits of separating the analytic functions of developing
risk assessments from the regulatory
functions of making policy decisions; to consider
the feasibility of designating a single organization
to do risk assessments for all regulatory agencies;
and to consider the feasibility of developing uniform
risk assessment guidelines for use by all
regulatory agencies.3
Project Director Lawrence McCray, in his
working paper for the NAS Committee, labeled as
"naive" the underlying premise that "matters of
science" could be segregated from "matters of
value" and "left to an organization primarily responsive
to scientific authority."4 In fact, the
Committee's final report suggested that there
were multiple places in risk assessments where
risk to human health could only be inferred. In
those situations, the NAS Committee commented
"how difficult it is to disentangle the
mixture of fact, experience (often called intuition),
and personal values,"5 and it concluded that
throughout the process of risk assessment it was
possible to make choices among assumptions
which would increase the likelihood that a substance
would be judged to be a significant risk to
human health.6 The Committee recommended
establishing a board of risk assessment which
would make explicit "underlying assumption and
policy ramifications" of the different choices
which face scientists who perform a risk assess-
ment.7
The research on which this paper is based,
while conceived and executed before the National
Academy of Sciences' report on risk assessment,
is an empirical complement. The research considers
non-scientific influences on scientists' selection
of assumptions that form the basis for
quantitative risk assessment and risk-benefit
analysis.
Empirical Findings
The study confirmed the initial hypotheses
that there were links between political values,
place of employment, and scientific beliefs. Even
after controlling for the influence of such standard
demographic variables as age, sex, region, religion,
and family background, scientists employed
by industry tended to be politically and
socially more conservative than government and
university scientists. They chose scientific assumptions
that decreased the likelihood that a
substance would be deemed a risk to human health
and increased the likelihood that a higher level of
exposure would be accepted as safe. Government
scientists were the most liberal politically and
most protective in choosing among scientific assumptions.
University scientists fell in between
their governmental and industrial colleagues.
Perceptions of the Risks Society Faces
The respondents were asked questions that attempted
to tap their general perceptions about the
risks of technology as well as what they considered
to be appropriate degrees of regulation. One
such question was adopted from a Louis Harris
poll "Risk in a Complex Society."8 The question
asked whether the respondents felt that the "risks
associated with advanced technology have been
exaggerated by events such as Three Mile Island
and Love Canal."
Figure 1 shows how occupational health scientists
interviewed for the study responded to that
question. The proportion of industry scientists
who supported this statement (82%) is almost
identical to that of the sample of corporate executives
interviewed by Harris. In the Harris survey,
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
42 Science, Technology, & Human Values-Spring 1986
100
90
82 AGREE
80
70 DISAGREE
60
z 51~~~~~~~~~~~~~~5
U. 50 ...c.....
40
30
20 18
10 INDUSTRY
GOVERNMENT UNIVERSITY
Figure 1. Percent who believed that the risks were
exaggerated by TMI.
88% of the corporate executives agreed with the
statement. Government and university occupational
health scientists were more evenly divided
in their response-echoing what Harris found
when he asked the question of members of Congress,
regulators, and the general public.
In another question taken from the Harris poll,
respondents were asked whether
American society is becoming overly sensitive to
risk, and that we now expect to be sheltered from
almost all dangers ... for] ... simply becoming
more aware of risks and starting to take realistic
precautions.
In the Harris poll, corporate leaders were four
times more likely than either the public or regulators,
and three times more likely than members
of Congress, to characterize American society as
"overly sensitive to risk and wanting to be protected
from nearly all dangers." In this study of
industrial hygienists and physicians, those working
for industry were three times more likely than
government and university scientists to believe
that Americans are overly sensitive to risk and
want to be protected from nearly all dangers
(Figure 2).
Respondents were also asked to agree or disagree
with the statement "society has only perceived
the tip of the iceberg with regard to the
risk associated with modem technology." Sixtyeight
percent of the government and university
occupational health scientists agreed with this
statement, compared to thirty-two percent who
worked in industry.
OVERLY SENSITIVE
80 MORE AWARE
70W BOTH
, 60 56 54
L) 50__
oS1
INDUSTRY GOVERNMENT UNIVERSITY
Figure 2. Feelings about Ameri
toward risks.
