Risk Analysis, Vol. 15. No.

5, 1995

Characterizing Perception of Ecological Risk

Timothy McDaniels,J Lawrence J. Axelrod, and Paul Slovic2
Received December 2, 1994; revised June 7, 1995

Relatively little attention has been paid to the role of human perception and judgment in ecological
risk management. This paper attempts to characterize perceived ecological risk, using the psychometric paradigm developed in the domain of human health risk perception. The research began by
eliciting a set of scale characteristics and risk items (e.g., technologies, actions, events, beliefs)
from focus group participants. Participants in the main study were 68 university students who
completed a survey instrument that elicited ratings for each of 65 items on 30 characteristic scales
and one scale regarding general risk to natural environments. The results are presented in terms
of mean responses over individuals for each scale and item combination. Factor analyses show
that five factors characterize the judgment data. These have been termed impact on species, human
benefits, impact on humans, avoidability, and knowledge of impacts. The factor results correspond
with initial expectations and provide a plausible characterization of judgments regarding ecological
risk. Some comparisons of mean responses for selected individual items are also presented.

KEY WORDS: Risk perception; ecological risk analysis.

1. INTRODUCTION

matization of products, places, and technologies,@)and

the determinants of value judgments underlying health
risk tradeoffsJ)
In recent years, ecological risks (threats to the
health and productivity of species and ecosystems) have
also arisen as a topic of great public concern, in parallel
with heightened attention to resource sustainability and
concern over environmental degradation. Examples of
ecological risks range from specific threats to localized
ecosystems from development or pollution to threats to
global ecosystems from climate change. While the risk
management community has recognized the increasing
need for serious research on ecological risk managementJaI0) much of the work thus far has been undertaken
from the perspective of the physical and biological sciences. Relatively little effort has been devoted to social
science questions regarding human perception, mental
characterization, value assessment, or decision-making
structures regarding ecological risks.
This article begins to address that gap. It presents
a framework for characterizing ecological risk perception, building on the psychometric risk perception par-

Managing risk to human health and safety has, over

the last two decades, become a dominant theme in government policy, public debate, media attention, and academic research. A striking aspect of this growth is the
increasing attention paid to human perception and judgment in debates that were initially characterized as based
completely in science and technology. One line of research involving human judgment had its origins in experimental work that adapted psychometric scaling
methods to characterize peoples perceptions of the relative riskiness of technologiesJl2) Recent research on
perceived health risk has provided insight into key social
aspects of health risk management, including how best
to communicate information about health risksJ3)the social amplification of risk impa~ts:~.~)
risk-induced stiglUniversity of British Columbia, Westwater Research Centre and
School of Community and Regional Planning, 433-6333 Memorial
Rd., Vancouver, British Columbia V6T 122, Canada.
* Decision Research, 1201 Oak St., Eugene, Oregon 97401-3575.
To whom correspondence should be addressed.

575
0272-4332/95/1ooo6575$07.50/1 0 I995 Society for Risk Analysis

McDaniels, Axelrod, and Slovic

576

adigm that emerged from the study of risks to human

health.",2) In simple terms, our primary objective is to
clarify what people mean when they say something is
risky to the environment. We attempt to identify the
Characteristics that lead individuals to perceive one activity as a high ecological risk, and another activity as
less so.
The psychometric paradigm is an approach for
identifying the characteristics influencing people's perceptions of risk. The approach assumes that risk is inherently multidimensional, with many characteristics
other than the probability of harm affecting individual
judgments. Applying the method to human health risk
perception includes:
1. Developing a list of hazard items or risky events,
technologies, and practices that span a broad domain of potential hazards.
2. Developing a number of psychometric scales
that reflect characteristics of risks that are important in shaping human perception of, and response to, different hazards.
3. Asking people to evaluate the list of items on
each of the scales.
4. Using multivariate statistical methods (such as
factor analysis), to identify and interpret a set of
underlying factors that capture the variation in
the individual and group responses.(2)
The present study follows these basic steps. However, in this study we assume that there are substantial
differences between judgments of perceived health risk
and perceived ecological risk, with the most fundamental
difference being the greater complexity of ecological
risk judgments.(*)One source of complexity is the wider
range of possible end states of interest. Ecological health
is less well defined than human health and will have a
much wider array of meanings.@)For example, does a
naturally occumng hazard (e.g., a flood or an earthquake) or a hazard that threatens humans pose ecological
risk? A second source of complexity is the potentially
greater influence of personal characteristics of people
(e.g., worldviews, value orientations, and prior experience with nature and potential hazards). These variables
could be more influential because of the greater diversity
of opinion regarding what ecological risk means in systems where natural forces themselves create massive
changes in species and their habitats. Still another source
of complexity is the great variation in the physical scale
of ecological systems, which can range from a few
square meters of plants to the global level. One final and
subtle source of complexity may be the concern for entire ecological systems and species, rather than effects
on individuals.

Given this complexity, we assumed that many new

sets of scales and items would be required to characterize respondents' judgments about ecological risks. We
developed these scales and items through a series of focus groups, as described in the next section. In sum,
even though we followed the basic steps of the human
health risk perception approach, the survey instrument
in this study and the concepts it examines were developed specifically to address ecological risks.
We began the study with a number of expectations,
drawn from various sources, regarding the kinds of factors that would eventually characterize ecological risk
perception. We expected that possible influences could
include the potential for loss of species and ecosystems;
the potential for more conventional environmental impacts (e.g., water pollution) that directly affect human
uses of natural resources; the potential scope and destructiveness of impacts; the potential influence of dread
and knowledge as indicated in the human health-risk
perception literature; and the perceived benefits to hum a n ~ . ( ' ~In. l sum,
~ ) we expected that a substantial number
of dimensions could be important in characterizing perceived ecological risk.

