www.epjournal.net – 2009. 7(2): 146-159
Sex Differences in the Use of Indirect Aggression in Adult Canadians
Gail Moroschan, Psychology Department, University of Alberta, Edmonton, Canada. Email:
firstname.lastname@example.org (Corresponding author).
Peter L. Hurd, Psychology Department, University of Alberta, Edmonton, Canada.
Elena Nicoladis, Psychology Department, University of Alberta, Edmonton, Canada.
Abstract: Evolutionary psychologists have argued that the emergence of language was
associated with reducing direct physical aggression and easing social functioning in small
groups. If this is so, then males should use verbal or indirect aggression more frequently
than females since they engage in more direct aggression. A recent study found no
significant differences between men and women’s self-reports of indirect aggression in a
U.K. sample. We administered the same questionnaire to 175 male and 311 female
Canadian university students. Men in this population reported using indirect aggression
more frequently than women. The Canadian participants generally reported using indirect
aggression less frequently than the U.K. study sample did, particularly the women. These
results suggest that there are cultural differences in adults’ frequency of use of indirect
aggression. We review a number of possible reasons to account for these different results.
Keywords: Sex differences, indirect aggression, self-reports.
Evolutionary biologists view aggression as a range of strategies with which an
individual may solve the problem of competition (Archer, 1988; Riechert, 1998). Agonistic
behaviors may allow an individual to gain access to limiting resources, such as food,
territories, mates, or social dominance, and should therefore be under strong evolutionary
selection. In addition to benefits, there can be high costs associated with physical
aggression (Maynard Smith and Parker, 1976; Maynard Smith and Price, 1973) including
harm to the aggressor (Geist, 1974). Physical attacks are not the only form of aggression.
Other, more indirect or covert, acts can serve some of the same functions (Cairns, 1986;
Little, Jones, Henrich, and Hawley, 2003). A number of different terms have been applied
to these less direct forms of aggression, including relational (Crick and Grotpeter, 1995;
Little et al., 2003; Werner and Crick, 1999), covert (Björkqvist, Österman, and Lagerspetz,
1994), relational-appearing (Björkqvist et al., 1994), social aggression or social
manipulation (Björkqvist et al., 1994; Crick and Grotpeter, 1995; Galen and Underwood,
1997), and verbal, indirect, hostile, or emotional aggression (see Björkqvist, 1992, for a
review). While these different terms have different nuances (see Archer, 2001), we are
primarily concerned with distinguishing between physical, verbal, and indirect aggression,
following Björkqvist, Lagerspetz, and colleagues (1988, 1992a, 1992b, and 1994). Physical
aggression involves physical contact with objects or another person—behaviors such as
hitting, kicking, and pushing. Verbal aggression involves behaviors such as yelling,
speaking hurtful remarks, and making threats. Indirect aggression is different from physical
and verbal aggression in that it uses indirect methods to cause harm, sometimes without the
aggressor being identifiable, and includes acts such as the manipulation of social
environments to hurt the target. Behaviors of indirect aggression include damaging
another’s self-esteem or social status, using humor hurtfully, spreading rumors behind
someone’s back, damage to interpersonal relationships by excluding others from a group,
purposeful manipulation of others, or secretive acts intended to harm another person
socially or emotionally. Previous researchers have sometimes distinguished between
different kinds of indirect or covert aggression and have labeled them with different terms.
For our purposes in this paper, indirect aggression refers broadly to all acts intended to
harm through use of social and/or emotional means.
Evolutionary psychologists have argued that a strong selection pressure in the
evolution of language among humans’ ancestors was the replacement of high-cost physical
aggression with less damaging verbal and indirect means to negotiate status and power
within a social group (e.g., Cairns, 1986; Locke and Bogin, 2006). If so, then one might
predict that males would use indirect aggression more frequently than females (Locke and
Bogin, 2006; cf. Cairns, 1986). Alternatively, the use of indirect aggression may have more
to do with the vulnerability of the target. This would suggest that females would use more
indirect aggression, as they value their social relationships higher than males do.
