Guns, Testosterone, and
An Experimental Test of a Mediational Hypothesis
Jennifer Klinesmith, Tim Kasser, and Francis T. McAndrew
ABSTRACT—We tested whether interacting with a gun in-
creased testosterone levels and later aggressive behavior.
Thirty male college students provided a saliva sample (for
testosterone assay), interacted with either a gun or a
children’s toy for 15 min, and then provided another saliva
sample. Next, subjects added as much hot sauce as they
wanted to a cup of water they believed another subject
would have to drink. Males who interacted with the gun
showed significantly greater increases in testosterone and
added more hot sauce to the water than did those who
interacted with the children’s toy. Moreover, increases in
testosterone partially mediated the effects of interacting
with the gun on this aggressive behavior.
Substantial evidence suggests that aggression can be increased
by the presence of weapons in the environment and by the
hormone testosterone. Several studies show that the presence of
aggressive environmental cues such as weapons can increase
the accessibility of hostile, aggressive thoughts and lead to more
aggressive behavior (Anderson, Benjamin, & Bartholow, 1998;
Bartholow, Anderson, Carnagey, & Benjamin, 2005; Berkowitz
& LePage, 1967; Bettencourt & Kernahan, 1997; Killias &
Haas, 2002). Regarding testosterone, in animal species ranging
from chickens to monkeys, the injection of this hormone in-
creases aggressiveness and social dominance behavior, re-
gardless of whether the animals are males or females (Ellis,
1986); in humans, however, the results are more mixed, with
many laboratory and ﬁeld studies revealing strong positive re-
lations between testosterone and levels of restlessness, tense-
ness, and tendency toward violence (Archer, 1994; Campbell,
Muncer, & Odber, 1997; Dabbs, Carr, Frady, & Riad, 1995;
Dabbs, Jurkovic, & Frady, 1991) and other studies failing to
replicate such effects (Archer, 1991; Archer, Birring, & Wu,
1998; O’Connor, Archer, Hair, & Wu, 2001; Rowe, Maughan,
Worthman, Costello, & Angold, 2004).
Surprisingly, we were unable to ﬁnd any studies that examined
whether testosterone and the presence of a weapon might work
together to increase aggressive behavior. Perhaps the presence
of a stimulus such as a gun triggers increases in testosterone
levels, which in turn increase aggressive behavior. Such a chain
of events would be predicted by the challenge hypothesis de-
veloped by Wingﬁeld, Hegner, Dufty, and Ball (1990) to explain
aggressive behavior in male pair-bonded birds. According to
this hypothesis, testosterone rises in response to situational cues
that represent either a threat to a male’s status or a signal that
competition with other males is imminent; such increases in
testosterone then facilitate whatever competitive behaviors
(including potentially aggressive responses) are necessary for
meeting the challenge. The challenge hypothesis has been
supported by studies across a wide range of vertebrate species
(Cavigelli & Pereira, 2000; Ferree, Wikelski, & Anderson,
2004; Hirschenhauser, Taborsky, Oliveira, Canario, & Oliveira,
2004; Muller & Wrangham, 2004); most studies in humans have
focused on how males’ testosterone levels rise and fall de-
pending on success or failure in competitions (Archer, 1991;
Booth, Shelley, Mazur, Tharp, & Kittok, 1989; Gladue, Boechler,
& McCaul, 1989; Mazur, Booth, & Dabbs, 1992; Mazur & Lamb,
1980) or in response to insults (Cohen, Nisbett, Bowdle, &
Schwarz, 1996; see Archer, 2006, for a review of the applica-
bility of the challenge hypothesis to humans).
In this study, we examined whether the presence of a gun (vs. a
control object) might act as a stimulus signaling competition and
a threat to status; if so, according to the challenge hypothesis,
it should cause increases in males’ testosterone levels, which in
turn should increase their aggressive behavior. We assessed
males’ testosterone levels both before and after interacting with
a gun or a children’s toy; to measure aggression, we adapted a
method developed by Lieberman, Solomon, Greenberg, and
Address correspondence to Tim Kasser, Box K-83, Knox College,
Galesburg, IL 61401, e-mail: email@example.com.