2. METHOD
2.1. Participants

Participants in the study were 40 women and 28

men who were recruited from the student population at
the University of British Columbia. They averaged 23
years in age, with a range of 18-39. The sample included
students from most of the faculties and academic disciplines at the university. Although this sample is small,
and its members are clearly not representative of the
general public (being younger and better educated on
average), several previous studies have used such samples to explore risk perception issues.") The sample was
recruited through advertisements seeking individuals interested in a few hours of paid work filling out a survey.
Participants took between 2 and 3 hours to complete the
survey instrument and were paid either $20.00 or
$25.00, depending on the time required.
2.2. Item and Scale Development
As a first step in developing the survey instrument,
four focus groups were held to help identify (a) items
that may be perceived as risks to the health and produc-

Characterizing Perception of Ecological Risk

tivity of natural environments, and (b) scales reflecting
characteristics of these items that may influence the
judgment of risk. Two of the groups consisted of participants from a range of backgrounds including environmental managers, environmental activists, service
workers, union members, and university students. A
third group consisted of academic specialists concerned
with the biological and societal dimensions of natural
environments. Participants in the fourth group were
members of a high school English class. An open discussion format was employed in each focus group, in
which participants were prompted to discuss freely the
two issues noted above. These discussions were lively
and expansive, arousing enthusiasm and emotion in the
participants. Information gathered from these focus
groups was structured using content analytic procedur e ~ and
, ~ comprehensive lists of risk items and characteristics were generated. These lists were then reviewed
by the researchers to eliminate any redundancies and to
ensure item and scale clarity.

577

me). Items in this latter grouping generated substantial

discussion (and emotion) during the focus groups. Although the link between some items and ecological risk
may not be obvious (e.g., television), their inclusion in
the final list was consistent with the goal of representing
the diverse perspectives raised in the focus groups.
The list also included items that are indirect sources
of risk (e.g., air conditioning, aerosol cans), direct
sources of risk (e.g., emission of ozone depleting gases),
and consequences of risks (e.g., ozone depletion). Thus,
several of the items are directly related, with some being
causes of others. We adopted this approach to clarify
differences in peoples judgments about these linked
items. In addition, although all 65 items could be perceived as posing some level of risk, items were included
that were reasonably expected to be rated as not very
risky (e.g., scuba diving, outdoor recreation) as well as
items that likely posed substantial risk (e.g., global
warming, acid rain, population growth, loss of animal
species). The entire set of items is presented later, in
Table 111, with the results.

2.2.1. Items
2.2.2. Scales
The final list included 65 items that were perceived
as posing some level of ecological risk. The set of items
derived from the focus groups was extremely broad, and
included a vast range of human endeavors and natural
phenomena. Four general groupings in these items could
be identified, although some overlap among classifications is unavoidable. One grouping contained natural
disasters (i.e., earthquakes, volcanos, drought, floods,
and meteors colliding with Earth); a second involved
technologies and their applications (e.g., fertilizers, pesticides, mass production farming practices, oil transportation, incineration). A third grouping included human
practices that were seen as potentially having some negative environmental impact (e.g., poaching, disposal of
different kinds of waste products, driving automobiles,
cigarette smoking, beef production, tourism and travel,
scuba diving). The fourth grouping included human beliefs and political/social systems (e.g., capitalism, consumer-oriented society, disconnection of modem life
from natural environments, human dominion over naTwo researchers took notes in each group. Each person then coded
their notes as to potential items and scales. A graduate student compiled the content of each set of notes into a comprehensive list of
items and scales using a simple union procedure. The researchers
then reviewed the lists to eliminate redundancies. This process determined the lists of potential scales and items, which were edited
slightly to insure clarity and make the judgment task feasible for
individuals to complete at one administration.

In contrast to previous research in risk perception,

where the relevant characteristics of risks have been
largely based on theory and literature reviews,) we developed judgment scales reflecting risk characteristics
based on information obtained from the focus groups.
From this approach, 31 scales (including one general
risk scale) were developed to characterize the ecologically risky items. A number of the scales paralleled
those found to have explanatory value in studies of human health risk-perception (e.g., Ref. 1). Examples of
these include observability of potential consequences associated with the item, knowledge of the risks, the severity of the consequences, and the controllability of the
potential impacts. The dread scale, which plays such a
prominent role in health risk perception research, has in
this study an analog worded in terms of negative emotion. That is because focus group participants sometimes
mentioned sadness, anger, disgust, or frustration in discussion of certain ecological risks, but the notion of personal dread (fear) was never mentioned or a ~ p a r e n t . ~
Several scales were identified in the focus groups that
We were surprised and moved by the extent of emotional reactions
in the focus groups. In groups with individuals from diverse backgrounds, people were at times close to tears when reflecting on ecological risk. The profound sadness felt in response to threats to
nature, and the frustration arising from an inability to reduce these
threats was palpable.

McDaniels, Axelrod, and Slovic

578

Table 1. Description of Scales and Response Categories in Order Presented in the Survey
Scale end points

Description
of scale

Low (1)

High (7)

Certainty

Please rate how certain it is that the event has an impact on natural environments

Not at all

Very

Adaptability

Please rate how well natural environments maintain their health and productivity
in response to the current level of each event

Cannot at all

can hlly

Avoidability

Please rate how avoidable the event is, in terms of how easy or difficult it would
be to avoid the Occurrence of the event

Not at all

Completely

Relevance to life

Please rate how relevant the event is to your life, in terms of its impact on natural
environments

No

Direct

Controllability

Please rate how controllable is each event, in terms of peoples ability to control
its impact on natural environments

Not at all

Very

Duration of impacts Please rate the duration of the impacts that each event has on natural environments

Short-term

Long-term

Societal benefits

Please rate how much you think the event may benefit the functioning of your
society

No

Great

Personal benefits

Please rate how much you think that you personally can or do benefit from the
event

No

Great

Scope of impacts

Please rate the scope of the impacts of the event, in terms of the size of the area
affected

Small

Widespread

Number of people

Please rate how many people are, or could be, affected by the impact the event
may have on natural environments

Very few

A great number

Species loss

Please rate the impacts of each event in terms of any potential for loss of animal
or plant species

No species

Many species

Destructiveness

Please rate how destructive the event is or can be, in terms of its impacts on
affected natural environments

No adverse
impacts

Complete
destruction

Emotionality

Please rate how much negative emotion (ie., anger, fear, disgust) you feel when
you think about the event and its impacts on natural environments

No

High

Equitableness of
outcomes

Please rate the equity of each event in terms of whether those who receive the
benefits are the same people who incur the costs

Inequitable

Equitable

Ethicality of event

Please rate how ethical you perceive each event to be, in terms of its impact on
natural environments

Very unethical

Completely ethical

Immediacy of
effects

Please rate the immediacy of each event, in terms of how soon its effects on
natural environments may be experienced