No Anti-Regulation
Although they disagreed about the extent of
risk facing society, those interviewed for this
study, including those working for industry, were
not necessarily anti-regulatory. Although the majority
of industry scientists were Republicans (see
Figure 31, had voted for President Reagan (74%33
and were self-identified as conservative (56%),
they expressed surprise at the Reagan Administration's
vehement attacks on environmental regulation.
When read a list of different types of environmental
regulations (e.g., air, water, consumer
products and asked whether the regulations
should be made "more strict "less strict or
skept the sameiln very few industry respondent
wanted the regulations weakened.
On the other hand, very few wanted them
strengthened. For examples when asked about reg-
100
90
80
70
60 56
Q 50 46Wkb
ulati on. Whentra icitio of dfrentoyens ofenv"kept,
thepusiame," very fneewndutrnrspndnt100
90
40~~~~~~~~~~~2
70 1
60 56
40ur 334ryietfiaino epodns(eo
30t eulcn o needn)
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
Lynn: Environmental Risk 43
100. STRICTER
90
80 - 80 SAME
70 | 67 LESS
80
Figu0 47Pretwo eivdtargltost
30~~~~~~~~~~~~~~~~5
20 18
10
333
INDUSTRY GOVERNMENT UNIVERSITY
Figure 4. Percent who believed that regulations to
protect employees from health risks like cancer should
be stricter, be more lenient, or remain the same.
ulations to protect employees from working conditions
that cause health risks like cancer, the
majority of industry scientists (67%) wanted the
regulations kept the same. This finding contrasts
with the responses of government scientists, 80%
of whom wanted the regulations made stricter.
Academic opinion was evenly divided between
wanting the regulations made more strict or keeping
them as is (Figure 4).
The dominant sentiment across all institutional
settings, however, was that a minimum
level of safety had to be maintained. Majorities in
all three subsamples (Figure 5) disagreed with the
question taken from the Harris poll, which asked
whether "a consumer should be allowed to
choose between a very safe product at a higher
price and the same product at a lower price without
safety equipment." By inference, respondents
IlI AGREE
90 DISAGREE
80
71
70 64
U 50,
6 40 36
30 2
20~~~~~~~~~~~2
INDUSTRY GOVERNMENT UNIVWRSITY
Figure 5. Attitudes toward consumer choices between
price and safety.
seemed to disagree with a 1983 suggestion made
by James Miller III, chairman of the Federal
Trade Commission and now head of the Office of
Management and Budget that "imperfect products
should be available because consumers have
different preferences for defect avoidance."9
Attitudes Toward Cost-Benefit Analysis
Cost-benefit analysis, a major proposal for
regulatory reform, was most strongly supported
by industry scientists (Figure 6). Proponents view
cost-benefit analysis and its cousin, risk-benefit
analysis, as means by which to make environmental
decisionmaking more systematic.'l Critics
challenge the use of cost-benefit analysis,
suggesting that the technique has major methodological
flaws and has imbedded within it philosophically
conservative assumptions about issues
of distribution, equity, and individual rights."
The response of the sample to questions that
probed the basis of cost-benefit analysis suggest
the technique is not well understood. For, although
71% of the industry sample supported the
use of cost-benefit analysis (Figure 6), only 52%
agreed with the statement that "society must attempt
to place an economic value on human life
in order to allocate scarce resources."
When presented with the dominant methods
currently used to value life in cost-benefit analysis
(human capital, willingness-to-pay, and wage
differentials), almost half of the sample was dissatisfied
with all three options (Table 1). In fact,
only 5% of the total sample, including 5% of industry
scientists, supported the method currently
100
FAVOR
90
0 OPPOSE
80
70 7'll|m||NOT SURE
1- ~~~~~~~~~~~~~~~~53
U 50
aw 43
40 3
30
22 ~~~~~~~~24
20
10,
INDUSTRY GOVERNMENT UNIVERSITY
Figure 6. Attitudes toward the government's use of
cost-benefit analysis.
