Content uploaded by Bogusław Pawłowski
Author content
All content in this area was uploaded by Bogusław Pawłowski
Content may be subject to copyright.
Evolutionary Psychology
www.epjournal.net – 2008.6(1): 29-42
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Original Article
Sex Differences in Everyday Risk-Taking Behavior in Humans
B. Pawlowski, Department of Anthropology, University of Wroclaw, ul. Kużnicza 35, 50-138, Wroclaw,
Poland & Institute of Anthropology, Polish Academy of Sciences, ul. Kużnicza 35, 50-138, Wroclaw, Poland.
Rajinder Atwal, School of Biological Sciences, University of Liverpool, Liverpool L69 7ZB, United
Kingdom.
R.I.M. Dunbar, Institute of Cognitive & Evolutionary Anthropology, University of Oxford, 51 Banbury Road,
Oxford OX2 3PE. Email: robin.dunbar@anthro.ox.ac.uk (Corresponding author)
Abstract: Sexual selection theory predicts that males will tend to behave in ways that are
more risky than females. We explored this in humans by studying two everyday situations
(catching a bus and crossing a busy road). We show that humans are competent optimizers
on such tasks, adjusting their arrival times at a bus stop so as to minimize waiting time.
Nonetheless, single males pursue a more risky strategy than single females by cutting
waiting times much finer. Males are also more likely than females to cross busy roads
when it is risky to do so. More importantly, males are more likely to initiate a crossing in
high risk conditions when there are females present in the immediate vicinity, but females
do not show a comparable effect in relation to the number of males present. These results
support the suggestion that risk-taking is a form of “showing off” used as mate
advertisement.
Keywords: risk, sex differences, optimization, behavioral decisions, road-crossing
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Introduction
Evolutionary theory predicts that, in polygamously mating species, young males
will be more willing to take risks in an effort to breed successfully than young females.
This is most conspicuous in lekking species, where males may exhibit bright plumage or
conspicuous displays that make them more susceptible to predation than is the case for
females. In species like humans where risk-taking may itself become a form of display, this
sex difference may be exaggerated and risk-taking may characterize many aspects of
behavior. Many studies have noted that young human males are more prone than females to
take risks in relation to conflict (Campbell, 1999; Daly and Wilson, 1988; Wilson and
Daly, 1993,) and sexual behavior (Clift, Wilkins, and Davidson, 1993; Poppen 1995), as
Sex differences in risk-taking
well as in such situations as car driving (Chen, Baker, Braver, and Li, 2000; Flisher,
Ziervogel, Charlton, Leger, and Roberston, 1993; Harre, Field, and Kirkwood, 1996),
accident risk (Fetchenhauer and Rohde, 2002), drug-taking (Tyler and Lichtenstein, 1997),
gambling and financial decisions (Bruce and Johnson, 1994, Powell and Ansic, 1997) and
outdoor activities (Howland, Hingson, Mangione, and Bell, 1996, Wilson, Daly, Gordon,
and Pratt, 1996). Indeed, psychological studies have found that females find risky
situations more stressful than males do (Kerr and Vlaminkx, 1997). In this context, risk-
taking by males may be a form of mating display (Hawkes, 1990, 1991).
Most of these studies of humans focus on situations that, in one degree or another,
are life-threatening (or which may have serious consequences for subsequent health and
wealth). In this paper, we report results showing that sex differences in risk-taking occur
even in simple everyday situations that do not necessarily incur significant risks. We
consider two examples: catching a bus and crossing a road.
We use these data to explore two separate issues. The first offers us the opportunity
to examine the trade-offs between the costs and benefits of alternative courses of action:
individuals can either arrive early (but thereby incur a cost because they have to wait in the
cold: playing safe) or they can arrive closer to departure time (but at the risk that the bus
may already have left by the time they arrive at the bus stop: risky strategy). The riskiness
of the latter option is created by the fact that buses often leave before their official
departure times (especially if the bus has already filled to capacity). The second study
allows us to explore more directly the possibility that risk-taking may be a form of male
mating display (sensu Hawkes, 1991).