Lagerspetz et al. (1988) found that 11- and 12-year old girls formed tighter friendships with
other girls, while boys of the same age had very loose friendships with other boys of their
cohort. If females tend to value close friendships more than males, then more indirect
aggression aimed at gaining status through social means might be more effective for
females. In fact, research on sex differences in the use of indirect aggression has yielded
The majority of research on aggression has been done on school-aged children.
Many studies show that young boys often use more physical aggression than young girls
(e.g., Björkqvist, 1992, 1994; Björkqvist et al., 1992a, 1992b; Cairns, Cairns, Neckerman,
Ferguson, and Gariépy, 1989; Crick and Grotpeter, 1995). Some studies of school-aged
children show that girls are just as aggressive as boys, but that they use more indirect
means than boys do (Björkqvist, 1992, 1994; Björkqvist et al., 1992a, 1992b; Cairns et al.,
1989; Crick and Grotpeter, 1995; Lagerspetz and Björkqvist, 1994; Lagerspetz et al.,
1988), while other studies have shown that adolescent boys use more relational aggression
than girls (Little et al., 2003).
Some studies (Björkqvist et al., 1992a, 1992b; Cairns et al., 1989) suggest a
developmental trend for aggression. Björkqvist et al. (1992a) studied Finns of four different
age groups: 8-, 11-, 15-, and 18-year olds. Boys tended to display physical aggression the
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -147-
most at ages 8 and 11. Verbal aggression increased by ages 11 and 15, and indirect
aggression was used later on, during ages 11, 15, and 18. Girls displayed substantially less
physical aggression than boys at all ages, and used slightly more verbal and indirect
aggression than boys at ages 11, 15, and 18. Indirect aggression appears to require
awareness and understanding of social relationships. As social intelligence increases with
age, so should the ability to use indirect aggression effectively. Girls typically have tighter
social relationships than boys do (Lagerspetz, et al., 1988) and therefore, are more able to
use indirect measures of aggression to their advantage. Green, Richardson, and Lago
(1996) directly tested whether the size of social networks of young adults predicted the rate
of self-reported aggression. They found that the size of social networks did not predict the
rate of direct or indirect aggression in females. However, males with high-density social
networks used more indirect than direct aggression.
The few studies to examine the use of direct and indirect aggression in adults
suggest that aggression does not necessarily decrease when compared to children’s usage,
but it is employed in subtler, and less direct forms. Björkqvist et al. (1994) showed that
adult male Swedes used more direct relational-appearing aggression, such as criticizing
work, while females used more indirect social manipulations such as gossiping. When
measures of aggression were considered in these two categories, women were equally as
aggressive as men. Sex differences were also found in bullying in an adult prison
population in the U.K. (Ireland and Archer, 1996). Women reported perceiving more
bullying overall and specifically more indirect bullying, whereas men reported perceiving
more direct bullying than indirect methods, and less bullying overall.
Many studies of indirect aggression, especially in children, use peer-report methods
rather than self-report, with the reasoning that since aggression is socially unacceptable,
tests that rely on admissions of aggression will not be reliable. With peer-reports, peers can
describe how often they have seen direct or indirect aggressive behavior in a fellow peer.
However, the very nature of indirect aggression draws on the ability to remain anonymous,
therefore, peer-reports may not capture the true amount of indirect aggressive behavior
used by specific individuals, and self-report methods might be better at measuring indirect
Forrest, Eatough, and Shevlin (2005) asked British university students to report on
the frequency with which they used, or were the targets of, indirect aggression. Indirect
aggression was classified by three categories: social exclusion, use of malicious humor, and
guilt induction. They found no sex differences on the frequency of self-reported indirect
aggression in any of these categories or the overall frequency of using indirect aggression.