Immediately

Far in the hture

Infringement on
rights

Please rate to what extent the event infringes on the rights of nonhuman species

Does not

Greatly

are specific to ecological risk (e.g., ability of natural environments to adapt, species loss). In addition, two other
types of scales were derived. First, in keeping with recent research,(14J5)
some focus group members suggested
that the benefits associated with an item may influence
the perceived risk associated with that item. Thus, three
scales were included (i.e., benefits to society, benefits to
persons, and overall goodness). Second, ethical dimensions were mentioned in each of the focus groups. In
response, several scales were developed that addressed
these considerations (i.e., ethicality, infringement on the
rights of nonhuman species, extent of suffering to humans and nonhuman species). Finally, the general risk
posed by each item to the health and productivity of

natural environments was assessed. The complete set

of rating scales is shown in Table I along with the response categories as provided to the participants.
2.3. The Questionnaire

The questionnaire began with an introduction that

provided a definition of ecological risk as uncertain
potential for harm to the health and productivity of natural environments. Participants were instructed to rate
each item on judgment scales that ranged from 1 to 7,
with 4 being the midpoint. They rated each of the 65
items on the set of 31 scales, each of which reflected

579

Characterizing Perception of Ecological Risk

Table 1. Continued
Scale end points

Description
of scale
Revetsibility of
impacts

Please rate the extent to which the impacts on natural environments associated
with the event are reversible (i.e., the ability of natural environments to return to
pre-event conditions)

[rreversible

Reversible

Human suffering

Please rate how much human suffering could result ftom the event as a result of
its impact on natural environments
Please rate how much suffering by animals or plants could occur as a result of
the event

No

Great

No

Great

Understandability

Please rate how easy or difficult it is to understand the impacts each event has
on natural environments

Simple to

Hard to

Predictability

Please rate how well impacts on natural environments associated with the event
can be predicted

Not at all

Very

Recognition of
impacts

Please rate how recently potential impacts on natural environments associated

with each event have been recognized by experts

Recently

For a long time

Observability of
impacts

Please rate how observable are the impacts on natural environments associated
with the event

Not at all

Very

Media attention

Please rate how much attention the media has given to the event, in terms of its
impact on natural environments

No

A great deal

Regulatability
of risk

Please rate the extent to which the event can be regulated by governments

Cannot be

Can be completely

Availability of
alternatives

Please rate the extent to which there are reasonable alternatives to the event, or
to the practices that lead to the event

Not available

Are available

Goodness
Human health
risk

Please rate whether you think,in general, the event is good or bad
Please rate the extent to which the event and its impact on natural environments
pose a risk to human health

Very bad

Very good

No risk

A great risk

General acceptability of
event

Please rate the acceptability of each event, in terms of its general impact on
human life and natural environments

Not at all

Completely

General riskiness

Please rate how risky in general you think each event is in terms of its impacts
on the health and productivity of natural environments

Poses no risk

Poses great risk

AnimaVplant
suffering

one characteristic of the items. Participants rated the entire set of 65 items on one characteristic scale before
going on the next scale. A final section of the questionnaire collected demographic information (i.e., age, sex,
major area of study) and attitudinal information.

3. RESULTS
3.1. Mean Ratings
An initial step in analyzing the data was to recode
the 1-7 scale into a scale with endpoints of -3 and 3,
and a midpoint of 0. This recording procedure was done
to highlight the relationship between responses and the
scale midpoint. A second step was to create a data matrix
of mean responses over all individuals, for each com-

bination of scale and item. Table I1 presents the means

and standard deviations for all the scales across all respondents and across the 65 items ordered in terms of
the mean rating. Several scales had means well dispersed
from the scale midpoint. The most extreme mean was
for the certainty of impacts scale (M = 1.54), followed
by the goodness scale (M = - 1.33), the infringement
on the rights on nonhuman species scale (M = 1.31),
and the animallplant suffering scale (M = 1.31). On average, respondents perceived the items to be bad, and,
with a high degree of certainty, to have substantial impact on nonhuman species.
The mean for the animallplant suffering scale (M =
1.31) was higher than the mean for the human suffering
scale (A4= .59, t = 7.20, p C .01). This difference not
only suggests that greater suffering is perceived to occur
in nature as a result of the rated items, but that respon-

580

McDaniels, Axelrod, and Slovic

Table 11. Means and Standard Deviations of Judgment Scale Ratings

Characteristic

Scale end points

Mean

Certainty
Animaliplant suffering
Infnngement on rights
Duration of impacts
General acceptability of risk
Number of people
Availability of alternatives
Destructiveness
Predictability
Scope of impacts
Species loss
Human health risk
Relevance to life
Controllability
Observability of impacts
Regulatability of risk
Emotionality
Human suffering
Recognition of impacts
Avoidability
Media attention
Reversibility of impacts
Immediacy of effects
Adaptability
Equitableness of outcomes
Ethicality of event
General acceptability of event
Societal benefits
Understandability
Personal benefits
Goodness

Not at all + very

No + great
Does not 4 greatly
Short-term + long-term
Poses no risk + poses great risk
Very few + a great number
Not available + are available
No adverse impacts + complete destruction
Not at all -+ very
Small + widespread
No species + many species
No risk -+ a great risk
No -+ direct
Not at all + very
Not at all 4 very
Cannot be + can be completely
No + high
No + great
Recently + for a long time
Not at all -+ completely
No + a great deal
Irreversible --f reversible
Immediate + far in the future
Cannot at all 4 can fully
Inequitable 9 equitable
Very unethical + completely ethical
Not at all -+ completely
No --f great
Simple to -+ hard to
No -+ great
Very bad + very good

1.54
1.31
1.31
1.25
1.13
1.02
1.oo
.90
.8 I
.73
.73
.68
.68

dents were differentiating risks to nature from risks to humans. Another notable difference can be observed by
comparing the social benefit and personal benefit scales,
where on average respondents indicated they benefitted less
personally from the group of items (M = - 1.29) than did
society as a whole ( M = - 1.OO, t = 6.04, p < .01).
In terms of the variability of responses across items,
the highest standard deviations were found for the two
benefit scales (societal = 1.31, personal = 1.32), followed closely by the regulatability scale. The emotionality scale also had a high standard deviation (1.23), with
some items eliciting very negative emotional responses
and other items being perceived as emotionally benign.
The emotionality scale also had the widest range of mean
responses with a low score of -2.43 (outdoor recreation)
as compared to a high score of 2.93 (nuclear war).
Table 111 presents the 65 items ordered in terms of
their mean rating of overall risk to natural environments.
These means were also very diverse, ranging from a low