Methods
The observations for the first study were carried out at a single bus stop which
students habitually use to get to the University of Liverpool campus (about 2.5km distant).
Most of the housing in the immediate area of the bus stop is student accommodation. A bus
to the university campus specifically provided for students starts its journey at this bus stop.
The bus usually arrives up to 12 minutes before its official departure time and waits at the
stop. All the observations were carried out on the bus that was officially timed to depart at
0940 a.m.: departure times were, however, randomly distributed around the official
departure time. Figure 1 gives the cumulative frequency distribution for all bus departure
times, relative to the official departure time of 0940 hrs. For convenience, all times have
been converted to the number of minutes before the official bus departure time. The data
are given as the proportion of all buses (n = 32) that departed up to and including the
minute shown.
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -30-
Sex differences in risk-taking
Figure 1. Cumulative probability distribution of actual bus departures times (n = 32 days).
The vertical line marks the official bus departure time. The X-axis is scaled as the number
of minutes prior to the official departure time of the bus.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
-14-12-10-8-6-4-20
Minutes Prior to Official Departure Time
Cumulative Probability
2
The arrival times of individual males and females and the subsequent bus departure
times were noted on 32 mornings over a four-month period during the winter months.
However, only those days on which the bus left at the appointed time (or later) were
included in the analyses of risk-taking behavior in order to be sure that all arrivals up to and
including the minute at which the bus should have gone could be counted. This yielded a
sample of 20 mornings. On two of these, the bus left later than the appointed time, but only
arrivals up to and including the official departure time were counted for analysis. The
arrival times of 475 females and 524 males were recorded in the samples for the 20 days
used in the analysis.
The second study was carried out at a busy road crossing in the middle of the
University campus over the midday period. The crossing point was provided with a light-
controlled pedestrian crossing. Individual subjects were selected as they approached the
crossing and the following variables recorded: sex and approximate age (by decade) of
subject, risk state of road on approach, whether the subject crossed or waited, risk state of
the road when the subject crossed, whether the subject was a leader or a follower when
he/she crossed, number and sexes of all individuals on the subject’s side of the crossing
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -31-
Sex differences in risk-taking
point at the moment he/she crossed. Over a three month period, a total of 500 males and
500 females were sampled in this way. The risk state of the road was defined in terms of
the risk of being hit by a vehicle when crossing at that moment (see Table 1).
Table 1. Risk state of the road crossing.
Risk State Definition
______________________________________________________________
No risk Traffic stationary at traffic signal; safe to cross
Low risk Traffic can approach, but no vehicles within 50m
Risky Vehicle approaching (within 50m) but not at crossing
High risk Vehicles passing through the crossing
______________________________________________________________
Results
Optimizing bus waiting time
The bus departure times given in Figure 1 can be used to calculate the mean delay
to the next bus at each minute, noting that individuals who miss the bus at time 0 minutes
have to wait on average a further 14.12 minutes for the next student bus. (This is calculated
as the weighted delay to the departure of the next bus, given that the official departure time
is 15 minutes after the first, but with a probability density function for departure times
similar to that shown in Figure 1.) The results are shown in Figure 2. The minimum delay
occurs at minute 5, and this represents the optimal arrival time that trades off time spent
waiting in the cold by arriving early against the risk of missing the bus.
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -32-
Sex differences in risk-taking
Figure 2. Predicted waiting time for individuals arriving at indicated times prior to the
official departure time of the bus, estimated from the actual departure times given in Figure
1.
Females were more likely to arrive in groups than were males (42.0% vs. 28.6%,
respectively:
χ
2 = 19.71, df = 1, p < 0.001). Therefore, to avoid confounds due to social
factors and pseudoreplication when individuals arrive in groups, we separate out
individuals who arrive alone from those who arrive in groups. Figure 3 gives the mean
minutes prior to departure at which males and females in different group types arrived.