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
44 Science, Technology, & Human Values-Spring 1986
Table 1. Attitudes toward dominant methods used by
economists to value life. Respondents were asked the
following question: "A number of economists think
that ultimately one must place a value on human life,
that is, decide how much money society is prepared to
invest in order to prevent one additional death or save
one additional life year. Three methods are currently
being used by economists. If you had to choose a technique
for valuing life, which would you select? (A)
Compute the amount of earnings that would be lost in
the case of premature death or disability and equate
this with the value of life/disability (HUMAN CAPITAL
APPROACH); (B) Ask individuals how much they
would be willing to pay to reduce the probability of
death or disability (WILLINGNESS-TO-PAY APPROACH);
or (C) Analyze wage differentials in occupations
involving varying risk of death or injury and use
wage rate differentials as reflections of societal willingness-to-pay
for decreases in risk (WAGE DIFFEREN-
TIALS)."
Willing- Wage
Human ness to DifferCapital
pay entials None
Industry 38% 17% 5% 40%
N = 42
Government 32% 17% 5% 46%
N = 41
University 9% 26% 6% 59%
N = 34
TOTAL 27% 20% 5% 47%
N = 117 (32) (23) (6) (56)
popular among economists, using wage differentials
(i.e., hazard pay) as proxies for the value that
people place on measures to reduce risks of injury
or death.'2 The wage differential method uses regression
analysis to measure job earnings against
a measure of the risk level of each job, holding
constant other variables that also influence variations
in observed earnings. Studies using this
technique have yielded estimates of life values
ranging from $300,000 to $35 million.
This research attempted to involve the scientists
in a willingness-to-pay exercise. The willingness-to-pay
approach is another technique currently
popular with economists as a means to
value life. Respondents were asked to select an
acceptable level of risk, stated in an annual probability
of death. They were next asked what salary
increase would make them accept a job with a
higher probability of death. Many refused to participate
because they found such an exercise odious.
Some wanted a list of common risks and
their probabilities of death in order to make comparisons.
But many others said that given their
existing income, they would never have to make
such choices and would not now.
Critics of the willingness-to-pay approach question
the technique's assumption that workers
possess accurate knowledge of risks and can
readily move to jobs posing lower risks.'3 Critics
also claim that the technique ignores existing income
distributions. The rich will pay more (or
earn less) to reduce the risk of their death.'4
Even those in the sample who generally favored
the use of cost-benefit analysis did not advocate it
as the deciding factor in standard setting. Ninetyeight
percent of the entire sample viewed costbenefit
analysis as a "decision tool," not a "decision
rule. " Moreover, majorities in all
employment categories in the sample (including
89% of industry scientists) supported the "public
policy goal of decreasing cancer risks" even if it
"caused the average price of goods and services to
increase" (Figure 7). Similarly, majorities "supported
the public policy goal of decreasing cancer
risks . . . [even if it] . . . caused some factories to
close down and increased unemployment."
Given the lack of support for the dominant
techniques used to value life and a willingness to
accept price rises for prevention of disease, why
did respondents tend to support cost-benefit
analysis? For some, support may be symbolic of
the desire to lessen regulatory burdens on business.
But for others, support may be a part of a
hope that a technique can be found to make environmental
decisionmaking easier. In this sense,
cost-benefit analysis and risk assessment share a
common attribute. Both are expected to provide
98 100.
90 ......89 FAVOR
8 * OPPOSE
70
60
cc' 50'
40'
30-
20~~~~~~~~~~~~
INDUSTRY GOVERNMENT UNIVERSITY
Figure 7. Attitudes toward preventing cancer at the
risk of rise in prices.
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
Lynn: Environmental Risk 45
100
80 AGREE
70 ......69 DISAGREE
60
Z 52
0 ......
40
31
30 2
20
10
INDUSTRY GOVERNMENT UNIVERSITY
Figure 8. Attitudes toward using animal data to identify
risks to people.
objective answers to uncertain and value-laden
processes.
