Gender Differences in Risk Assessment: Why do Women Take Fewer Risks than Men?

Text-only Preview

Judgment and Decision Making, Vol. 1, No. 1, July 2006, pp. 48–63
Gender Differences in Risk Assessment: Why do Women Take
Fewer Risks than Men?
Christine R. Harris?, Michael Jenkins
University of California, San Diego
and Dale Glaser
Glaser Consulting Firm, San Diego
Across many real-world domains, men engage in more risky behaviors than do women. To examine some of the
beliefs and preferences that underlie this difference, 657 participants assessed their likelihood of engaging in various
risky activities relating to four different domains (gambling, health, recreation, and social), and reported their perceptions
of (1) probability of negative outcomes, (2) severity of potential negative outcomes, and (3) enjoyment expected from
the risky activities. Women’s greater perceived likelihood of negative outcomes and lesser expectation of enjoyment
partially mediated their lower propensity toward risky choices in gambling, recreation, and health domains. Perceptions
of severity of potential outcomes was a partial mediator in the gambling and health domains. The genders did not differ
in their propensity towards taking social risks. A ?fth domain of activities associated with high potential payoffs and
?xed minor costs was also assessed. In contrast to other domains, women reported being more likely to engage in
behaviors in this domain. This gender difference was partially mediated by women’s more optimistic judgments of the
probability of good outcomes and of outcomes being more intensely positive.
Keywords: sex differences, gender differences, risk perception
1 Introduction
drowning or accidental poisoning throughout the West-
ern world (Waldron, et al., 2005). Thus, there seems little
Accidents are a very frequent cause of death, particularly
doubt that men must be engaging in more risky behaviors
among young adults and teenagers (U.S. Center for Dis-
across a broad range of domains.
ease Control [CDC], 2004), and men are more often the
Despite its obvious practical importance, some key as-
victims of accidents than are women (CDC, 2004; Wal-
pects of the psychological underpinnings of gender dif-
dron, McCloskey, & Earle, 2005). For example, for ev-
ferences in risk taking have not been examined. The
ery 100,000 US drivers, men are three times as likely as
present article seeks to shed new light on these under-
women to be involved in fatal car accidents (U.S. De-
pinnings, by asking a substantial sample of college men
partment of Transportation, 2004). While some of this
and women to report various perceptions and preferences
well-known difference in automobile death rates prob-
related to a wide range of risk-taking scenarios.
ably re?ects differences in the average amount of time
men and women spend driving, it seems likely that an-
1.1 Gender differences in risk taking and
other important cause is that males voluntarily engage in
risk perception
risky behaviors more often than do females. For exam-
ple, US women report usually using seat belts substan-
The existence of gender differences in propensity to take
tially more often than men (Waldron, et al., 2005), and
risks has been documented in a large number of ques-
men have been shown to run yellow lights more often
tionnaire and experimental studies. For example, a meta-
than women (Konecni, Ebbesen, & Konecni, 1976). Fur-
analysis by Byrnes, Miller, and Schafer (1999) reviewed
thermore, similar differences are seen in a wide variety of
over 150 papers on gender differences in risk perception.
other forms of accident statistics. Male pedestrians in the
They concluded that the literature “clearly” indicated that
UK are involved in accidents about 80% more often than
“male participants are more likely to take risks than fe-
female pedestrians, and men die much more often from
male participants” (p. 377).
Recent work has begun to examine the generality and
?Correspondence concerning this article should be addressed to
cognitive underpinnings of these differences in greater
Christine R. Harris, Department of Psychology, University of Califor-
nia San Diego, 9500 Gilman Drive #0109, La Jolla, CA 92093–0109.