Males arriving alone did so significantly later than females arriving alone (mean arrival
times: 4.22 ± 0.133SE vs. 4.98 ± 0.176SE minutes prior to departure, respectively;
Kolmogorov-Smirnov test: Z = 1.609, n = 375,275, p = 0.011). Note that the mean arrival
time for females (4.98 minutes prior to departure) is almost exactly the optimal arrival time
identified in Figure 2 (five minutes). Since there is considerable variance in the data, we
divided arrival times into three blocks differing in riskiness: minutes 12-6 (cautious
period), minutes 5-4 (optimal period) and minutes 3-0 (risky period). Females arriving
alone (i.e. those whose arrival times are not influenced by others) were significantly more
likely to arrive during the cautious period, whereas males arriving alone were more likely
to arrive during the risky period (
χ
2 = 10.75, df = 2, p = 0.001). These results stand even
when days are considered as the unit of analysis (for subjects arriving alone, medians of
43.7% of males vs. 28.2% of females arrived during the risky period: Wilcoxon matched
pairs test, n = 20 days, p = 0.033 2-tailed).
When individuals arrived on their own, males did so significantly later than females
(i.e. they cut the waiting time to a minimum). However, when they arrived in single sex
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -33-
Sex differences in risk-taking
groups, males were more likely to arrive earlier than females (though not significantly so:
means of 5.55±0.303SE vs. 4.58±0.230SE; Kolmogorov-Smirnov test, n = 86 and 138, p
= 0.157). A likely explanation for this is that these males were actually late for the
previous bus: this suggestion is reinforced by the fact that female arrival times do not vary
significantly across group type (Kruskal Wallis χ2 = 1.92, n = 424, df = 3, p = 0.589),
whereas those for males do (Kruskal Wallis χ2 = 17.38, n = 446, df = 3, p = 0.001).
Moreover, when males accompanied females in couples, their arrival times were the same
as those for single females (Kolmogorov-Smirnov Z = 0.614, n = 43,275, p = 0.846), but
significantly different from (and later than) those for single males (Z = 2.382, n = 43,375, p
= 0.044), suggesting that male behaviour was being driven by that of the females.
Figure 3. Mean ± SE of number of minutes prior to the departure of the next bus that males
(solid symbols) and females (open symbols) arrived at the bus stop in groups of different
composition.
Alone Single-sex
group Couples Mixed-sex
group
3
4
5
6
7
Mean +- 1 SE minutes
Road-crossing
Figure 4 shows the frequencies with which males and females crossed the road at
different risk states. Males are significantly more likely to cross the road at higher risk
states than females (χ2 = 32.56, df = 3, p < 0.001). This is reinforced by a comparison of
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -34-
Sex differences in risk-taking
sex differences in the riskiness at road crossing for individuals who approached the road
when it was on the highest risk state (vehicles on the crossing): males arriving at this risk
state crossed at significantly more risky states than females did (χ2 = 52.27, df = 3, p < <
0.001).
Figure 4. Frequencies with which males and females crossed the road at different risk
states.
0
0.2
0.4
0.6
0.8
1
1.2
No risk Low risk Risky High risk
Probability of Crossing
Males
Females
In a significant number of instances, the subject arrived at the crossing when a
vehicle was approaching or on the crossing (high risk states). In these cases, subjects either
crossed or waited until the risk state of the crossing was reduced. Figure 5 plots the
probability of crossing for the two sexes in relation to the risk state of the road at the
moment of arrival. Overall, males are less likely to wait than females (χ2 = 19.02, df = 1, p
< 0.001), and this difference was true at each risk state except the lowest (no risk: χ2 = 2.49,
df = 1, p > 0.05).
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -35-
Sex differences in risk-taking
Figure 5. Relative frequency with which males and females would wait rather than cross in
relation to the risk state of the road on their arrival at the crossing.