568 Volume 17—Number 7Copyright r2006 Association for Psychological Science
McGregor (1999) that gives subjects the opportunity to anony-
mously put hot sauce in a cup of water that they believe another
person will have to drink. We hypothesized that males who in-
teracted with the gun would show both a greater increase in
testosterone levels and more aggression than would males who
interacted with the children’s toy. We also hypothesized that
changes in testosterone levels would be correlated with ag-
gression levels and would indeed mediate the effects of inter-
acting with a gun on later aggressive behavior.
Subjects were 30 male college students (age range: 18–22) who
received extra course credit or a small monetary reward for their
participation. All subjects were run during the afternoon or early
Procedure and Materials
When recruited, subjects were informed that the study would
examine taste sensitivity in males and that they would therefore
need to provide saliva for hormone analysis; subjects were asked
not to eat, drink, smoke, or brush their teeth for 1 hr prior to
testing in order to minimize impurities in the saliva samples.
When subjects arrived at the lab, a female experimenter con-
ﬁrmed that the subjects had indeed followed these instructions
before she administered consent procedures. Next, participants
provided an approximately 6-ml sample of saliva by spitting into
a cup; this saliva was used to assess baseline, or Time 1,
All subjects were then led into a room containing a television,
a chair, and a table with an object and some paper on it. For
experimental subjects, the object was a pellet gun identical in
size, shape, and feel to a Desert Eagle automatic handgun; for
control subjects, the object was the children’s game Mouse
Trapt. Subjects were told that the study was investigating
whether taste sensitivity was associated with the attention to
detail required for creating instructions concerning the object.
Subjects were therefore asked to spend 15 min handling the
object and writing a set of instructions about how to assemble
and disassemble it; a drawing of the object was also provided for
subjects to label the object’s parts. The handgun and children’s
game were similar in number and complexity of parts.
After 15 min, the experimenter reentered the room, asked the
subject to stop working on the instructions, and obtained a Time
2 saliva sample from the subject. The subject was told he would
next perform the taste-sensitivity portion of the study. He was
given a cup ﬁlled with 85 g of water and a single drop of Frank’s
Red Hot Sauce. The subject was told that the sample had been
prepared by a previous subject, was instructed to take a sip of the
sample, and was then asked to rate the taste of the sample on a
The experimenter left and then returned with a tray containing
a cup of 85 g of water, a nearly full bottle of Frank’s Red Hot
Sauce, and a lid. The subject was asked to prepare a sample for
the next subject by placing as much hot sauce in the water as he
wanted. He was assured that neither the person who drank it nor
the experimenter would know how much hot sauce he had put in
the water, as the lid was to be put on the cup after the hot sauce
was added. The experimenter then left the room, and the sub-
ject signaled when he was ﬁnished adding the hot sauce.
(Throughout this process, the gun or the game remained in the
room.) The cup was then removed from the room, and the ex-
perimenter weighed it again to obtain a measure of the amount of
hot sauce, in grams, the subject had added to the water. This
served as our primary measure of aggression (see Lieberman et
Because of the potentially arousing nature of the experiment,
we wanted to ensure that all subjects were reasonably calm when
they left the lab. Therefore, all subjects next watched a relaxing
video of nature scenes and classical music. Given that subjects
had been deceived, we next debriefed them, emphasizing that
they should not feel badly about any aggressive behavior they
exhibited. Interestingly, several subjects were disappointed
when told that the sample of hot sauce and water they had
prepared would not actually be given to the next subject. No
subjects expressed suspicion as to the true nature of the study.