detail (Slovic, 1997). In one important study that pro-
Email: [email protected]
vides a backdrop for the present investigation, Weber,

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
Blais, and Betz (2002) assessed the risks that men and
Note that in the ?eld of ?nance, where distribution of
women perceived in behaviors spanning ?ve different
potential outcomes is obviously continuous, risk is often
content domains (?nancial, health/safety, recreational,
conceptualized as the variability of the returns offered
ethical, and social decisions). Gender differences were
by a choice. Following that approach, some theorists
found in four of the ?ve domains — social decision-
have found it useful to conceive of people’s generalized
making being the exception — with males perceiving
risk preferences in terms of how this variability affects
less risk and indicating a greater likelihood of engag-
an individual’s disposition to choose an option (see We-
ing in risky behaviors. Similar gender differences have
ber, 1999, for a discussion). While this seems quite rea-
been found in these domains in a large German sample
sonable, in many real world risky choice scenarios (e.g.,
(Johnson, Wilke, & Weber, 2004). Across studies, the so-
riding motorcycle without helmet; not using sunscreen;
cial domain is unique in that either no gender differences
etc.), it would seem to be a reasonable simpli?cation to
are found or when they are found, it is women who re-
view the potential negative outcomes as a unitary event,
port greater propensity to engage in risky behaviors and
having a probability and some degree of (un-)desirability.
perceive overall greater bene?t and less risk in doing so
This approach will be followed here, although in the Gen-
(Johnson et al., 2004; Weber et al., 2002). Of interest,
eral Discussion we will point out the potential for follow-
these authors also found great variability in an individ-
up work that would consider risks involving more than a
ual’s willingness to engage in risk across domains, sug-
single discrete negative outcome.
gesting that risk taking is not simply the product of some
Remarkably, the literature with adults does not seem
general personality trait that promotes risk seeking. In-
to contain any studies that seek to decompose the per-
stead, individual and group differences are substantially
ceptions of risk involved in real-world risky behaviors,
due to differing perceptions of risk in different domains.
in order to determine whether the genders differ in their
For the most part, previous research has relied on a
evaluations of the likelihoods and costs of negative out-
unitary and subject-de?ned notion of “risk” (e.g., “how
comes. A number of plausible hypotheses immediately
risky is the behavior or situation?”). A number of re-
present themselves. One such hypothesis is that women
searchers have examined the role of various affect di-
do not evaluate the probability of negative outcomes dif-
mensions in determining overall perceptions of riskiness.
ferently than men; they simply assume (perhaps rightly;
Slovic (1997) proposes that several psychological risk di-
perhaps not) that they would be more emotionally upset
mensions (including dread, control, and knowledge) con-
or harmed by negative outcomes, should these occur. Al-
tribute to perceived riskiness. Follow-up research has
ternatively, one may hypothesize that women assess as
shown the material as well as emotional factors also im-
greater the probability of unfavorable outcomes, without
pact overall risk judgments (Holtgrave & Weber, 1993).
projecting any stronger negative reactions to these out-
Any global assessment of perceived risk combines el-
comes than do men.
ements of a belief (“how likely is it that something bad
While studies of gender effects in adult risk prefer-
will happen?”) and a subjective valuation of that outcome
ences — with the exception of Gurmankin Levy and
(“how bad would that be?”). Thus, in common parlance
Baron (2005) — have not addressed this issue, there is
a given behavior might be said to be riskier than another
one study within the developmental literature that ex-
behavior if the former has more severe potential conse-
plored this question. Hillier and Morrongiello (1998)
quences, or if it has a higher risk of potential negative
examined gender differences in perceptions involved in
consequences, or both. For example, leaving one’s bike
physical risk taking in children. Using pictorial descrip-
unattended for a day in a busy city, and bungie jumping
tions (e.g., riding bicycle with no helmet in street) and an
could both be described as risky behaviors, and yet the
interview to determine how children assessed risks, they
probabilities and potential bad outcomes are enormously
found that girls appraised more general risk (i.e., judged
different in the two cases. Past research shows that de-
the situations as more unsafe) than boys. The genders
composing these elements can shed important light on
also differed in the factors that contributed to their over-
individual and group differences in responses to risky sit-
all risk judgments. Boys’ risk judgments were signi?-
uations. Gurmankin Levy and Baron (2005) had subjects
cantly predicted by their ratings of injury severity while
assess badness of unfortunate medical outcomes associ-
girls’ risk judgments were better predicted by their rat-
ated with a de?ned probability (e.g., 32% chance of loss
ings of vulnerability to any type of injury. This suggests
of a big toe). Different groups (men vs. women; physi-
that girls may avoid risky situations with any likelihood
cians vs. non-physicians) were differentially sensitive to
of perceived injury and boys may avoid risky situations
probability as against severity. The present article pur-
only if the possible perceived injuries are judged as being
sues a similar approach to explore the determinants of
men’s and women’s willingness to engage in different
As noted above, the literature with adults has not exam-
risky activities.