0
0.2
0.4
0.6
0.8
1
1.2
No risk Low risk Risky High Risk
Probability of Crossing
Males
Females
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -36-
Sex differences in risk-taking
Overall, males were more likely to act as leaders (initiators of a crossing) than were
females (Figure 6: χ2 = 12.15, df = 1, p < 0.001). This was true at all risk levels except very
risky, where, paradoxically, females were more likely to lead than follow compared to
males. However, males were three times more likely to cross in the highest risk category
than females were (90 vs. 33 occasions) and, in absolute terms, were twice as likely to be
leaders (50 vs. 24 occasions).
Figure 6. Probability that males and females act as leaders (initiate a crossing) in relation
to the risk state of the road on crossing.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
No risk Low risk Risky High risk
Probability of Leading
Males
Females
To examine spectator effects on risk-taking, we analyzed the risk at crossing in
relation to the number of males and females on the subject’s side of the road at the moment
of crossing for those occasions when the subject was a leader (i.e. initiated a crossing).
Table 2 gives the resulting analyses of variance, with risk state at crossing as the dependent
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -37-
Sex differences in risk-taking
variable. The number of females present (but not the number of males) has a significant
effect on the male’s crossing pattern, but females show only a weak (non-significant) effect
due to the number of males present. Figure 7 plots these data in the form of the probability
of initiating a crossing (i.e., acting as “leader”) as a function of the number of members of
the opposite sex present for high and low risk states at crossing. While the probability of
crossing typically declines as audience size increases for both sexes, it remains high for
males in the high risk condition, suggesting a greater willingness to take risks when there
are females present to display to.
Table 2. ANOVA for risk state at crossing for each sex in relation to numbers of males
and females present at the time.
Variable F df p
________________________________________________________
(a) Males
Number of males 1.24 13,406 0.251
Number of females 1.78 14,406 0.039
Males x females 1.11 66,406 0.271
________________________________________________________
(b) Females
Number of males 1.69 13,413 0.060
Number of females 0.54 12,413 0.891
Males x females 1.32 60,413 0.065
________________________________________________________
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -38-
Sex differences in risk-taking
Figure 7. Probability with which males and females initiated a road crossing (i.e. were
leaders) in relation to risk state and the number of members of the opposite sex who were
waiting on their side of the road. Low risk states are “no risk” and “slight risk;” high risk
states are “risky” and “very risky”.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Low risk High risk Low risk High risk
Males Males Females Females
Probability of Crossing
0
1 or 2
3+
Spectators
Discussion
We have shown that males are more likely to take risks than females, even in
everyday situations that are relatively unlikely to incur life-threatening costs. This suggests
that risk-taking is a pervasive feature of human male psychology. In addition, we have
shown that males’ risk-proneness even at this level is related to the presence of females in
the immediate vicinity. We infer from this that male risk-taking is a form of mating display
(sensu Hawkes, 1990), comparable to other forms of mate advertising peculiar to modern
humans (e.g. display of mobile phones: Lycett and Dunbar, 2000). In this case, risk-taking
is assumed to reflect something about gene quality rather than the resources a male has to
offer.
Even though both sexes seemed to be quite efficient at optimizing their arrival times
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -39-
Sex differences in risk-taking
at a bus stop so as to trade off the risk of missing the bus against waiting time, males and
females appear to differ in their willingness to pursue risky strategies in this context. Males
are more likely to cut their arrival times fine (and sometimes miss the bus as a result,
thereby incurring longer waiting times than they need have done in cold winter
temperatures, as well as being late for a class). An alternative explanation for these results
might be that human females are more resistant to cold and are therefore more willing to
spend longer in cool temperatures by arriving early. However, sex differences in
subcutaneous fat volume might be expected to have exactly the opposite effect: the issue is
not how long you spend waiting for a particular bus but the risk of having to wait for the
next bus because you arrived after your target bus had left. With their lower subcutaneous
fat volumes, males ought to be more risk-averse than females because they would bear
more significant costs from heat-loss if they missed the bus. In contrast, heat-loss cannot
possibly be an explanation for the sex differences in road-crossing behavior, yet similar
sex-biases in risk-taking emerge from this study. This suggests that the sex differences
observed in the bus study are genuine differences in risk-sensitivity.