.64
.64
.64
.60
.59
.39
.27
.I8
-.21
-.71
- .72
-.76
- .83
- .92
- 1.oo
-1.14
-1.29
-1.33

Standard
deviation
.83
.97
I .04
.99
1

.oo

1.01
1.13
.99

.58
1.19
1.05
I .09
.89
I .07
.85
I .30
1.23
1.05
.69
1.19
1.12
.83
.69
.89
.61

I .05
1.10

1.31
.73
1.32
1.05

of - 1.85 to a high of 2.69. On average, though, these

items were perceived to pose a moderate level of risk to
natural environments (M = 1.13) and somewhat less risk
to human health ( M = .68). In terms of specific items,
outdoor recreation ( M = - 1.85), scuba diving ( M =
- 1.78), fireplaces ( M = - 1.42), travel and tourism ( M
= -.86), golf courses ( M = -.72), television ( M =
-.56), and collecting wilderness souvenirs ( M = -.53)
were rated as posing the least risk to natural environments. No other item had a negative rating. In contrast,
nuclear war ( M = 2.69), loss of animal species ( M =
2.53), ozone depletion ( M = 2.51), and loss of plant
species (M = 2.5 1) were rated as posing the highest risks
to natural environments.

3.2. Intercorrelations Among Scales

Table IV presents the intercorrelations among mean
ratings for all 31 scales. A review of the matrix shows

Characterizing Perception of Ecological Risk

581
Table 111. Continued

Table 111. Ratings of 65 Items on Overall Risk to Natural

Environments
ltem
Item

Mean

Nuclear war
Loss of animal species
Depletion of ozone layer

2.69
2.53
2.51

Loss of habitats for animalslfish

Loss of plant species
Deforestation (permanent removal of forest cover)

2.5 1
2.5 1
2.43

Loss of wetlands
Air pollution
Disposal of untreated sewage in oceans

2.42
2.26
2.25

Emission of ozone depleting gases (CFCs)

Clearcutting forests
Climate change (e.g., global warming)

2.22
2.1 1
2.06

Acid rain
Conventional warfare
Production and disposal of toxic chemicals

1.99
1.99
1.99

Belief that humans have dominion over nature

Waste production in modem society
Consumption levels in modem society

1.77
1.68
1.63

Population growth
Lack of regard for nonhuman rights
Nuclear power plants

1.61
1.60
1.57

Intensive commercial fishing

Value system oriented toward material wealth
Aerosol cans

1.50
1.44
1.43

Driftnet fishing
Energy production from nonrenewable resources
Drought

1.39
1.38
1.33

Driving automobiles
Earthquakes
Urbanization (continued growth of large cities)

1.28

Poaching (illegal harvest of wild animals)

Transporting oil
Cigarette smoking

1.26
1.15
1.15

Disposal of treated sewage in oceans or lakes

Burning of waste materials (incineration)
Societys desire for continued economic growth

1.11
1.07
1.07

1.28
1.28

high associations between some scales (e.g., social benefit and personal benefit, r = .96) and no association
between others (e.g., social benefit and availability of
alternatives, r = .03). More than half of the characteristics had correlations of .80 or higher with general riskiness, and five scales (i.e., certainty of impacts,
destructiveness, emotionality, goodness, and acceptability) had correlations of .90 or higher.
The bottom row of Table IV shows that several
scales had relatively low correlations with general risk

Mean

Soil erosion
Floods
Large scale/multinational business

1.06
1.04
1.04

Disposal of municipal waste in landfills

Increasing reliance on technology
Biotechnology (genetically altering plants and animals)

I .03
.99
.99

Development of land for housing

Pesticides
Meteors colliding with Earth

.92
.90
.89

Dams on rivers
Hunting of animals
Volcanos

36
.85
.82

Mass production fanning practices

Disconnection of modem life from natural environments
Beef production

.82
.76
75

Air conditioning
Mining
Capitalism

.75
.74
,151

Fertilizers
Urban water usage
lmgated agriculture

.53
SO
.47
.32
- .53
-.56

Transplanting of animal and plant species

Collecting wilderness souvenirs (e.g., plants, seashells)
Television
Golf courses
Tourism and travel
Fireplaces

- .72
-36
- 1.42

Scuba diving
Outdoor recreation (e.g., skiing, hiking, climbing)

-1.78
-1.85

Scale ranged from -3 (poses no risk) to + 3 (poses great risk)

to nature including avoidability (r = .13), controllability

(r = - .22), ability to regulate (r = .1 l), and availability
of alternatives (r = .21). Interestingly, these four scales
all reflect aspects of societys ability to manage the risk.
The findings of such low associations with risk to nature
is a contrast with studies of perceived risk to humans,
where controllability over the item has been found to be
highly correlated with overall riskiness (e.g., Ref. 16).

3.3. Factor Analysis of Scale Intercorrelations

The matrix in Table IV indicated a substantial degree of correlation for many pairs of scales. This suggests that there may be some underlying dimensions that
could more compactly explain the overall variance in the
data. Factor analysis has been employed in many human

16.-

69'-

E8'
56'

16-

IZ'

5P'Z0'-

LL'-

IZ-

8L.88'

L8'
08'-

5L'

L5'
16

8L'

LL'-

9P'

81'-

8P'

E8'-

W'
18'-

6P'
P5'W'

8L'
9s
99-

II'
EL'
ZS'L9'
E8'

P8'-

EL'56'
18'96'

9E'-

LO'-

0Z'08'-

P6'99'

81'

55'-

EO

EL'
65'6L'

85.