Quantitative Risk Assessment
The National Academy of Sciences in its 1983
report, Risk Assessment in the Federal Government,
characterized risk assessment as the use of
a "factual basis to define the health effects of exposure
to individuals and populations to hazardous
materials and situations."5 The NAS viewed
risk assessment as both "quantitative and qualitative"
and involving "the interplay of science
and policy." The NAS identified close to fifty areas
in a risk assessment where a scientist or risk
assessor must make choices among "several scientifically
plausible options" in which "policy
considerations inevitably affect and perhaps determine,
some of the choices."16
Figure 8 shows this sample's response to one of
those inference points: using animal data to predict
risks for humans. Respondents were asked
whether they agreed or disagreed that "a substance
which is shown conclusively to cause tumors
in experimental animals should be considered
a carcinogen, thereby posing a risk to
humans." The majority of government (69%) and
university scientists (52%) agreed with this statement,
while only 27% of the industry sample
agreed. The National Academy of Sciences feels
that
the inference that results from animal tests are
applicable to humans is fundamental to toxicological
research; this premise underlies much of
experimental biology and is logically extended to
the experimental observation of carcinogenic ef-
fects.'7
A similar question is whether or not threshold
levels exist for carcinogens below which there are
no negative effects. A threshold model for carcinogenesis
assumes that there is a dose below
which cellular or tissue damage does not occur.
Under this model, one assumes that the body has
mechanisms that withstand toxic events at low
doses. A linear model suggests, especially in the
case of carcinogens, that a single interaction with
a single cell can trigger a toxic reaction. The National
Academy of Sciences Committee on Risk
Assessment concluded that there was no conclusive
biological evidence to support either type of
model.'8 The regulatory policy implications of
support for the existence of thresholds is viewed
as less protective of public health. The State of
California,'9 as well as the Office of Science and
Technology Policy,20 rejected the concept of
thresholds for carcinogens in their guidelines.
As in the choice of using animal data to identify
cancer risks for humans, the occupational health
scientists working in industry held views about
thresholds that differed substantially from those
of government scientists. Eighty percent of the
industry scientists interviewed agreed that
thresholds exist for carcinogens, compared to
37% of government employees (Figure 9). Noteworthy
are the responses of academics, the
majority of whom supported the existence of
thresholds. Among academics, age made a difference
in attitudes; older academics were
more likely to believe in thresholds than younger
ones. A likely explanation for this difference is
that when older academics entered the field of
100
90 EJ AGREE
80 . ....
70 DISAGREE
63 61
60....
U 50-
40 3 7
30-
20-
10.
INDUSTRY GOVERNMENT UNIVERSITY
Figure 9. Attitudes toward the existence
for carcinogens.
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
46 Science, Technology, & Human Values-Spring 1986
100
90 mn
80 11lI~AGREESO 74 A EE
70 DISAGREE
61
60
0 50
CL ~~~~~~~~~39
40
30 2
20
10
THRESHOLD THRESHOLD
DOESN'T EXIST
Figure 10. The relationship between belief in a
threshold to attitudes about whether TMI exaggerated
the risks.
occupational and environmental health in the
late 1930s and 1940s, levels of exposure were high
and scientific recommendations seemed clear.
The older academics may have believed that
government policy of the 1970s was becoming
overprotective.
Figures 10 and 11 and Tables 2 and 3 show that
the samples' beliefs toward the existence of
thresholds correlate positively with social and political
attitudes reported earlier in this article.
The key finding of these data is that support for
the existence of thresholds is associated with conservative
political attitudes and with perceptions
that suggest that American society is overly sensitive
and overreacting to environmental risks.
Similarly, those who question the use of animal
data to predict carcinogenesis in humans are
more likely to hold conservative political attitudes
and believe that Americans are too risk ad-
verse.
100
90
80 78 I FAVOR
70 OPPOSE
62
60
U 50
40 3
30
22
20
10,
THRESHOLD THRESHOLD
DOESN'T EXIST
Figure 11. Attitudes toward the existence of a threshold
and the government's use of cost-benefit analysis.
Table 2. The relationship between attitudes toward
the existence of thresholds and the respondent's vote
in the 1980 Presidential election (p < 0.02; N = 75).