It is worth pointing out here that, in evaluating the optimal arrival time in the bus
study, we considered only the expected waiting time. Part of the costs of waiting lie in the
rate of heat loss when exposed to low ambient temperatures. While we treated this as more
or less constant (and modest), a change in local climatic conditions can be expected to have
a significant impact on the optimal arrival time. As ambient temperatures fall, so the cost
of waiting increases and the optimal arrival time can be expected to drift towards the
official bus departure time (minute 0) and the payoff function (as shown in Figure 2) will
become more strongly peaked. Conversely, as temperatures rise, so the constraints on
waiting time will weaken and the payoff function should become flatter. We should thus
expect to see differences both between locations and, within locations, between seasons and
days in both the shape of the payoff function and the consequent behavior of individuals.
The difference in risk-taking behavior might be the result of either of two rather
different effects. One is that males are more prone to risk-taking as a form of display and
merely implement it under all conditions; the other is that males are more reluctant to waste
time on inessential activities (such as waiting at a bus stop). For females, such an activity
may have additional functional benefits in terms of social interaction (e.g. servicing social
relationships). Because males may be less social than females, waiting around at bus stops
may be less advantageous for them. Some support for this second interpretation is given by
the suggestion that males were less social than females in our sample: they were more
likely than females to arrive at the bus stop alone rather than in groups. In either case, we
interpret these proximate functional considerations as being derivative of differences
between the sexes in long term reproductive strategies.
However, perhaps the strongest evidence in support of a generic advertising
hypothesis is the fact that risk-taking by males during road-crossings is directly affected by
the presence of female (but not male) spectators, whereas the female’s behavior is only
marginally (and non-significantly) influenced by the presence of males. Indeed, females are
more likely to act as followers than leaders in these situations. These observations make it
much less likely that it is social factors that are driving the sex differences in risk-taking
behavior.
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -40-
Sex differences in risk-taking
Groups clearly impose constraints on the behavior of individuals, since the stability
of a group through space and time depends on its members compromising on their
preferred options. Males in mixed sex groups seem to give way to females’ demands as far
as arrival time is concerned, and are thus more likely to optimize arrival time than lone
males (itself a suggestion that mate-searching tactics may be more important than purely
social considerations). In contrast, the apparently optimal behavior of males in single-sex
groups (at least compared to single males) is probably an artifact: the arrival times for
males in single-sex groups are clearly bimodal, with peaks at three minutes and eight
minutes before departure. No less than a third of males in groups arrived nearer to the
departure time of the previous bus (i.e. more than seven minutes before the bus’s official
departure time). One explanation might be that males in single sex groups in fact behave
like single males, except that grouping has the effect of slowing them down in order to
accommodate the slowest member, such that they are more likely to miss the bus. These
individuals thus appear in our data as early arrivals when they are in fact late-comers. This
is reflected in the fact that the minute eight peak is much higher in single sex group males
than it is for single males (where the minute three and minute one peaks are higher). In
contrast, males in mixed-sex groups behave very differently: reproductive and mate choice
considerations seem to dictate that males fall into line with female behavioral patterns in
these groups.
Whatever the proximate reasons for the behavior patterns shown by males might be
(e.g. the desire to spend longer in bed, a dislike of waiting around), the fact is that their
behavior is shifted into a more risky part of the state space relative to that for females.
Although they did not consider sex differences as such, Hoffrage, Weber, Hertwig, and
Chase, (2003) have shown, at least for children, that risk-taking does seem to generalize
across contexts: children who were more prone to taking risks when crossing roads were
also more prone to taking risks in gambling games. Sexual selection theory would suggest
that this is a consequence of males’ behavior being driven more strongly by the demands of
mating tactics than that of females.