ined whether the genders differ in their evaluations of (1)

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
the likelihood of potential negative outcomes and (2) their
low-frequency outcomes (whether good or bad) as more
appraisals of the severity of these potential outcomes. In
likely to occur, in which cases they should show greater
adults, either or both of these aspects of risk may mediate
attraction to choices in the positive domain.
gender differences in engaging in “risky” behaviors. A
third factor may also be responsible for the gender differ-
ences in propensity to engage in risky behaviors: the gen-
2 Method
ders may differ in their estimates of the enjoyment offered
by the activity, assuming that negative outcomes do not
take place. This last possibility ?nds some support from
2.1 Participants
Weber et al. (2002) and Johnson et al. (2004), who found
A sample of 657 subjects (389 female and 268 male) from
that relative to women, men judged they would obtain
undergraduate psychology classes at the University of
greater bene?ts from engaging in risky behaviors in all
California, San Diego participated in the study for course
domains except social.1 Using a risk-return framework,
credit. Their average age was 18.5 years. Three addi-
Weber and colleagues have suggested that risky decision
tional subjects participated but were excluded because
making can be seen as a trade-off between fear (risk) and
they did not indicate their gender.
hope (expected returns).
1.2 Present study
2.2 Survey design
The present study had two major goals. The ?rst was to
Sixteen of the risk behavior scenarios consisted of a sub-
separately assess gender differences in the three kinds of
set of those used by Weber et al. (2002). These fell into
assessments just mentioned. To put it in simple terms,
4 domains: gambling (e.g., betting at a race track), health
the present study asks: do women tend, for example, to
(e.g., deciding whether or not to use sunscreen), recre-
engage in dangerous recreational activities less often be-
ational (e.g., engaging in an extreme sport such as moun-
cause (a) they think the likelihood of injury is greater,
tain climbing), and social decisions (e.g., discussing op-
(b) they think the severity of an injury, were it to occur,
posing viewpoints with a friend). For each domain, we
would be greater, and/or (c) because they simply do not
chose the four items that had the highest risk perception
?nd the positive aspects of such activities as attractive as
factor loadings in Weber et al. (2002). Given the mixed
men do? In addition, we examined whether such assess-
results regarding gender differences in the social domain
ments vary depending upon the domain of behavior and
reported by Weber et al. (2002) and Johnson et al. (2004),
compared patterns of risk perception with individuals’ re-
two additional social domain scenarios were created for
ports of engaging in risky behaviors in the past.
the current work to further examine potential gender dif-
A second aim was to explore an important category of
ferences in this domain. These items were designed to
choices (popularly referred to as “taking a chance”) that
include behaviors that while having potential social risk
have not, to our knowledge, been examined in previous
also had potential social bene?t. For each scenario (listed
studies of individual differences in risk: decisions to en-
in Appendix A), subjects rated (1) their likelihood of en-
gage or not engage in behaviors that offer a small proba-
gaging in the activity, (2) the probability of a risky be-
bility of a large positive reward in return for some small
havior incurring negative consequences, (3) the severity
but certain cost. An example is trying to be the 12th caller
of these potential consequences, should they occur, and
to a radio station in order to win a large sum of money.
(4) how positive or enjoyable the given activity would be,
This type of scenario will be referred to as the “positive
if there were no bad outcomes. Following Weber et al.
domain”. One possible explanation for why women en-
(2002), subjects responded to the likelihood of engag-
gage in fewer risky activities is that they are relatively
ing question with a 5-pt. scale (1 = very unlikely; 5 =
pessimistic and feel themselves relatively “unlucky” (i.e.,
very likely). The three additional questions were also an-
prone to experience the least desirable possible outcome
swered on a 5-pt. scale (1 = not at all; 5 = extremely).