Time 1 and Time 2 saliva samples were stored for 24 hr at room
temperature, centrifuged, and then frozen at 20 1C until the
time of the assay (Erikkson & Von Der Pahlen, 2002). The
samples were then brought to room temperature, transferred to
Eppendorf tubes, centrifuged for 15 min at 3,000 rpm to remove
debris, and then assayed in duplicate using a commercially
available microwell kit for testosterone level (Salimetrics, LLC,
State College, PA). All samples were assayed in house in a single
batch using a standard radioimmunoassay (RIA) procedure
under the supervision of an experienced RIA technician; at both
Time 1 and Time 2, the duplicates were averaged to yield our
measures of testosterone level. The intra-assay coefﬁcient of
variation for subjects was 5.3%, and the sensitivity of the assay
was less than 1.5 pg/ml from zero for men. Mean Time 1 and
Time 2 testosterone levels were 222.59 pg/ml (SD 597.17) and
253.92 pg/ml (SD 598.32), respectively. We subtracted each
subject’s Time 1 level from his Time 2 level to obtain a measure
of change in testosterone.
Our ﬁrst hypothesis was conﬁrmed: Subjects who interacted with
the handgun showed a greater increase in testosterone from
Time 1 to Time 2 (mean change 562.05 pg/ml, SD 548.86)
than did those who interacted with the children’s game (mean
Volume 17—Number 7 569
Jennifer Klinesmith, Tim Kasser, and Francis T. McAndrew
change 50.68 pg/ml, SD 528.57), t(28) 54.20, p
d51.53. Thus, interacting with the gun increased testosterone
Our second hypothesis was also conﬁrmed: Subjects who in-
teracted with the gun added more hot sauce to the water (M5
13.61 g, SD 58.35) than did those who interacted with the
children’s toy (M54.23 g, SD 52.62), t(28) 54.16, p
.99, d51.52. Thus, interacting with the gun increased ag-
Our third hypothesis was also conﬁrmed: The amount of hot
sauce placed in the cup was positively correlated with changes
in testosterone level (r5.64, p
5.41). Given that
all three of Baron and Kenny’s (1986) prerequisites for a me-
diational model were met, we next examined whether the size of
the association between the predictor variable (i.e., gun vs.
game) and the outcome variable (i.e., grams of hot sauce added)
diminished once the effects of the purported mediating variable
(i.e., changes in testosterone) were controlled. Indeed, the size of
the correlation between group membership (experimental vs.
control) and aggression dropped from r5.62 (p
pr 5.36 (p
5.91) after controlling for changes in testos-
terone. Finally, to more stringently test our mediational hy-
pothesis, we computed Sobel’s (1982) test for mediation, using
the Web site developed by Preacher and Leonardelli (2001). As
before, the evidence suggested that the effect of guns on ag-
gression was significantly mediated by changes in testosterone
levels, Sobel’s test 52.09, p
Past research shows that both testosterone and exposure to guns
are associated with aggressive behavior, but no studies, to our
knowledge, have examined how the two factors might work to-
gether. The present results demonstrated that males who inter-
acted with a gun showed a greater increase in testosterone levels
and more aggressive behavior than did males who interacted
with a children’s toy. Mediational analyses suggested that part of
the reason that guns increase aggression is that they cause in-
creases in testosterone levels.
Such ﬁndings not only are consistent with the challenge hy-
pothesis (Wingﬁeld et al., 1990), but also provide new support
for it both by examining a new type of ‘‘challenging’’ stimulus
and by assessing later aggressive behavior. Additionally, the
results provide evidence against some interpretations suggest-
ing that experimental effects of guns on aggression are due to
subtle demand characteristics. That is, it seems unlikely that
subtle experimenter pressures could significantly increase
subjects’ testosterone levels; instead, it seems more reasonable
to believe that the presence of the gun had this effect.