Vote in 1980 Presidential
Attitudes about Election
Thresholds Carter Reagan Total
Thresholds 36% 63% 99%
exist (18) (31) (49)
Thresholds 69% 30% 99%
don't exist (18) (8) (26)
Total
For instance, 74% of those who believe in
thresholds also believe that the reactions to TMI
have exaggerated the risks associated with advanced
technology. By contrast, only 39% of
those who question the existence of thresholds
believe that the risks of advanced technology
have been exaggerated by Three Mile Island (Figure
10).
A similar pattern held on the issue of risk sensitivity.
Fifty-seven percent of those who believe in
thresholds also believe that "Americans are
overly sensitive to risk and [want] to be protecte
from all dangers." Only 16% of those who question
the existence of thresholds for carcinogens
believe that Americans are overly sensitive to
risk.
Belief in the existence of thresholds correlates
with voting in the 1980 Presidential election (Table
2) and self-identification as conservative and
Republican. Support for the existence of thresholds
for carcinogens is also linked to support for
the use of cost-benefit analysis (Figure 11) as well
as to a decreased willingness to strengthen environmental
regulations (Table 3). In sum, in areas
of science where there is no data to distinguish
choices of models or assumptions, scientific
choices correlate highly with personal political
Table 3. The relationship between attitudes toward
the existence of thresholds and the regulations to protect
employees from cancer should be made more strict
or kept the same (p < 00; N = 86).
Regulations to Protect
Employees
Attitudes about More
Thresholds Strict Kept Same Totals
Thresholds 43% 56% 99%
exist (53)
Thresholds 75% 25% 100%
don't exist _ (33)
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
Lynn: Environmental Risk 47
beliefs. The policy implications of these findings
are treated in the final section of this article.
Policy Implications
The idea for this study grew from the debates
over OSHA's attempt to adopt a generic method
to identify and classify carcinogens. OSHA's was
the first in a series of continuing efforts by both
public and private bodies to promulgate guidelines
for identifying and assessing cancer risks. In
recent years, the debate over the use of a linear
dose-response model and the role for animal data
has been joined by issues such as the incorporation
of pharmokinetic data, differentiation among
genotoxic and non-genotoxic data, the role given
to different routes of exposure or to tumor sites in
animals not found in humans, and whether to use
"best estimates" or "worse case" characterizations
of data.
A reading of the record in the OSHA hearings as
well as of contemporaneous articles in scientific
journals suggests a lack of consensus on the
methods and models in the rapidly emerging field
of carcinogenic risk assessment. The issues
seemed to fall into the category of "trans-science,"
a term that Alvin Weinberg used in a 1972 article
in the journal Minerva.2l Weinberg cited the determination
of the biological effect of low-level
radiation insults as an example of a trans-scientific
issue. He suggested that the argument about lowlevel
radiation insults would have been far more
sensible had it been "admitted at the onset that
this was a question which went beyond science.
The matter could have been dealt with initially
on moral or aesthetic grounds."
Weinberg's observations in the early 1970s
seemed to go unnoticed in much of the debate
surrounding the OSHA standard and ensuing discussions
of assessing the risks of cancer. As late
as 1983, EPA Administrator William Ruckelshaus
was reiterating the distinction between risk
assessment and risk management, saying in a
speech before the National Academy of Sciences
that
Nothing will erode public confidence faster than
the suspicion that policy considerations have
been allowed to influence the assessment of
risk.22
A year later, however, in a February 1984 speech
at Princeton University, Ruckelshaus qualified
his position by saying that he had found that separating
the assessment of risk from its
management is rather more difficult to accomplish
in practice .. . values, which are supposed
to be safely sequestered in risk management, also
appear as important influences on the outcome of
risk assessments.23
Nonetheless, Ruckelshaus's qualification seemed
to fall on deaf ears, for on 4 January 1985 the
White House issued Executive Order 12498
which stated that risk assessments are to be "scientifically
objective."
The results of the study on which this article is
based call into question the characterization of
risk assessment as "objective." The results suggest
that under conditions of scientific uncertainty,
regulatory policy implications either consciously
or unconsciously influence the models,
assumptions and theories that scientists' choose.