more often than would be expected based on overall fre-
An additional set of questions assessed possible gen-
quencies). If this is so, then women should also show less
der differences in relation to choices associated with high
interest than men in options offering a low probability of
potential payoffs and relatively minor but certain costs,
positive reward. Another possibility is that women see
referred to as the “positive domain”. An example would
1It should be noted that Weber et al. (2002) did not ask subjects to
be calling a radio station to win money. For each scenario
assess the bene?ts of risky behaviors conditionalized on the absence of
(see Appendix B), subjects rated (1) their likelihood of
any negative outcomes; hence, it is possible that in giving their judg-
engaging in the activity, (2) the likelihood of the behavior
ments about positive bene?ts, respondents were “folding in” the risks,
thus potentially explaining why females might have given lower scores
incurring positive outcomes, (3) the intensity of these po-
on this.
tential positive consequences, should they occur, and (4)

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
the degree of unpleasantness of the activity, if there were
Relative to women, men reported a greater overall like-
no good outcomes.2
lihood of engaging in risky behaviors in the gambling,
Finally, additional questions dealing with risky past
health, and recreational domains. In all three domains,
behaviors were created for the present study, including
women judged potential negative consequences as more
some that were adapted from Gibbons and Gerrard (1995)
likely to occur and they judged the potential negative con-
(see Appendix C). Subjects were asked how frequently
sequences as signi?cantly more severe in two of these
they had actually engaged in behaviors that correspond to
domains (gambling and health). The genders also signi?-
the four negative domains of gambling, recreation, health,
cantly differed in their ratings of the enjoyment of engag-
and social.
ing in risky behaviors (assuming no negative outcome) in
all three domains, with men rating the scenarios as more
2.3 Procedures
The social domain showed a very different pattern of
Subjects were recruited from the UCSD psychology sub-
responses than the three domains just described. There
ject pool and completed questionnaires through a spe-
was no overall gender difference in reports of likelihood
cially created web program that was generated using PHP.
of engaging in behaviors carrying social risks. An ex-
The scenarios listed in Appendix A were presented in a
amination of individual items suggested that the gender
random order and subjects assessed their likelihood of en-
differences were not consistent in direction. For exam-
gaging in each described behavior. These scenarios were
ple, women reported signi?cantly greater propensity for
then presented a second time in a random order and sub-
taking risks on two scenarios (admitting tastes are dif-
jects answered the three additional risk questions (prob-
ferent than friends’; disagreeing with parent on a major
ability of negative outcomes, severity of negative out-
issue) while men reported signi?cantly greater propen-
comes, and enjoyment). Two practice scenarios appeared
sity on two different scenarios (defending unpopular is-
before the actual stimuli to familiarize the subjects with
sue; asking someone on a date) as well as a signi?cant
the types of scenarios and the response scales. The pos-
trend (p = .06) on a third scenario (arguing with a friend).
itive domain scenarios were presented next and followed
There were also no gender differences in overall ratings
the same procedures as the negative domain (e.g., like-
of likelihood of negative consequences or enjoyment of
lihood of engaging in the activity was ?rst assessed and
the behaviors. However, women did rate the severity of
then the scenarios were presented a second time with the
possible negative consequences as greater than men for
three additional questions about outcomes). Lastly, sub-
this domain as a whole.3
jects answered questions regarding past risky behavior.
The positive domain — behavioral choices offering a
chance of substantial gain and imposing a relatively small
but certain cost — is one that has not to our knowledge
3 Results
been examined in any previous studies of gender dif-
ferences and risk. In contrast to the ?ndings from the
3.1 Basic gender differences
domains described above, women reported being more
likely to engage in these behaviors. They also gave sig-
For each type of question (willingness to engage in be-
ni?cantly higher probability estimates for positive conse-
havior, perceived bene?ts, etc.), an individual’s responses
quences occurring and showed a trend towards reporting
to the scenarios composing each domain were averaged
that the potential favorable consequences would be more
together to form a composite score for that domain. As
positive. The genders did not signi?cantly differ in their
noted above, the categorization followed Weber et al.
assessments of degree of unpleasantness associated with
(2002). All the analyses described below were performed
the costs incurred by these behaviors.
on these mean responses. For each negative risk domain
(gambling, health, recreation, and social), four separate t-
tests were performed to determine the existence of gender
differences in perceptions of (1) likelihood of engaging;
3.2 Gender differences in reports of past
(2) probability of negative consequences due to engag-
risky behaviors
ing; (3) severity of potential negative consequences; and
(4) enjoyment. The overall mean responses for each type
The frequency of reporting engaging in speci?c risky be-
of question in each domain by gender are shown in Table
haviors as a function of gender is shown in Table 2. Every
1. T-tests were also performed on the positive domain for
each question type and are shown in Table 1.