Future research could explore a variety of avenues for cor-
recting some of the limitations of the present study. For example,
would females’ biological and behavioral responses to guns be
similar to males’? Perhaps they would, as other female animals
act more aggressively when injected with testosterone (Ellis,
1986). But perhaps they would not, as the types of evolutionary
challenges faced by ancestral females were different from those
faced by males, and thus females may react to guns differently
than males. Another topic worthy of further study concerns the
fact that subjects in this study only had the opportunity to ag-
gress in an anonymous, rather indirect fashion against an un-
known individual. Would the same pattern of results hold if the
aggression was directed against a particular individual, or if
there was a possibility of retribution from the victim? Finally,
would past experience with guns moderate these effects? That is,
would individuals who frequently handle guns (such as hunters
or soldiers) respond similarly to those with little or no experience
with weapons? Previous research (Bartholow et al., 2005) sug-
gests that this may be the case, but a link with testosterone has
yet to be established.
In sum, the present study replicates past research showing
that exposure to guns may increase later interpersonal aggres-
sion, but further demonstrates that, at least for males, it does so
in part by increasing testosterone levels. Such ﬁndings raise
many of the usual questions about whether the presence of guns
in modern society contributes to violent behavior. Although our
study is clearly far from deﬁnitive, its results suggest that guns
may indeed increase aggressiveness partially via changes in the
Acknowledgments—We thank Heather Hoffmann, Janet Kirk-
ley, Glen Normile, and Neil Schmitzer-Torbert for their assistance.
This research was funded by a grant from the Richter Memorial
Anderson, C.A., Benjamin, A.J., Jr., & Bartholow, B.D. (1998). Does the
gun pull the trigger? Automatic priming effects of weapon pictures
and weapon names. Psychological Science,9, 308–314.
Archer, J. (1991). The inﬂuence of testosterone on human aggression.
British Journal of Psychology,82, 1–28.
Archer, J. (1994). Violence between men. In J. Archer (Ed.), Male vi-
olence (pp. 121–140). London: Routledge.
Archer, J. (2006). Testosterone and human aggression: An evaluation of
the challenge hypothesis. Neuroscience and Biobehavioral Re-
Archer, J., Birring, S.S., & Wu, F.C.W. (1998). The association between
testosterone and aggression among young men: Empirical ﬁndings
and a meta-analysis. Aggressive Behavior,24, 411–420.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strate-
gic, and statistical considerations. Journal of Personality and
Social Psychology,51, 1173–1182.
Bartholow, B.D., Anderson, C.A., Carnagey, N.L., & Benjamin, A.J., Jr.
(2005). Interactive effects of life experience and situational cues
on aggression: The weapons priming effect in hunters and non-
hunters. Journal of Experimental Social Psychology,41, 48–60.
Berkowitz, L., & LePage, A. (1967). Weapons as aggression-eliciting
stimuli. Journal of Personality and Social Psychology,7, 202–207.
570 Volume 17—Number 7
Guns, Testosterone, and Aggression
Bettencourt, B.A., & Kernahan, C. (1997). A meta-analysis of aggres-
sion in the presence of violent cues: Effects of gender differences
and aversive provocation. Aggressive Behavior,23, 447–456.
Booth, A., Shelley, G., Mazur, A., Tharp, G., & Kittok, R. (1989).
Testosterone and winning and losing in human competition. Hor-
mones and Behavior,23, 556–571.
Campbell, A., Muncer, S., & Odber, J. (1997). Aggression and testos-
terone: Testing a bio-social model. Aggressive Behavior,23, 229–
Cavigelli, S.A., & Pereira, M.E. (2000). Mating season aggression and
fecal testosterone levels in male ring-tailed lemurs, Lemur catta.
Hormones and Behavior,37, 246–255.
Cohen, D., Nisbett, R.E., Bowdle, B.F., & Schwarz, N. (1996). Insult,
aggression, and the Southern culture of honor: An ‘‘experimental
ethnography.’’ Journal of Personality and Social Psychology,70,
Dabbs, J.M., Jr., Carr, T.S., Frady, R.L., & Riad, J.K. (1995). Testos-
terone, crime, and misbehavior among 692 male prison inmates.