Why are there such varied results with regard to sex differences in the use of
indirect aggression? One possibility is cultural differences in what qualifies as indirect
aggression. To understand how such cultural differences might be manifested, we briefly
review the literature related to sex differences on humor used aggressively. We focus on
humor because more research has been done on humor than on social exclusion or guilt
induction. For example, Baumeister, Stillwell, and Heatherton (1995) reported that guilt
induction could serve several functions in interpersonal relationships, such as inducing
action on the part of the other person or redistributing emotional distress. They did not,
however, test for the possibility of sex differences in the use of guilt induction. Coyne and
Archer (2004) reviewed the direct and indirect aggressive acts in television shows popular
among British adolescents and found that females were portrayed as committing most of
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -148-
the acts of indirect aggression, particularly in terms of social exclusion and malicious
Humor may function to allow coping with stress, anxiety, and hostile or aggressive
feelings in a socially acceptable way (Abel, 1998; McGhee and Lloyd, 1982; Woods,
1983). Using humor can be an effective way to gain status, particularly in a leadership role
(Priest and Swain, 2002). Males have been shown to use humor more than females (Hay,
2000; Honeycutt and Brown, 1998; McGhee, 1974). This sex difference can be observed as
young as the early school years (McGhee and Lloyd, 1982) and continues through
adolescence (Führ, 2002) and into adulthood (e.g., Hay, 2000; Honeycutt and Brown,
1998). Some evidence comes from studies of class clowns, who use humor to challenge
authority and reduce stress among classmates resulting from novel or strange situations
(Hobday-Kusch and McVittie, 2002; Jewell, 2005). Class clowns are overwhelmingly boys
(Damico and Pukey, 1978), probably because boys can use humor to gain prestige,
particularly with other boys (Hobday-Kusch and McVittie, 2002; Suitor, Powers and
Brown, 2004). Men are also more likely than women to use humor to negotiate status and
cope with difficult situations (Hay, 2000). It should be noted that these sex differences are
generally limited to humor production—few sex differences have been reported in humor
appreciation (Martin and Kuiper, 1999; Azim, Mobbs, Jo, Menon, and Reiss, 2005).
While sex differences in humor production have been found in some studies, others
report no sex differences. Sullivan (2002) reported no differences between mothers’ and
fathers’ use of humor to cope with having a child with Down syndrome. Forrest et al.
(2005) also reported no sex differences in using humor in an indirectly aggressive way.
Both of these studies relied on self-reports and were run in Great Britain. It is not clear if
sex differences are more likely to appear when observing actual behavior or asking for peer
reports (e.g., Hay, 2000; Little et al., 2003; but see Honeycutt and Brown, 1998, who found
sex differences with self-report measures). Alternatively, it is possible that there are
cultural differences in the use of humor. For example, Kazarian and Martin (2004) found
that Lebanese adults used less aggressive humor than Belgians.
We know of no evidence suggesting that Britons and Canadians use aggressive or
malicious humor differently. However, there is some evidence that the use of humor more
generally by Britons might be different from that by North Americans. For example, British
advertising executives approve a broader use of humor in advertising than their American
counterparts (Weinberger and Spotts, 1989). Similarly, a comparison of British and
American beer ads showed that British ads used humor as the major source of appeal, while
American ads used emotional or sexual appeal (Caillat and Mueller, 1996). These studies
suggest that there might be differences in how humor is used in the two countries (see
Coyne and Archer, 2004).
The present study
In this study, we measured the self-reported rates of indirect aggression among
Canadian adults. We replicated Forrest et al.’s (2005) British questionnaire on a Canadian
population. We chose Forrest et al.’s (2005) measure because it has strong psychometric
properties and was designed to measure indirect aggression among adults. There are a
number of cultural similarities between the United Kingdom and Canada. Canada is a
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -149-
Commonwealth country and the majority of its citizens speak English. Canadians often use
British spellings, rather than American spellings. Canadians also share many cultural
similarities with Americans— standard oral Canadian English is only subtly different from
standard oral American English (some small pronunciation differences and some different
words). Caillat and Mueller (1996) review some of the outstanding cultural differences
between the United Kingdom and the United States (such as individualism and use of direct
or indirect speech). In our experience, most of those differences would apply to a United
Kingdom-Canada comparison as well.
Materials and Methods
175 male and 311 female University of Alberta undergraduates participated in this
study and received credit toward their introductory psychology class for their participation.