3Results from analyses using just the four original items from We-
ber et al. revealed the same pattern of results with the exception that
2 Subjects also completed additional questions on other topics not
the gender difference in predictions of severity of outcome no longer
reported here.
remained signi?cant.

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
Table 1: Means (SD) of gender differences in risk perceptions by domain and question type.
Likelihood of engaging in risky behavior
1.84 (0.94)
1.45 (0.67)
t(654) = 6.18***
Probability of negative consequences
3.66 (0.74)
3.88 (0.73)
t(654) = 3.69***
Severity of potential negative consequences
3.62 (0.88)
3.77 (0.81)
t(654) = 2.33*
Enjoyment of experience
3.88 (1.04)
3.41 (1.24)
t(654) = 5.14***
Likelihood of engaging in risky behavior
2.58 (0.66)
2.25 (0.63)
t(654) = 6.40***
Probability of negative consequences
2.99 (0.74)
3.50 (0.73)
t(654) = 8.80***
Severity of potential negative consequences
4.24 (0.61)
4.48 (0.50)
t(654) = 5.52***
Enjoyment of experience
2.44 (0.83)
2.31 (0.73)
t(654) = 2.01*
Likelihood of engaging in risky behavior
2.96 (0.91)
2.54 (0.91)
t(654) = 5.75***
Probability of negative consequences
3.07 (0.72)
3.37 (0.65)
t(654) = 5.67***
Severity of potential negative consequences
4.37 (0.67)
4.42 (0.62)
t(654) = 1.09
Enjoyment of experience
4.17 (0.85)
3.98 (0.91)
t(654) = 2.78**
Likelihood of engaging in risky behavior
3.53 (0.61)
3.45 (0.59)
t(654) = 1.81†
Probability of negative consequences
2.46 (0.63)
2.54 (0.60)
t(654) = 1.58
Severity of potential negative consequences
2.58 (0.69)
2.69 (0.65)
t(654) = 2.05*
Enjoyment of experience
3.31 (0.69)
3.28 (0.70)
t(654) = 0.40
Likelihood of engaging in behavior
2.94 (0.86)
3.23 (0.83)
t(655) = 4.34***
Probability of positive consequences
3.23 (0.58)
3.40 (0.60)
t(655) = 3.57***
Intensity of potential positive consequences
4.48 (0.57)
4.56 (0.50)
t(655) = 1.84†
Unpleasantness of experience
2.72 (0.74)
2.68 (0.77)
t(655) = 0.69
†p < .10, *p < .05, **p < .01, ***p < .001

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
Table 2: Gender differences in reports of actual past risky behaviors.
Risk Behavior Questions
Males Means (SDs) Females Means (SDs)
Gender Difference
Do you smoke?
1.30 (0.73)
1.17 (0.55)
t(654) = 2.48, p < .013*
How many alcoholic beverages do
1.99 (1.16)
1.67 (0.80)
t(652) = 4.13, p < .001**
you typically drink in a week?
How often have you had too much
2.72 (1.51)
2.40 (1.35)
t(654) = 2.90, p < .004**
to drink or gotten drunk?
How often do you drive over the
3.91 (1.08)
3.65 (1.15)
t(653) = 2.90, p < .004**
speed limit?
How often do you “bend” or break
3.13 (1.14)
2.85 (1.08)
t(654) = 3.30, p < .001**
traf?c laws?
How often do you gamble?
2.18 (1.12)
1.47 (0.81)
t(653) = 9.53, p < .001**
How often do you engage in risky
2.21 (1.11)
1.73 (0.92)
t(655) = 5.99, p < .001**
recreational activities?
How often do you get into argu-
2.48 (1.00)
2.36 (0.95)
t(655) = 1.58, p < .115
ments with friends or family?