U.S. District Court Judge David L. Bazelon in a
1979 speech, "Risk and Responsibility," warned
that
in reaction to the public's often emotional response
to risk, scientists are tempted to disguise
controversial values decisions in the cloak of scientific
objectivity, obscuring those decisions
from political accountability.24
Currently, American industry is seeking to create
and lodge increasing decisionmaking responsibility
in science advisory panels. The most public of
these efforts was a 1983 proposal by the American
Industrial Health Council (AIHC), an association
of over 90 different companies and trade associations
formed in reaction to OSHA's cancer proposal,
to establish a Central Board of Scientific
Risk Analysis under the National Academy of
Sciences to review regulatory agencies' risk assessments.
A bill to accomplish this was introduced
by Representative James Martin (R-North
Carolina) and was viewed by the AIHC as a means
of getting "statutory recognition of good science"
and part of the AIHC's plan to "spearhead efforts
to ensure distinction between risk assessment
and risk management."25
Representative Martin withdrew his bill in the
spring of 1984. It had been criticized not only by
environmental groups but also by professional organizations
such as the American Chemical Society
and by prominent scientists. Epidemiologist
Alice Whitmore, in an article in the Journal of the
Society for Risk Analysis, argued that
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
48 Science, Technology, & Human Values-Spring 1986
attempts to view toxicant risk analysis as involving
two stages frisk assessment and risk management)
with risk assessment relying on scientific
activity and scientific judgment ... [was] ... an
erroneous description of reality and an unattainable
goal for the regulatory process.26
Her research suggested that values unavoidably
enter virtually every aspect of risk analysis. Whitmore
contended that
creating a central authoritative panel of distinguised
scientists who will resolve on a case-bycase
basis, complex or difficult issues for regulatory
agencies. . . would move the scientists from
the frying pan into the fire ... [and] . . . would
ensure that a scientist's values would drive the
decision.27
The potential danger of these boards and panels,
especially in areas with pervasive uncertainties
such as quantitative risk assessment, is that scientists
alone will make decisions with important
political and ethical implications for the protection
of human health. The question is whether
scientists are the only ones that should be in-
volved.
One could argue that the most appropriate role
for the scientists in this situation would be to
self-consciously provide decisionmakers and the
public with as much information as possible
about the uncertainties in his or her work. Social
scientists and ethicists could be involved in analyzing
which, if any, social, ethical and political
implications flow from selecting one assumption,
model or theory instead of another.
This will mean that a new type of debate will
be occurring in policy discussions, one which is
more conscious of those places in the scientific
process where non-scientific values play a role
and hard political choices must be made. This
will be to the benefit of science for it may avoid
the spectacle of the dueling scientist and place
the very difficult decision of degrees of protection
and the acceptability of a risk into the political
arena, where, in a democracy, it belongs.
Appendix
This appendix focuses on the choice, construction,
and evaluation of the research instrument. It
also describes the sampling process and data analysis
techniques.
The research instrument used was an interview
pre-tested and conducted in person by the author.
The decision to conduct interviews as opposed to
a mailed questionnaire, and hence a larger sample,
was made primarily because of the sensitivity
and relative newness of the subject matter under
investigation, and because of a desire to have an
explanatory richness difficult to obtain through
the questionnaire process. Sensitive questions included
probing institutional constraints and ethical
dilemmas which arise in the process of an occupational
health professional's job. In the case of
issues which involved scientific uncertainty and/
or controversy such as the existence of thresholds
and the use of cost-benefit analysis, the interview
format offered the potential for probing explanations
and possibly a better understanding of the
basis for attitudes.
Over half of the interview items came from
other studies of risk. The main source of these
questions was from a study, "Risk in a Complex
Society," conducted by the Louis Harris survey
organization for Marsh and McLennan, Inc., the
world's largest insurance broker. The survey, the
most expensive ever conducted by the Harris organization,
was administered in 1980. A second
source of questions came from a study, "Public
and Worker Attitudes Toward Carcinogens and
Cancer Risk," prepared for Shell Oil by Cambridge
Reports.
One hundred thirty-six randomly-selected, inperson
interviews were conducted in the New
York and Cincinnati metropolitan areas and the
Research Triangle and Triad of North Carolina.