We used one of the scales created by Forrest et al. (2005) to measure indirect
aggression. They created 2 versions of the Indirect Aggression Scales (IAS): the Aggressor
version (IAS-A) that measures usage of aggression towards someone else, and the Target
version (IAS-T) that measures the experience of being a victim of indirect aggression. We
used the IAS-A version only since we were interested in the use of aggressive behavior but
not the experience of being a victim. Forrest et al. (2005) constructed the IAS using
interviews and survey questions. The original IAS-A questionnaire consisted of 35 items
but was reduced to 25 items after item analyses and factor analyses were conducted. We
used these 25 items in our study. The use of this scale was granted ethical approval by the
University of Alberta, Faculty of Arts, Science and Law Research Ethics Board.
The IAS-A was administered in mass testing in introductory psychology classes.
The procedure was the same as in the Forrest et al. (2005) study. Participants read and
signed consent forms at the beginning of the mass testing session indicating that they had
been informed of their rights and could withdraw from the testing session at anytime
without penalty. Participants received a paper version of the scale and were asked to
indicate how often they had used the 25 listed behaviors in the past 12 months by using the
following scale: A = Never, B = Once or Twice, C = Sometimes, D = Often, and E =
Regularly. These answers were converted to numerical scores with A = 1, B = 2, C = 3, D
= 4, and E = 5. With this method of scoring, the lowest possible score that an individual
could receive was 25 and the highest possible score was 125. Names were not required on
the forms, providing a high level of anonymity to the participants.
Validity of the IAS-A
Forrest et al. (2005) constructed their scale with items that were based on
information gathered during qualitative interviews. Interviewees were asked about personal
experiences with indirect aggression over a variety of contexts. From these interviews
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -150-
behaviors were listed and scale items were constructed, giving the scale high face validity.
The internal construct validity of the IAS-A was assessed by Forrest et al. (2005) using
Maximum Likelihood exploratory factor analyses on the original 35 items. This resulted in
3 factors that cumulatively accounted for 35.89% of the variance: factor 1 items were
associated with malicious humor, factor 2 items were associated with social exclusion, and
factor 3 items were associated with guilt induction. Ten items failed to load consistently on
any one factor and did not make theoretical sense with the other items in each factor and
were therefore removed. The remaining 25 items were split into 3 subscales: Social
Exclusionary, 10 items; Malicious Humor, 9 items; and Guilt Induction, 6 items. See
appendix for all items.
We also did a Maximum Likelihood factor analysis on the data we collected using
the IAS-A. We cannot directly compare the eigen values or amount of variance accounted
for between our data and Forrest et al.’s (2005) because their factor analyses were
conducted on the original 35 items and our data uses the conclusive 25 items. However, we
did find three factors with eigen values above one, just like the original factor analyses, and
almost all of the items loaded on the same factors as Forrest et al. (2005) found. The few
differences we found were only marginally numerically different from the Forrest et al.
(2005) analyses. In order to compare our data with the data in the British study we grouped
the items into the same subscales used by Forrest et al. (2005).
Reliability of the IAS-A subscales
Forrest et al. (2005) measured the reliability of the subscales of the IAS-A and
found Cronbach’s alpha coefficient to be 0.82 for the Social Exclusion items, 0.84 for items
that focused on the use of Malicious Humor, and 0.81 for items regarding Guilt Induction.
Using the same items as Forrest et al. (2005) did for each subscale, we calculated
Cronbach’s alpha coefficient for the data collected from our Canadian sample and found it
to be similarly reliable: 0.86 for Social Exclusion, 0.85 for Malicious Humor, and 0.78 for
Canadians: gender differences
We were interested in seeing if there is a gender difference in how indirect
aggression is conducted in Canada. The 3 subscales of the IAS-A categorize 3 different
ways indirectly aggressive behavior can be performed and measured. Mean (standard
errors) scores for the total indirect aggression scale and each subscale, are shown in Table
1. A 3 x 2 (Subscale x Gender) ANOVA, with the Subscale as a repeated measure was
used. This revealed a main effect for gender, (F (1, 484) = 21.22, p < .01) and a main effect
for Subscale (F (2, 968) = 383.73, p < .01). There was an interaction between gender and
the subscales, (F (2, 968) = 23.91, p < .01). Planned t-tests revealed significant differences
between genders on all subscales, as summarized in Table 1.