EL'PS'
EL'88'-

OZ'

L5'98'

8P'
15'-

L8'

If5Z'
89

LP'ZS'
L5'-

91'SO'
P5'-

0P'-

09'1E'E5'-

tZ'0Z'ZE'
EE'
9P'-

90'
91'
-

W'9L'
89-

LO'10'
ZL'
-

E5'
LE'
6P'

LL'

ZZ'-

EP'

El'

E6'
08'

68'-

0L'-

~~

W'-

99'

OL'

L8'

56'

ZP'ZL'
95'

8P'-

pp'

L9'

PL'

P8'-

EE'-

t8'

00'

E8'-

58'-

EZ-

ZP'60'-

08'-

8t'

65'8L'

LE'-

LL'
L9'-

ZL'
Z9PL'-

6P'
PS

Z9'-

18'
21'Z5'

85.-

p9'

E9'

E8'

18'80
5P'-

IS'-

LZ
6L'
Po'

51'
2 6-

88'

Z Z88'

E8'

ZL'
65'P8'

PZ

8P'- 19.LS'
Z5'-

EL'E5'
19'-

89'
W '-

18'

8P'

5E'

PZ'
PZ'
6L'

PS'
PL'-

IL'
28'51'

I['LZ'S9'-

0Z'-

L9'
00'

LE'
Z5'

8P'

LE'
61'-

IL'
09'59-

LL'LL'
Z1'-

SI'

SP'

0P'-

58'

P8'

0p'S9'

82'96
-

&I.-

6E'
82'
ZP'
EO
62
95'-

08'

PO'
PZ.
6P'-

W'LE'58'-

9L'28'
LL'

8E'
91'
LP'

9s'

If'9E'61'

8L'
P8'W.-

61'
I I'
LS'

81'
90'
69'

10
51'

LL'PL'
8P'-

I Z'

PI'-

98'
6L'LP'

9L'E9'L8.

25'-

LZ'
PE'

LI'Lt'
SZ'

08'
69
88'-

01'-

PE'LE'
01'

95'PZ'9E'

L2

5E'I0

19-

EZ'
09'

ZS'

06'

80-

L1'LZ'
EO

26L8'E5'-

LE.
Po'
P5'

PL'9P'-

90
PZ
60-

05'09'95'-

L9'

Z8'

LV
LV
LL'

LP'L5'55'-

59'
08'

9Z'Po'11'-

69'

ZI'
Zl'80'-

EC

p9'-

9L'

ZO'

81'
ZO'

5Z'
50'0Z'LP'
OL'

19'ES'22-

8E'

89'

If'
9L'16'

L9'E6'
E9'

60' -

91'-

LO-

8P'-

5s'-

18'SP' - Z 9 - 19'
6 9 - LZ'IV- 6L'
65'
89
65'-

88'

PL'
PL.
06.

58'

El

ZP'

ZE'
ZO'
PS'

LL'

89'

55'
If'
LL'

IZ'0Z'PS'-

LO
90'
8 9 - Z9'
28'19'-

E5'LP'-

62
EL'

P8'85'-

SP'

81'

P8'

08'

05'- 19'EP'
65'

SZ-

OL'

IV- El'-

LL06

85'
19'

PL'-

ss-

P8'8L'
8L'EL'68'

E8'

E6'9L'-

5P'-

IL'-

ZL'
8L'89'-

IZ'
ZL'
59

6L'-

19'
8L'85'-

6E'

IP'
9E'0Z'99'-

8P'
EL'

08'

LP'EZ'19-

8L'EL'E9
55'
65'

88'-

IL'
85'
59'
ZL'85'SL'
85'W-

65'
-

LL'
-

59'69'
09'68'
EP'
59-

99'
Z9'89

Lt'
LZ-

6L'

69'
28'LE'

E6
-

69'
09'
-

EZ'62-

OZ

12-

ssautqsu pauat)

Characterizing Perception of Ecological Risk

583

Table V. Rotated Factor Loadings for 30 Risk Characteristic Scalesa

Characteristic
Species loss
Infringement on rights
AniiaVplant suffering
Destructiveness
Adaptability
Reversibility of impacts
Duration of impacts
Emotionality
Ethicality of event
Certainty

Factor Factor Factor Factor Factor

1
2
3
4
5
.84
.80

.I1
.I3
-.I2
.I 1
.63
-.62
.54

Societal benefits
Personal benefits
Goodness
-.46
Equitableness of outcomes
General acceptability of event -.58
Human suffering
Number of people
Relevance to life
Scope of impacts
Human health risk

.45

-.I2

- .43

.59
-.41
.58
.49
.88
.88
.I6
.I3
.65
-.60

.59

-.59

.91
.80
.I4
.65

.56

Controllability
Avoidability
Availability of alternatives
Regulatability of risk
Observability of impacts
Pre&ctability
Recognition of impacts
Understandability
Immediacy of effects
Media attention
a

SO

.92
.89
.88
.86
.48
-.45
-.41

.45
.48

.41

.I4
.I3
.69
- .69
-.59

.56

Loadings with absolute values below .40 are omitted from the table.
Names for the factors are discussed in the text.

health risk perception studies to identify such dimensionsJ2)Thus, we conducted a factor analysis of the correlation matrix in Table IV.
Table V presents the summary of a principle components factor analysis with varimax rotation performed
on the interrelation among the mean responses for the
30 risk characteristics. Five orthogonal factors with eigenvalues greater than 1.0 emerged from the analysis.
In the unstated results the first factor accounted for
slightly under 56% of the variance in the data; factor 2
accounted for 18%, followed by 9%, 5%, and 3% for
factors 3, 4, and 5, respectively. Even though the last
three factors accounted for substantially smaller amounts
of variance than the first two, they were retained because
of their conceptual relevance and to maintain clarity and
comprehensiveness in the factor structure.

Loss of animal and plant species, infringement on

rights of species and rate of animavplant suffering were
the three highest loaded scales on Factor 1 in Table V.
Thus, we label this factor impact on species. The highest
loaded scales on Factor 2 in Table V were benefits for
society resulting from the items and benefits for individuals. We label this factor human benefits. The third factor included a number of scales related to the extent of
impacts, particularly to humans, including number of
people affected, risks to human health, and scope of impacts. This factor could be viewed as impact on humans,
although from an ecological perspective it could be
viewed as the scope of impacts. We adopt impact on
humans here because of the parallels with the first two
factor labels. The fourth factor had the tightest grouping
of scales and included the four characteristics related to
risk management and control (i.e., controllability, avoidability of impacts, availability of alternatives, ability to
regulate). We label this factor avoidability, though controllability might also be an appropriate label. The fifth
and weakest factor, both in terms of variance explained
and magnitude of factor loading scores, represented the
ability to observe, predict, recognize, and understand the
impacts of the items. We refer to this factor as howledge of impacts.
Factor scores for each item were computed by
weighting the ratings on each risk scale proportionally
to the scales importance in determining each factor and
then summing across all scales, resulting in five factor
scores for each item. Table VI shows the 20 extreme
items (10 highest and 10 lowest) on each factor. As
would be expected, the loss of wetlands, plant species,
animal species, and habitats were perceived as the most
extreme items in terms of Factor 1, impact on species.
In contrast, cigarettes, scuba diving, and fieplaces had
the lowest impact on species. Interestingly, three natural
hazards (i.e., earthquakes, floods, and drought) also were
ranked low on impacts on species. Earthquakes and
floods also appeared as two of the items that ranked
lowest on Factor 2, human benefits. Cigarette smoking
was rated as the least beneficial of the 65 items, whereas
outdoor recreation, housing, travel and automobiles were
perceived as the most beneficial to humans. Turning to
Factor 3, impact on humans, automobiles were ranked
as having the highest impact, followed closely by ozone
depletion, air pollution, and CFC emissions. Somewhat
surprising is the absence of cigarette smoking from the
items ranked high in impact on humans. Collecting wilderness souvenirs, scuba diving, golf courses, poachmg,
and hunting were perceived as lowest in terms of impact
on humans. As expected, the natural hazards were perceived as the least avoidable items (Factor 4). There was