All geographic areas contain a sufficient number
of academic, government and industrial institutions
to provide an adequate sampling frame.
An effort was made to make the selection process
replicable. Names were selected from the
main professional organizations. For industrial
hygienists, these groups included the American
Industrial Hygiene Association (AIHA), the
American Conference on Governmental and Industrial
Hygienists (ACGIH), and the American
Academy of Industrial Hygiene (AAIH). Within
occupational medicine there is no one or even
two organizations to which all physicians, regardless
of employment, belong. Physicians working
for industry, but not for academia or the government,
belong to the American Occupational Medicine
Association AOMNA). The industry sample,
therefore, was drawn from the membership rosters
of AOMNA. The department heads of univerThis
content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
Lynn: Environmental Risk 49
sity-based occupational health programs supplied
the names of academic physicians. Heads of govemment
agencies likewise advised on government
physicians.
The sampling frame was defined by zip codes,
stratified by institutional setting and profession
(industrial hygiene or occupational medicine),
and then randomly selected. In all but two categories
(industrial and academic physicians), the response
from the initial postcard request ranged
from 75% to 85%. Follow-up calls to physicians
in industry and academia were successful in filling
quotas. These are unusually high response
rates and can probably be attributed to the salience
of the topic and the specialized nature of the
sample. Several respondents said that they were
intrigued by the topic and felt that they could
learn something from the interview. The sample
was predominantly white (98%), male (88%) and
Protestant (56%).
The data analyzed was categorical (e.g., information
measured on nominal or ordinal scales or
grouped continuous data). Statistics used to analyze
the data included frequencies, chi-squares
and weighted least-squares, an application of the
general linear model to categorical data. In the
weighted least-squares technique, one looks at
variations in cell probabilities and models hypothetical
regression lines for associations or interactions
among variables. One uses the goodnessof-fit
chi-square statistic to compare expected
and observed frequencies, asking the question
whether "the departures between observed and
expected values . . [can] . . . reasonably be attributed
to chance or ... [whether] ... they are so
large that the model itself seems wrong?"28
Initial contingency tables showed extremely
strong statistical significance between institutional
affiliation and a wide variety of dependent
variables. Methodologist Hubert Blalock feels
that we may be saying "quite a bit when we can
establish significance with small samples" because
small samples require a "much more striking
relationship in order to obtain significance."29
In addition to institutional affiliation, controls
were run for the following variables: age, region,
religion and father's education. Age was the only
variable which appeared to make a substantive
difference on the effect of institutional affiliation.
Age did not prove to be statistically significant in
altering the effect of employment setting.
An unanticipated result of doing personal interviews
was the opportunity, on a limited basis, to
experiment with a more interactive questioning
mode. In the last set of interviews, respondents
were given the option of filling out the fixed choice
questions themselves. This technique permitted
more time for open-ended questions and probes.
In at least six interviews involving research scientists,
the interviewees were asked to reflect on
the process by which he or she made a hazard
determination and to identify those points in the
process where professional and/or personal judgment,
as opposed to scientific certainty, were involved.
The interviewees seemed to welcome the
opportunity to reflect on the research process.
The role of the interviewer was to keep the respondent
on as straight a path as possible and to
question constantly the basis for decisions. This
is a fruitful method to be used in research of this
type.
Notes
1. U.S. Department of Labor Occupational Safety and
Health Administration, Identification, Classification
and Regulation of Potential Carcinogins Federal
Register, Volume 45, No. 15 Book 2, 1980.
2. William W. Lowrance, Of Acceptable Risk: Science
and the Determination of Safety (Los Altos,
CA: William Kaufman, 1976).
3. National Academy of Sciences, Risk Assessment
in the Federal Government (Washington, DC: National
Academy Press, 1983).
4. Lawrence McCray, An Anatomy of Risk Assessment:
Scientific and Extra-Scientific Components
in the Assessment of Scientific Data on Cancer
Risks, Working Paper for the National Academy of
Sciences Committee on the Institutional Means
for Assessment of Risks to Public Health (Washington,
DC: National Academy Press, 1983), p. 13.