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -151-
Table 1. Average (SE) Scores on Indirect Aggression Scale for Canadian Students
p < .0001
2.40, p < .016
6.57, p < .0001
2.30, p < .022
Note: Mean (SE) indirect aggression scores of overall test and subscales (with possible
ranges) for male and female Canadians. t-tests indicate that males scored significantly
higher than females on all measures (df = 484).
Figure 1 summarizes the scores of the males and females of the present study
compared to the results of Forrest et al. (2005). When our sample is compared to the
sample presented by Forrest et al. (2005), we find that Canadian females score lower than
the Britons on all subscales: social exclusion (t (694) = -4.33, p < .001), malicious humor (t
(694) = -8.20, p < .001) and guilt induction (t (694) = -7.09, p < .001). The males in the
present sample scored lower than those in the Forrest et al. (2005) sample on guilt
induction (t (350) = -3.42, p < .001), but did not differ in social exclusion (t (350) = -0.20, p
= .84) or malicious humor (t (350) = -0.97, p = .33).
In summary, Canadians use less indirect aggression than Britons, particularly
Canadian women, and Canadian men are more indirectly aggressive than Canadian women
on all subscales.
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -152-
Figure 1. Sex differences in the indirect aggression subscales, within and between a
Canadian sample (current study) and a British sample (Forrest et al. study).
Note: *p < 0.05, ***p < 0.001.
We found that an individual’s gender influenced the self-reported frequency of
indirectly aggressive behavior in Canadians. Males reported significantly higher rates of
indirect aggression (social exclusion, malicious humor, and guilt induction) than females.
These results differ from those reported by Forrest et al. (2005), who found no sex
differences within any indirect aggression subtype. The Canadians in this study also
generally reported significantly fewer acts of indirect aggression than did the British
participants, particularly the females. Why did we get different results from Forrest et al.
(2005)? We consider four possible sources for the differences in these two studies, in
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -153-
roughly increasing order of likelihood.
First, it is possible that there are cultural differences in willingness to report indirect
aggression. If so, the British adults might be more willing to report their own acts of
indirect aggression than Canadians. Some studies have suggested that sex differences in
self-report studies may be due to males being more comfortable with their aggression than
females (e.g., Little et al., 2003). At the moment, we know of no reason to suspect that
there should be differences between Britons and Canadians in terms of self-report biases.
Nonetheless, it is clearly important to complement self-report data with other measures.
A second possibility is that demographic differences between the samples may
explain some of the differences in the two populations. British adults might feel a higher
degree of perceived threat than Canadian adults. The population density of Nottingham
(where the Forrest et al. (2005) study was conducted) was 3579.0 people per square
kilometer, while the population density of Edmonton (where the present study was carried
out) was 974.0 people per square kilometer (2001 census data from respective countries).
Crowding and associated factors may increase the perceived risk of violence (Choldin,
1978; Kposowa, Breault, and Harrison, 1995). The responses to density may differ between
the sexes, as they do in rhesus monkeys, where males use more coping mechanisms (such
as huddling together) as population density increases, while females increase all forms of
aggression (Judge and de Waal, 1997). According to the British Home Office report on
crime (www.homeoffice.gov.uk) and Correctional Service Canada (www.csc-scc.gc.ca) the
percentage of inmates that were female was higher in Britain than in Canada in 2004 (5.7%
vs. 3.1%, respectively), and the percentage of inmates that were female and serving life
sentences, was also higher in Britain than in Canada in the same year (5.0% vs. 2.4%
respectively). To the extent that criminal activity is related to aggressive behavior, it is
possible that Britons, and particularly women, are more aggressive overall (i.e., both direct
and indirect aggression) than Canadians as a result of a higher perception of threat due to a
higher population density. To confirm this possibility, it would be minimally necessary to
compare the rates of direct aggression between the two countries.