How often do you raise your hand to
2.28 (1.12)
2.02 (1.00)
t(655) = 3.22, p < .001**
answer or ask questions in class?
*p < .05, **p < .01.
Note. Some n’s may be slightly reduced for some individual analyses due to missing data points.
category of behavior showed a signi?cant gender differ-
3.4 Correlations between different judg-
ence with the exception of one question associated with
ments of risk
the social domain.
How are perceptions of the likelihood of and the severity
of negative outcomes related? This can be addressed by
3.3 Correlations of risk perceptions and
examining a correlation computed across subjects, ask-
past risky behaviors
ing “do people who rate a behavior as risking a severe
outcome rate this outcome as more — or less — likely
How are subjects’ assessments of likelihood of engaging
to occur?” As shown in Table 4, those rating the bad
in risky behaviors in a given domain related to the fre-
outcomes as severe also rated them as more probable.
quency with which they have actually engaged in risky
Across different domain scales, all 18 correlations were
behaviors in that domain? Judgments of the likelihood
positive (all but one being statistically signi?cant), with
of engaging in risky behaviors in the recreational domain
an average correlation of .40. These positive correlations
were signi?cantly related to the responses on an actual
were present both in the sample as a whole, and within
risk behavior question in the same domain, r(656) = .588,
the male and female subsets of the population considered
p < .001, as were responses regarding likelihood of en-
separately. We also examined the relationship between
gage in risky gambling behavior and reports of past gam-
evaluations of the enjoyment associated with an activity
bling risk behavior, r(654) = 0.582, p < .001. Signi?-
and the probability and the severity of potential nega-
cant associations between predicted and actual behavior
tive outcomes. There was a weak negative relationship
also were found in the health domain, as shown in Table
between enjoyment and probability (across items, corre-
3. Reports of likelihood of engaging in risky behavior in
lations ranged between -.23 and .02, averaging -.10; of
the social domain were signi?cantly associated with past
these, 10 were signi?cant, all in a negative direction).
socially risky behaviors: r(656) = .41, p < .001 for the
There was no discernible consistent relationship between
question regarding raising ones hand in class and r(656)
pleasure and severity (across items, correlations ranged
= .25, p < .001 for the question regarding getting into
from -.21 and .12, averaging 0; of these, 6 were sig-
ni?cant, 4 in a positive direction, 2 in a negative direc-

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
Table 3: Health domain: correlation between reports of actual past risk behavior and likelihood of engaging in risky
Past risk behavior
Average likelihood of engaging
Do you smoke?
How many alcoholic beverages do you typically drink in a week?
How often have you had too much to drink or gotten drunk?
How often do you drive over the speed limit?
How often do you bend or break traf?c laws?
* p < 0.01, ** p < 0.001.
tion). In summary, people who evaluate potential harms
the independent variable and the mediator variable are in-
as likely also have a marked tendency to also evaluate
cluded as predictors of the outcome variable. Evidence of
them as being severe; however, assessing the activities as
mediation exists, if the effect of the independent variable
enjoyable says little or nothing about whether a person
is reduced in this third equation, a reduction that can be
will view the potential negative outcomes as likely or se-
tested by Sobel’s test. We applied this strategy to each of
the domains described here for each of the potential me-
Similar questions can be posed in the positive domain.
diators separately. For simplicity sake, we only present
First, do those viewing the positive rewards as greater
the Sobel test statistic for these analyses as well as corre-
also think them more probable? Here, the answer varied
lations and partial correlations between likelihood in en-
(see Table 5), with the correlations between judgments
gaging in risky behavior, gender, and mediators for each
of probability of good outcomes and intensity ranging
domain (see Table 6).
from -.07 to .47, averaging .20. Ratings of the antic-
ipated unpleasantness (costs) were not correlated with
For both the gambling and health domains, separate
probability of positive outcomes (average correlation =
analyses of each mediator revealed that perceptions of
.02, none signi?cant). Anticipated unpleasantness was
probability of negative consequences, severity of poten-
signi?cantly negatively correlated with intensity of posi-
tial negative consequences, and enjoyment each partially
tive consequences for only one of the four scenarios, and
mediated the gender effect in risky gambling behavior. In
the average correlation for the four scenarios was -.05.
the recreational domain, the gender difference in risk tak-
ing was partially mediated by perceptions of likelihood
3.5 Mediation analysis
of negative consequences and partially mediated by per-
ceptions of enjoyment from engaging in such behaviors.