5. National Academy of Sciences, op. cit., p. 36.
6. Ibid., p. 34.
7. Ibid., p. 171.
8. Louis Harris and Associates, Risk in a Complex
Society (New York: Marsh and McLennan Companies,
1980).
9. Mark Green, "The Gang that Can't Regulate," The
New Republic, Volume 188, (1984): 14-17.
10. C. Starr, Science, Volume 165, Number 3899
(1969); Ezra J. Mishan, Cost-Benefit Analysis
(New York: Praeger, 1976); Edith Stokey and Richard
Zeckhauser, A Primer for Policy Analysis
(New York: W.W. Norton and Co., 1978); Richard
Zeckhauser, "Procedures for Valuing Lives," Public
Policy, Number 23, (1975): 419-462.
11. Lawrence H. Tribe, "Policy Science: Analysis or
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms
50 Science, Technology, & Human Values-Spring 1986
Ideology?" Philosophy and Public Affairs, Volume
2, (1972): 66-1 10; Baruch Fischhoff, "Cost-Benefit
Analysis and the Art of Motorcycle Maintenance,"
Policy Sciences, Volume 8, (1977): 188-202; Nicholas
Ashford, "The Usefulness of Cost-Benefit
Analysis in Decisions Concerning Health, Safety,
and the Environment," Annals of the New York
Academy of Sciences, Volume 363, 1980; Michael
S. Baram, "Cost-Benefit Analysis: An Inadequate
Basis for Health Safety and Environmental Regulatory
Decision Making," Ecology Law Quarterly,
Volume 8, (1980): 474-531; Marguerite Connerton
and Mark MacCarthy, Cost-Benefit Analysis and
Regulation (Washington, DC: National Policy Exchange,
1982).
12. Robert Smith, The Occupational Safety and
Health Act (Washington, DC: American Enterprise
Institute, 1976); Richard Thaler, and Sherwin Rosen,
"The Value of Saving a Life: Evidence from the
Labor Market," in Nestor Terleckyi, ed., Household
Production and Consumption (New York:
Pergamon Press, 1976); Kip Viscusi, "Labor Market
Valuations of Life and Limb: Empirical Evidence
and Policy Implication," Public Policy (Summer,
1978): 369-372.
13. David Zimmerman, "Coercive Wage Offers," Philosophy
and Public Affairs, Volume 10, (1981):
121-145.
14. Thomas C. Schelling, "The Life You Save May Be
Your Own," in S.B. Chase, ed., Problems in Public
Expenditure Analysis (Washington, DC: Brookings
Institute, 1968).
15. National Academy of Sciences, op. cit., p. 3.
16. Ibid., p. 28 and p. 33.
17. Ibid., p. 22.
18. Ibid., p. 25.
19. State of California, Department of Health Services,
Health and Welfare Agency, Carcinogen Identification
Policy: A Statement of Science as a Basis of
Policy (1982): 19.
20. Office of Science and Technology Policy, Executive
Office of the President, Chemical Carcinogens:
A Review of the Science and Its Associated
Principles, Federal Register, Volume 50, Number
50 (1985).
21. Alvin Weinberg, "Science and Trans-Science,"
Minerva, Volume 10, (1972): 209-222.
22. William Ruckelshaus, "Science, Risk and Public
Policy," Science, Volume 221, (1982): 1026-1028.
23. William Ruckelshaus, "Risk in a Free Society,"
speech given at Princeton University, 18 February
1984, mimeo.
24. David Bazelon, "Risk and Responsibility," Science
Volume 205, (1979): 277-280.
25. American Industrial Health Council, Report to the
Membership, Scarsdale, NY, 1983.
26. Alice Whitmore, "Facts and Values in Risk Analysis
for Environmental Toxicants," Risk Analysis,
Volume 3, (1983): 23-33.
27. Ibid., p. 32.
28. H.T. Reynolds, Analysis of Nominal Data (Beverly
Hills, CA: Sage Publications, 1977), p. 58.
29. Hubert L. Blalock, Social Statistics (New York:
McGraw-Hill Book Co., 1960), p. 227.
This content downloaded from 147.251.55.10 on Thu, 14 Dec 2017 09:12:20 UTC
All use subject to http://about.jstor.org/terms