A third possible explanation for the differences between our study and that of
Forrest et al. (2005) is that there may be cultural differences in the time spent in same-sex
groups. Men and women use humor in a teasing and aggressive way more often when they
are in same-sex groups than when they are in mixed-sex groups (Hay, 2000). If British
adults spend more time in same-sex groups than Canadians, then they might report higher
levels of indirect aggression than Canadians. Future research needs to examine the
relationship between time spent in same-sex versus mixed-sex groups, and look at the
various forms of aggression that is observed while controlling for the gender of the
Finally, it is possible that we found differences in self-report frequencies relative to
Britons because there are cultural differences in what counts as aggressive behavior, and in
what is considered appropriate behavior for men and women. We know aggressive humor
can differ by culture (Kazarian and Martin, 2004; Weller, Amitsour, and Pazzi, 1976) so it
seems likely that aspects of indirect aggression might be affected by culture. This
possibility seems the most likely and may follow some of the United Kingdom-United
States cultural differences observed by other researchers (e.g., Caillat and Mueller, 1996).
An analogous conclusion can be drawn from the research on emotional expression. There
are similarities across cultures as to how some emotions are expressed (e.g., Darwin,
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -154-
1872/1924), particularly those emotions that are expressed in infancy (e.g., Malatesta,
Culver, Tesman, and Shepard, 1989). However, different cultures use emotions differently
in order to regulate interpersonal relationships (see, for example, Lutz and White, 1986).
By analogy, we might expect aggression to be a common human trait, but how it gets
expressed could be highly culture-specific. Such subtle differences between closely related
cultures and populations likely represent phenotypic plasticity; the behavioral differences
are culturally contingent manifestations of the same underlying strategy. The expected
costs and benefits to differing degrees of aggression are not exactly the same in the two
cultural backgrounds, but the life-history strategy is the same. For example, it is possible
that there is an evolved tendency for men to use more indirect aggression that women (e.g.,
Locke and Bogin, 2006), which is why we see male adolescents use more indirect
aggression than females (Little et al., 2003). However, with the effect of socialization
(which could differ by culture), the absolute frequencies of their use change. For example,
British adolescents watch many TV shows in which female characters use a lot of indirect
aggression (Coyne and Archer, 2004). So, it is possible that sex differences in the
frequency of indirect aggression will emerge as a result of socialization, adjustment by each
individual in the perceived optimally effective level of indirect aggression. Differences
between cultures in the perception of an act may be investigated by scoring how subjects in
the two cultures perceive the same scenario depicting a potential act of indirect aggression.
Cross-cultural developmental studies of aggressive behavior would be required to assess
how differences between cultures develop with age in comparison to differences between
the sexes. We predict that sex differences would develop first, and that culture effects
would appear later, after the physiological differences were established.
Before closing, it is important to note that this study relied on self-report data of
some behaviors that are not necessarily considered socially acceptable (see also Green et
al., 1996). We do not know if we have a valid measure of the frequency of indirect
aggression. However, echoing Green et al. (1996), we are confident that the comparisons of
the relative use of indirect aggression are solid. Future research on indirect aggression
should also include behavioral measures.
Individuals ought to use whatever behavioral strategy they judge to be most
probable to achieve the desired outcome while minimizing the risk of the least desired
outcomes. This choice of behavior ought to depend upon understanding of cultural norms
and ability, including the social intelligence required for indirect aggression, as well as the
perceived costs and benefits of failure and success. Further work investigating the ability,
real or perceived, and desirability to use indirect aggression, may identify the proximate
factors that account for variation in the use of this behavior.
Acknowledgements: We thank Jaime Werne for helping run participants as well as
NSERC funding to the second and third authors.
Received 17 December 2008; Revision submitted 12 February 2009; Accepted 23
Abel, M.H. (1998). Interaction of humor and gender in moderating stress and outcomes.
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(2). 2009. -155-