The analyses reported above show that in regard to gam-
Perceptions of severity of negative consequences were
bling, health, and recreational domains — but not so-
not analyzed since they were not signi?cantly correlated
cial domains — women tend to judge negative outcomes
with gender. The genders did not signi?cantly differ in
associated with risky behaviors as both more likely and
their average willingness to engage in social risk, there-
more severe; they also indicate a lower likelihood of en-
fore mediational analyses were not performed in this do-
gaging in these risky behaviors and judge the activities as
less enjoyable than do men (assuming that the negative
outcomes do not occur). Do these perceptions mediate
the gender differences in reported likelihood of engaging
Next we examined mediation in the positive domain,
in risky behaviors?
where potential payoffs were high but uncertain, and
To test for mediational effects, we began with the
costs were low. Unlike the most of the negative do-
commonly used approach laid out by Baron and Kenny
mains, women reported being more likely to engage in
(1986). Each mediational analysis requires three regres-
these types of behaviors. This difference was partially
sion equations. The ?rst tests for a signi?cant relation-
mediated by perceptions of probability of positive con-
ship between the independent variable and the mediator.
sequences. Intensity of positive consequences was also
The second looks at the relationship between the media-
a partial mediator, although only marginally so. Per-
tor and the outcome variable. If both of these correlations
ceptions of unpleasantness were not analyzed since they
are signi?cant, a third equation is computed in which both
were not signi?cantly correlated with gender.

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
Table 4: Correlations of judgments of probability of negative consequences, severity of negative consequences, and
enjoyment of activity for each item within each risky domain.
Probability and severity
Probability and enjoyment
Severity and enjoyment
Gambling domain
1 (sport event)
2 (horse races)
3 (poker)
4 (casino)
1 (walking home)
2 (seatbelt)
3 (helmet)
4 (sun exposure)
1 (rafting)
2 (sport)
3 (plane)
4 (tornado)
1 (tastes)
2 (disagreeing)
3 (defending)
4 (arguing)
5 (date)
6 (raising hand)
† p < .10, *p < .05, **p < .01, ***p < .001
Table 5: Correlations of judgments of probability of good outcomes, intensity of good outcomes, and unpleasantness
of activity for each item within the positive domain.
Probability with
Probability of Good Outcomes Intensity of Good Outcomes
Intensity of Good Outcomes
with Unpleasantness
with Unpleasantness
1 (screenplay)
2 (radio station)
3 (applications)
4 (visiting)
† p < .10, ***p < .001.

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
Table 6: Analyses of mediators of risk taking for each domain, with zero-order and partial correlations.
Zero-order correlations
Partial correlations
Sobel’s Test
Risk taking
Risk taking
Negative probability
Negative probability
z = -3.08, p < .003
Risk taking
z = -2.23, p < .03
Risk taking
z = -4.08, p < .001
Risk taking
Negative probability
Negative probability
z = -4.81, p < .001
Risk taking
z = -2.51, p < .02
Risk taking
z = -1.92, p = .055
Risk taking
Negative probability
Negative probability
z = -4.45, p < .001
Risk taking
Risk taking
z = -2.76, p = .01
Risk taking
Negative probability
Negative probability
Risk taking
Risk taking
Risk taking
Positive probability
Positive probability
z = 3.30, p < .001
Risk taking
Positive intensity
Positive intensity
z = 1.70, p = .09
Risk taking
Risk taking
† p < .10, *p < .05, **p < .01, ***p < .001., ?? criteria for mediational analyses not met, for gender M = 1 F = 2.