Personality and Individual Differences,18, 627–633.
Dabbs, J.M., Jr., Jurkovic, G., & Frady, R.L. (1991). Salivary testos-
terone and cortisol among late adolescent male offenders. Journal
of Abnormal Child Psychology,19, 469–478.
Ellis, L. (1986). Evidence of neuroandrogenic etiology of sex roles from
a combined analysis of human, nonhuman primate, and nonpri-
mate mammalian studies. Personality and Individual Differences,
Erikkson, C.J., & Von Der Pahlen, B. (2002). Testosterone, dihydro-
testosterone and cortisol in men with and without alcohol-related
aggression. Journal of Studies on Alcohol,63, 518–526.
Ferree, E.D., Wikelski, M.C., & Anderson, D.J. (2004). Hormonal
correlates of siblicide in Nazca boobies: Support for the challenge
hypothesis. Hormones and Behavior,46, 655–662.
Gladue, B.A., Boechler, M., & McCaul, K.D. (1989). Hormonal re-
sponse to competition in human males. Aggressive Behavior,15,
Hirschenhauser, K., Taborsky, M., Oliveira, T., Canario, A.V.M., &
Oliveira, R.F. (2004). A test of the ‘‘challenge hypothesis’’ in
cichlid ﬁsh: Simulated partner and territory intruder experiments.
Animal Behaviour,68, 741–750.
Killias, M., & Haas, H. (2002). The role of weapons in violent acts:
Some results of a Swiss national cohort study. Journal of Inter-
personal Violence,17, 14–32.
Lieberman, J.D., Solomon, S., Greenberg, J., & McGregor, H.A. (1999).
A hot new way to measure aggression: Hot sauce allocation. Ag-
gressive Behavior,25, 331–348.
Mazur, A., Booth, A., & Dabbs, J.M., Jr. (1992). Testosterone and chess
competition. Social Psychology Quarterly,55, 70–77.
Mazur, A., & Lamb, T.A. (1980). Testosterone, status, and mood in
human males. Hormones and Behavior,14, 236–246.
Muller, M.N., & Wrangham, R.W. (2004). Dominance, aggression, and
testosterone in wild chimpanzees: A test of the ‘challenge hy-
pothesis.’ Animal Behaviour,67, 113–123.
O’Connor, D.B., Archer, J., Hair, W.H., & Wu, F.C.W. (2001). Ex-
ogenous testosterone, aggression, and mood in eugonadal and
hypogonadal men. Physiology and Behavior,75, 557–566.
Preacher, K.J., & Leonardelli, G.J. (2001, March). Calculation for the
Sobel test:An interactive calculation tool for mediation tests.
Retrieved September 2005 from http://www.unc.edu/preacher/
Rowe, R., Maughan, B., Worthman, C.M., Costello, E.J., & Angold, A.
(2004). Testosterone, antisocial behavior, and social dominance in
boys: Pubertal development and biosocial interaction. Biological
Sobel, M.E. (1982). Asymptotic intervals for indirect effects in struc-
tural equations models. In S. Leinhart (Ed.), Sociological meth-
odology 1982 (pp. 290–312). San Francisco: Jossey-Bass.
Wingﬁeld, J.C., Hegner, R.E., Dufty, A.M., Jr., & Ball, G.F. (1990). The
‘‘challenge hypothesis’’: Theoretical implications for patterns of
testosterone secretion, mating systems, and breeding strategies.
American Naturalist,136, 829–846.
(RECEIVED 8/11/05; ACCEPTED 10/17/05;
FINAL MATERIALS RECEIVED 10/31/05)
Volume 17—Number 7 571
Jennifer Klinesmith, Tim Kasser, and Francis T. McAndrew