584

McDaniels, Axelrod, and Slovic

Table VI. Ten Highest and Ten Lowest Items fot the Five Factors that Characterize Perceived Ecological Risk"
~~

Factor I
Impact of species
Cigarettes
Scuba diving
Fireplaces
Earthquakes
Television
Recreation
Floods
Drought
Burning of waste
Travel
Climate change
Deforestation
Poaching
Nuclear war
Disregard rights
Belief in dominion
Loss of habitat
Animal loss
Plant loss
Wetland loss

Factor 2
Human benefits

-3.43
- 1.78
- 1.77
I .74
-1.67
- I .57
-1.29
- .95
p.87
- .86
-

34
1.oo
1.03
1.26
1.30
1.56
1.60
2.01
2.03
2.06

Cigarettes
CFC emissions
Conventional war
Ozone depletion
Aerosol cans
Nuclear war
Acid rain
Earthquakes
Floods
Poaching
Technology
Dams
Mass farming
Imgation
Urbanization
Urban water use
Automobiles
Travel
Housing
Recreation

-2.38
-1.60
-1.52
- 1.43
- 1.37
- 1.32
-1.13
-1.10
- 1.08
- 1.03
1.06
1.19
1.22
I .29
1.44
1.78
1.83
2.13
2.29
2.55

Collect souvenirs
Scuba diving
Golf courses
Poaching
Hunting
Fireplaces
Volcanos
Driftnet fishing
Dams
Meteors
Waste production
Monetary values
Urbanization
Technology
Population growth
Climate change
CFC emissions
Air pollution
Ozone depletion
Automobiles

~~~~

Factor 4
Avoidability

Factor 3
Impact on humans

Factor 5
Knowledge of impacts

-2.45
-2.29
-2.28
-1.97
-1.95
-1.66
-1.36
- 1.04
-1.03
- .89

Meteors
Volcanos
Earthquakes
Floods
Drought
Economic growth
Climate change
Population
Capitalism
Technology

-3.45
-2.92
-2.58
-2.57
-2.44
- .99
-.93
- .84
- .78
- .64

1.02
1.14
1.14
1.32
1.36
1.37
1.47
1.55
1.73
I .75

Hunting
Air conditioning
Poaching
Deforestation
Driftnet fishing
Untreated sewage
Clearcutting
Golf courses
Aerosol cans
Cigarettes

.85
36
.87
.92
I.05
I .09
I .09
1.26
1.52
1.55

Television
Biotechnology
Meteors
Scuba diving
Air conditioning
Fireplaces
Big business
Climate change
Monetary values
Disconnection
Hunting
Loss of habitat
Floods
Air pollution
Housing
Drought
Automobiles
Deforestation
Clearcutting
Earthquakes

- 1.84
- 1.74

-1.72
- 1.50
- 1.47
- 1.34
- 1.23

- 1.20
-1.20
-1.18
1.08
1.19
1.39
I .40
1.49
1.51
1.53
1.78
1.90
I .90

" The table entries are factor scores calculated using regression procedures, as described in the text.

also a perception that three dominant forces in North

American life (economic growth, capitalism, and reliance on technology) were not perceived as avoidable. In
addition, population growth was rated as not avoidable.
Smoking cigarettes, use of aerosol cans, golf courses,
and clearcutting of forests were rated as the most avoidable items. Untreated sewage was also seen as quite
avoidable. Several items were perceived as having ecological impacts that are relatively unknown (Factor 5).
These include television, biotechnology, meteors, and
scuba diving. Knowledge regarding global warming was
also considered as low. In contrast, respondents seemed
to think that a good deal is known about the ecological
impacts of earthquakes, clearcutting of forests, deforestation. and automobiles.
3.4. Risk Perception Maps

The relative position of each of the 65 items in

terms of the first two factors can be seen in Fig. 1. The
vertical axis represents Factor 1 (impact on species);
the horizontal axis represents Factor 2 (human benefits). Items at the extreme bottom of Factor 1 are judged
as having little adverse impact on species, whereas
items near the top are perceived as having a high im-

pact. On the horizontal dimension, items at the far right

are construed as offering great human benejits, whereas
items at the far left are seen to offer little or no human
benefits.
Items appearing in the upper right quadrant are
those that have a high impact on species, but are perceived as highly beneficial to humans. Development of
housing is the most extreme item in this quadrant, reflecting the difficult ecological tradeoffs inherent in satisfying this human requirement. Also in this quadrant
are dams, mass farming practices, and urbanization.
Items in the lower right quadrant are also perceived as
beneficial, but are seen as having little effect on natural
environments. Items in this quadrant include outdoor
recreation, travel and tourism, urban water usage, and
automobiles. The lower left quadrant consists of those
items perceived to provide little human benefits and to
have little impact on nature. By far, the most extreme
item in this quadrant is cigarette smolung, seen as having
few benefits and virtually no impact on nature. Also in
this quadrant are four of the five natural hazards. Finally,
items found in the upper left quadrant seem to represent
the practices that are most associated with ecological
risk. These include the loss of animal and plant species,
the loss of wetlands and habitats in general. Other items

585

Characterizing Perception of Ecological Risk

Factor 1 score
Impact on species
3.50

2.50

1.50

Factor 2 score
Human benefits

-1.50

-2.50

-3.50
-3.50

*
-2.50

-1.50

-0.50

0.50

150

2.50

3.50

Fig. 1. Map of Factor 1 vs. Factor 2.

perceived as having a high impact on nature and low

human benefits include nuclear war, poaching, and the
belief that humans have dominion over nature.
The relative position of each item in terms of Factor
1 (impact on species) and Factor 3 (impact on humans)
can be seen in Fig. 2. On this map, the vertical axis
represents Factor 1 and the horizontal axis represents
Factor 3. Items in the upper right quadrant are those
which are construed as posing high impacts on nonhuman species and high impact on humans. This quadrant
consists of the most notable environmental threats including climate change, ozone depletion, population
growth, and nuclear war. The right lower quadrant displays items that have minimal impact on species, but are
seen as having high impact on humans, including cigarettes, television, and air pollution. In the upper left
quadrant are the items that greatly affect species while
at the same time have limited impact on humans, including the loss of animal and plant species, the loss of
wetlands, poaching, and hunting. Finally, in the lower
left quadrant are the items that have minimum impact
on both species and humans, including collecting wilderness souvenirs, golf courses, scuba diving, and fireplaces.

3.5. Relationships Between Factors and Perceived

Risk to Nature
Next, we consider how these factors are correlated