Judgment and Decision Making, Vol. 1, No. 1, July 2006
Gender differences in risk assessment
3.6 Exploratory path analysis
of engaging in risky behaviors. Judged severity of po-
tential negative consequences was an additional partial
As a further check on the conclusions just described re-
mediator of the gender differences in engaging in risky
garding mediation, we utilized a path analysis (SEM us-
behaviors in the health and gambling domains.
ing LISREL 8.54) to test a model in which perceptions
The social domain showed more mixed results, as was
of probability of negative outcomes, severity of nega-
the case in the data of Weber et al. (2002). In one study,
tive outcomes, and perceived enjoyment are assessed as
they found that women reported greater propensity to-
potential mediators of gender differences in risk taking
wards taking social risks but in a second study this dif-
for each of the four different content domains. This
ference was not signi?cant. In a German sample, John-
framework assumes that the variables combine additively
son et al. (2004) also did not ?nd a sex difference in so-
with each other to determine the target variables. How-
cial risk taking, although women did perceive such ac-
ever, there is no decision-theoretic model that we are
tivities as providing greater bene?ts. It is interesting that
aware of that would predict that probability would com-
the genders do not show consistent differences with re-
bine additively with severity to determine an individual’s
spect to social risks, as they do in the other domains.
propensity to engage in a behavior.4 Thus, a path an-
Looking over the individual items, it appeared that men
alytic approach is perhaps best viewed as exploratory
tended more often to describe themselves as likely to en-
(Raykov & Marcoulides, 2000). Nonetheless, the results
gage in behaviors that could be perceived as ‘defending’
mirrored the individual Sobel mediation tests described
ideas (e.g., “Defending an unpopular issue that you be-
above quite closely.
lieve in at a social occasion”) whereas women appeared
to respond more positively than men to behaviors that in-
3.7 Full model regression analyses
volved social risks, but which were not phrased in this
way (e.g., “Admitting that your tastes are different from
The ?nal analyses focused on full regression models
those of your friends”). Indeed, men scored signi?cantly
where likelihood in engaging in risky behaviors was re-
higher on the former while women scored signi?cantly
gressed on gender, probability, severity and enjoyment
higher on the latter question in the social domain. This
for each domain. These results are presented in Table 7.
suggestion is obviously tentative, however; a more ?ne-
When included together, all four variables signi?cantly
grained analysis of the particular risks and bene?ts at is-
predicted risk taking in the gambling and recreational do-
sue in “risky” social decisions is plainly needed in order
mains. In the health domain, all variables except severity
to better characterize gender differences. What is clear
were signi?cant predictors of risk taking. Social risk tak-
is that the social domain, as assessed here, did not show
ing was only signi?cantly predicted by severity and en-
homogenous gender effects, which is quite different from
joyment. Finally, all variables except unpleasantness sig-
the other domains of risky behavior.
ni?cantly predicted behavioral inclinations in the positive
One category of risky choice examined in the present
data set that apparently has not been previously investi-
gated is what was termed the “positive domain”: behav-
ioral choices affording a small chance of a large bene-
4 Discussion
?t for a ?xed small cost. Interestingly, women reported
greater willingness to engage in the behaviors surveyed.
4.1 Summary of Findings
These results suggest that when there is no risk of severe
negative consequences, but rather a possibility of pre-
In the health, recreational, and gambling domains,
dominantly positive consequences in exchange for some
women reported a lower likelihood of engaging in risky
small ?xed cost, women more than men will engage in
behaviors. In all three domains, there were signi?cant
such behaviors. Mediational analyses suggest that the
gender differences in perceptions of probabilities of neg-
difference arises because women judge that these conse-
ative consequences from engaging in risky behaviors,
quences are more likely to occur, and to a lesser extent,
with women reporting greater probabilities. In addition,
because they judge the consequences as more worthwhile
women expected to obtain less enjoyment from these be-
than do men. The results clearly speak against the sug-
haviors than did men in each of these three domains,
gestion that women engage in risky behaviors less often
assuming that the potential negative outcomes did not
because they are pessimistic and “feel unlucky” in some
occur. The mediational analyses revealed that percep-
global sense.
tions of negative consequences and enjoyment signi?-
One category of real-world behavior that mirrors our
cantly partially mediated gender differences in likelihood
de?nition of the positive domain quite closely is the pur-
4However, it is not unprecedented to ?nd additive models ?tting data
chasing of lottery tickets. One recent survey disclosed
of this sort reasonably well (Mellers & Chang, 1994).
that while somewhat more men (56%) than women (43%)