Acknowledgements: BP was supported by a Royal Society NATO Postdoctoral
Fellowship for Partner Countries.
Received 15 August 2007; Revision submitted 4 January 2008; Accepted 4 January
2008
References
Bruce, A.C., and Johnson, J.E.V. (1994). Male and female betting behaviour: New
perspectives. Journal of Gambling Studies, 10, 183-198.
Campbell, A. (1999). Staying alive: Evolution, culture and women’s intrasexual
aggression. Behavioral and Brain Sciences, 22, 203-267.
Chen, L-H., Baker, S.P., Braver, E.R. and Li, G. (2000). Carrying passengers as a risk
factor for crashes fatal to 16- and 17-year-old drivers. Journal of the American Medical
Association, 283, 1578-1582.
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -41-
Sex differences in risk-taking
Clift, S.M., Wilkins, J.C., and Davidson, E.A. (1993). Impulsiveness, venturesomeness and
sexual risk-taking among heterosexual GUM clinic attenders. Personality and
Individual Differences, 15, 403-410.
Daly, M., and Wilson, M. (1988). Homicide. Hawthorne, NY: Aldine de Gruyter.
Fetchenhauser, D. and Rohde, P.A. (2002). Evolutionary personality psychology and
victimology – Sex differences in risk attitudes and short-term orientation and their
relation to sex differences in victimizations. Evolution and Human Behavior, 23, 233-
244.
Flisher, A.J., Ziervogel, C.F., Charlton, D.O., Leger, P.H., and Roberston, B.A. (1993).
Risk-taking behaviour of Cape Penninsula high-school students. 6. Road related
behaviour. South African Medical Journal, 83, 486-490.
Harre, N., Field, J., and Kirkwood, B. (1996). Gender differences and areas of common
concern in the driving behaviors and attitudes of adolescents. Journal of Safety
Research, 27, 163-173.
Hawkes, K. (1990). Why do men hunt? Benefits for risky choices. In E. Cashdan (Ed.)
Risk and uncertainty in tribal and peasant economies (pp. 145-166). Boulder, CO):
Westview Press.
Hawkes, K. (1991). Showing off: Tests of an hypothesis about men’s foraging goals.
Ethology and Sociobiology, 12, 29-54.
Hoofrage, U., Weber, A., Hertwig, R. and Chase, V.M. (2003). How to keep children safe:
Find the daredevils early. Journal of Experimental Psychology – Applied, 9, 249-260.
Howland, J., Hingson, R., Mangione, T.W., and Bell, N. (1996). Why are most drowning
victims men? Sex differences in aquatic skills behaviors. American Journal of Public
Health, 86, 93-96.
Kerr, J.H. and Vlaminkx, J. (1997). Gender differences in the experience of risk.
Personality and Individual Differences, 22, 293-295.
Lycett, J.E. and Dunbar, R.I.M. (2000). Mobile phones as lekking devices among human
males. Human Nature, 11, 93-104.
Poppen, P.J. (1995). Gender and patterns of sexual risk taking in college students. Sex
Roles, 32, 545-55.
Powell, M., and Ansic, D. (1997). Gender differences in risk behaviour in financial
decision-making: An experimental analysis. Journal of Economic Psychology, 18, 605-
628.
Tyler, J., and Lichtenstein, C. (1997). Risk, protective, AOD knowledge, attitude, and AOD
behavior: Factors associated with characteristics of high-risk youth. Evaluation and
Program Planning, 20, 27-45.
Wilson, M. and Daly, M. (1993). Lethal confrontational violence among young men. In
N.J. Bell and R.W. Bell (Eds.) Adolescent Risk-Taking (pp. 84-106). Newbury Park
(CA), Sage.
Wilson, M., Daly, M., Gordon, S., and Pratt, A. (1996). Sex differences in valuations of the
environment. Population and Environment, 18, 143-159.
Evolutionary Psychology – ISSN 1474-7049 – Volume 6(1).2008. -42-