with the respondents' perceptions of the overall riskiness
of items for natural environments. Although it will eventually be important to investigate the relation of this factor structure and expert assessment of ecological risks
associated with each item, currently we only have data
regarding the relation between the factors and our respondents' ratings of general ecological risk. Impact on
species (Factor 1) not only accounted for the greatest
amount of variance in the factor model, it also had the
strongest correlation with general riskiness (r = .58; p
< .01). In addition, perceived human benefits (r =
-.51; p < .Ol), and impact on humans (r = .48; p <
.O 1) were strongly correlated with general ecological
risk.
In contrast with other risk perception studies focused on technological hazards and human health, no
correlation was found between perceived avoidability/controllability and perceived general ecological risk.
This result suggests that avoidability may be seen as
more associated with those activities involved in risk

McDaniels, Axelrod, and Slovic

586
Factor 1 sco?
Impact on speaes
3.50

2.50

150

050

Factor 3 SCOW
Impact on
humans

-0.50

-1.50
etdNmm

-2 50

-3 50
-3.50

-2.50

-1 50

-0.50

--

0.50

1.50

2.50

3.50

Fig. 2. Map of Factor 1 vs. Factor 3.

management (e.g., current levels of regulation, time and

money spent preparing for, and responding to, consequences of the events) than in the absolute judgment of
risk. Another interpretation is that environmental risks
may be viewed as markedly less amenable to risk management efforts than are human health risks. This difference may be attributable to the extent to which
ecological risks are, at a global scale, the result of billions of individual decisions.

3.6. Selected Comparisons

We turn from the overall factor structure to briefly
consider two of many possible comparisons among selected items. These comparisons indicate the kinds of
insights to be drawn from detailed examination of responses among items. One comparison involves two
items concerned with management of sewage. The respondents clearly perceived the disposal of untreated
sewage in oceans as posing more overall risk (M = 2.25)
than the disposal of treated sewage in oceans or lakes
(M = 1.1 1). A review of the factor scores for these items
reveals substantial differences on three of the five factors. Untreated sewage was perceived as having a higher

impact on species (Factor 1) than treated sewage (.41 as

compared to - .3 l), offering fewer human benejits (Factor 2; -.95 as compared to .23), and being more avoidable (Factor 4; 1.09 as compared to S O ) . Regardless of
whether there is an actual difference in ecological risk
stemming from these practices (some scientific research
suggests that marine disposal of untreated sewage may
be benign in certain location~),(~~J~)
there clearly is a
perceptual difference in terms of impacts on species, human benejits, and avoidability. These sorts of findings
may help risk managers understand the public response
to controversial ecological practices, and help them develop effective ways of communicating with the public
regarding these issues.
A second set of comparisons examines indirect and
direct sources of risk, and the consequences of those
risks for ozone-related items (including air conditioning,
aerosol cans, CFC emissions, and ozone depletion). In
terms of overall riskiness, ozone depletion was considered the most risky (M = 2.51) closely followed by CFC
emissions (M = 2.22), its direct cause. Two major
sources of CFC emissions, air conditioning (M = .75)
and aerosol cans (A4 = 1.43) were each perceived as
posing less overall risk than CFCs, which makes sense
in light of the fact that each source is only part of the

Characterizing Perception of Ecological Risk

CFC problem. Ozone depletion was perceived as having
a much higher impact on species (Factor l), factor score
of .34, than were its sources (i.e., CFC emissions, -.33;
air conditioning, -.65; aerosol cans, -.77). Air conditioning was perceived as offering significantly more
human beneJits (Factor 2), factor score of -. 18, than did
the other three items (i.e., ozone depletion, - 1.43, CFC
emissions, -1.61, aerosol cans, -1.37). The large difference on this benefit factor between air conditioning
and aerosol cans may explain the heightened perception
of riskiness associated with aerosol cans as compared to
air conditioning. Ozone depletion was perceived as having the highest impact on humans (Factor 3), factor score
of 1.73, followed by CFC emissions (1.47), aerosol cans
(.81), and air conditioning (.31). In terms of avoidability
(Factor 4), aerosol cans were seen as the most avoidable
(1.52), substantially more avoidable than air conditioning (36) and CFC emissions (.84). Interestingly, the
consequence (i.e., ozone depletion) was perceived as the
least avoidable (.11) of the four items, presumably because causes are more directly controllable than consequences. Knowledge of impacts was low for all four
items, ranging from -.94 for ozone depletion to - 1.47
for air conditioning.
4. DISCUSSION AND CONCLUSIONS

While the limitations of the sample, in terms of size

and representativeness, should be borne in mind, we believe these results provide a plausible conceptual framework for characterizing perceived ecological risk. The
risk maps are simple, yet sensible. Given the complexity
of ecological risk judgments, and the fact that the list of
items and list of scales used in this study were so diverse, it is perhaps surprising that the five factor model
identified here explains as much of the variance in the
respondents judgments as it does.
Amidst the wealth of descriptive detail contained in
the factor maps and other analyses presented here, there
appear to be many notable and sometimes surprising
findings. For example, the differential perceptions of
consequences and their causes suggests the need for additional studies designed to characterizethe mental models responsible for such differences. Also noteworthy is
the finding that natural hazards rate relatively low on the
hierarchy of perceived risks to nature despite the immense damage they are capable of causing. For example,
the meteorite strike thought to have caused the extinction
of the dinosaurs is also believed to have extinguished
50% of the species then in existence. Whereas risk from
meteor strikes may be discounted because of their rarity,

587

damaging floods and droughts are not so rare. In general,

the relatively benign evaluation of natural forces in ecological risk perception parallels the benign view of nature as a contributor to human health risk (see, e.g., Ref.
19). The strong inverse relation between human benefit
(Factor 2) and perception of risk to nature also parallels
results found with human health risk perceptions. Alhakami and S10vic(~~)
have attributed the latter to an inability of people to clearly distinguish risk and benefit,
perhaps due to reliance on affective (goodhad) evaluations as a primary cue from which both risk and benefit
judgments are derived. However, we should caution
against assuming these findings reflect the views of the
general public until larger sample surveys are undertaken
with more representative samples.
While these results are descriptively interesting,
they are also likely to be prescriptively relevant for future ecological risk management efforts. One direct prescriptive use of these results may be to help understand
current controversies about ecological risks (and helping
to predict future ones) by clarifying the factors influencing public risk judgments. A second may be to help clarify key issues that should be emphasized in ecological
risk communication efforts. A third use may be to determine the factors that should be highlighted in programs designed to change individual behavior in
response to ecological risks, or to design societal incentives to foster cooperative efforts in commons dilemmas.
A fourth use would be to provide a starting point for
development of objective hierarchies that characterize
the interests of various groups in public environmental
decision contexts>13)
Future research should build on these results in several ways, akin to the extensions of research on human
health risk perception. One important step would be to
expand the sample size and representativeness, and obtain judgments for several specific societal groups (e.g.,
environmentalists,journalists). A second step would be
to obtain judgments from experts that could be compared
to the lay judgments considered here. Still another focus
might involve cross-cultural comparisons of perceived
ecological risk, and examination of perceived risks in a
specific hazard domain (e.g., risks to water or land resources). With further research, ecological risk perception may prove to be as rich, informative, and enduring
a construct as its human health risk predecessor.
ACKNOWLEDGMENTS
This research was supported by a Tri-Council EcoResearch grant from the Government of Canada to the

588

University of British Columbia, and by an assistance

agreement between the U.S. Environmental Protection
Agency and Decision Research. McDaniels was supported by Westwater Research Centre at the University
of British Columbia. We thank Mike Healey for providing an opportunity to conduct this research. We thank
Bob Clemen, Jim Flynn, Robin Gregory, Steve Hrudey,
Steve Johnson, Don MacGregor, C. K. Mertz, and Terre
Satterfield for advice and encouragement. We thank Milena Calendino for able research assistance. Finally, we
thank those who participated in the focus groups and the
survey for their very considerable contributions, in the
form of thoughts, feelings, and judgments about ecological risks.
Although the research described in this article has
been funded in part by the U S . Environmental Protection Agency under assistance agreement R-822464-01-0
to Decision Research, it has not been subjected to the
Agencys peer and administrative review, and, therefore,
may not necessarily reflect the views of the Agency and
no official endorsement should be inferred.

REFERENCES
1. P. Slovic, Perception of Risk, Science 236, 28&285 (1987).
2. P. Slovic, Perception of Risk: Reflections on the Psychometric
Paradigm, in S. Krimsky and D. Golding (eds.), Social Theories
of Risk (New York, Praeger, 1992), pp. 117-152.
3. National Research Council (NRC), Improving Risk Communication (National Academy Press, Washington, D.C., 1989).

McDaniels, Axelrod, and Slovic

4. W. J. Burns, P. Slovic, R. E. Kasperson, J. X. Kasperson, 0. Renn,
and S. Emani, Incorporating Structural Models into Research on
the Social Amplification of Risk: Implications for Theory Construction and Decision Making, Risk Anal. 13, 6 1 1 4 2 3 (1993).
5 . R. Kasperson, 0. Renn, P. Slovic, H. Brown, J. Emel, R. Goble,
J. Kasperson, and S. Ratick, The Social Amplification of Risk:
A Conceptual Framework, Risk Anal. 8, 177-187 (1988).
6. R. Gregory, J. Flynn, and P. Slovic, Technological Stigma, Am.
Scientist (in press).
7. T. McDaniels, M. Kamlet, and G. Fischer, Risk Perception and
the Value of Safety, Risk Anal. 12, 495-503 (1992).
8. R. Lackey, Ecological Risk Assessment, Symposium on Critical
Issues in Risk Assessment, New Orleans, April (1993).
9. G. Suter, Ecological Risk Assessment (Boca Raton, Louisiana.
Lewis, 1993).
10. C. Travis and J. Moms, The Emergence of Ecological Risk Assessment, Risk Anal. 12, 167-168 (1992).
11. B. Fischhoff, S. Lichtenstein, P. Slovic, S. Derby, and R. Keeney,
Acceptable Risk (Cambridge University Press. New York. 1981).
12. S. Rayner and R. Cantor, How Fair is Safe Enough? The Cultural
Approach to Societal Technology Choice, Risk Anal. 7, 3-13
(1987).
13. B. Fischhoff, S. Watson, and C. Hope, Defining Risk, Policv
Sciences 17, 123-139 (1984).
14. A. Alhakami and P. Slovic, A Psychological Study of the Inverse
Relationship Between Perceived Risk and Perceived Benefit,
Risk Anal. 14(6), 1084-1096 (1994).
15. R. Gregory and R. Mendelsohn, Perceived Risk, Dread, and
Benefits, Risk Anal. 13, 259-264 (1 993).
16. D. MacGregor and P. Slovic, Perception of Risk in Automotive
Systems, Human Factors 31, 377-389 (1989).
17. National Research Council (NRC), Managing Wastewater in
Coastal Urban Areas (National Academy Press, Washington,
D.C., 1993).
18. Capital Regional District, Sewage Discharge: Eflecrs on the OHshore Marine Environment, Liquid Waste Management Plan Information Paper No. 4, Victoria, British Columbia, Author,
September (1992).
19. N. Kraus, T. Malmfors, and P. Slovic, Intuitive Toxicology:
Expert and Lay Judgments of Chemical Risks, Risk Anal 12,
215-232 (1992).