Bonus Reading 6.

1
Heat and Violence

Source: Anderson, C. A. (2001). Heat and violence. Current

Directions in Psychological Science, 10, 33-38.

Social Psychology
Professor Scott Plous
Wesleyan University

Note: This reading material is being provided free of charge to Coursera

students through the generosity of Craig Anderson and the Association
for Psychological Science, all rights reserved. Unauthorized duplication,
sales, or distribution of this work is strictly forbidden and makes it less
likely that publishers and authors will freely share materials in the
future, so please respect the spirit of open education by limiting your use
to this course.
 CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 33

The heat hypothesis states that hot

Heat and Violence temperatures increase aggressive
motivation and (under some condi-
Craig A. Anderson1 tions) aggressive behavior. The heat
Department of Psychology, Iowa State University, Ames, Iowa effect is the observation of higher
rates of aggression by people who
are hot relative to people who are
For now, these hot days, is the mad cooler. Methodological difficulties
Abstract blood stirring. and the lack of modern statistical
The heat hypothesis states analyses in early studies made cau-
that hot temperatures can in- —William Shakespeare, sal statements risky, but causal is-
crease aggressive motives and Romeo and Juliet, Act 3, Scene 1 sues are crucial. For example, more
behaviors. Although alterna- assaults occur during the summer
tive explanations occasionally Does excessive heat increase months than during other months,
account for some portion of the violence? Social commentators but this could be a spurious artifact
observed increases in aggres- have long noted effects of weather of differences in the daily activities
sion when temperatures are on human behavior and have used people perform at different times
high, none are sufficient to ac- heat-related imagery in their works of the year. Perhaps people are out-
count for most such heat ef- (e.g., Cicero, 106–32 B.C.; Siouxsie side more during the summer, in-
fects. Hot temperatures in- and the Banshees, in their song creasing the opportunity for con-
crease aggression by directly “92°,” 1986). Empirical methods flicts. Routine activities associated
increasing feelings of hostility were first applied to this theory in with summer may increase assault
and indirectly increasing ag- the middle 1700s. Montesquieu rates, and heat-induced discomfort
gressive thoughts. Results show (1748/1989) noted that “you will may play no direct causal role in
that global warming trends may find in the northern climates this increase. Such mediated, or in-
well increase violent-crime peoples who have few vices, direct, heat effects are important in
rates. Better climate controls in enough virtues, and much sincerity their own right, of course.
many institutional settings (e.g., and frankness. As you move to-
prisons, schools, the workplace) ward the countries of the south,
may reduce aggression-related you will believe you have moved
away from morality itself: the live- MODERN STUDIES OF
problems in those settings.
THE HEAT HYPOTHESIS
liest passions will increase
Keywords crime . . . “(p. 234). In the late 1800s
temperature; aggression; vio- and early 1900s, a number of Euro- Modern studies (i.e., post-1950)
lence; global warming pean and North American scholars address these methodological is-
found that rates of violent crime in- sues in several ways. The research
creased during the hottest times of can be classed into three broad cat-
I pray thee, good Mercutio, let’s retire; the year, and were higher in re- egories: (a) field studies, all of
The day is hot, the Capulets abroad, gions with hotter climates (Ander- which focus on some form of ag-
And, if we meet, we shall not ’scape a son, 1989). Perhaps Shakespeare gression; (b) laboratory studies
brawl, was right. with a focus on aggression; and (c)

Copyright © 2001 American Psychological Society

Downloaded from cdp.sagepub.com at WESLEYAN UNIV on August 4, 2013
34 VOLUME 10, NUMBER 1, FEBRUARY 2001

laboratory studies with a focus on Field Studies of Heat and for the temperature, population
aggression-related variables, such Aggressive Behavior size, and socioeconomic status of
as hostile feelings, beliefs, and the cities. As shown by the positive
arousal. Field studies may be categorized path coefficient for the link be-
The results can be characterized according to whether they compare tween temperature and violent
with four summary statements. aggression rates (usually violent- crime in Figure 1 (.43), temperature
First, periodic claims that observed crime rates) across geographic re- was significantly and positively re-
heat effects result solely from arti- gions that are similar in many re- lated to violent-crime rate. That is,
factual processes have, to date, spects but differ in climate, or hotter cities were more violent than
proven false. Second, the ongoing whether they compare aggression cooler cities even after city-to-city
search for conditions under which rates in one geographic region but differences in Southernness, popu-
excessive heat may cause a decline across time periods that differ in lation size, and socioeconomic sta-
in aggression has largely failed. temperature. tus were statistically controlled.
Third, there has been a growing re- However, the path coefficient for
alization that other aggression- Studies Comparing the link between Southernness and
related processes sometimes ob- Geographic Regions violent crime (.14) was not reliably
scure, exaggerate, or modify the Data consistently show that vio- different from zero (no effect), cast-
heat effect. Fourth, a simple ver- lent-crime rates are higher in the ing further doubt on the claim that
sion of the heat hypothesis (e.g., South than in other regions of the a Southern culture of violence is the
Berkowitz, 1993)—that people get United States. Similar patterns ap- sole or primary cause of higher vio-
cranky when uncomfortable—has peared in the older European stud- lent-crime rates in hotter U.S. cities.
ies (Anderson, 1989).
proven surprisingly robust to all
The heat hypothesis is only one
challenges. In short, excessive heat
of several explanations of the U.S. Studies Comparing
appears to cause increases in ag-
version of the hot-region effect. Time Periods
gression in many settings.
One alternative explanation is that, Field studies comparing aggres-
In investigating the relation be-
for some reason, a culture of vio- sion rates in hotter versus cooler
tween heat and aggression, my stu-
lence (e.g., Nisbett, 1993) devel- time periods also support the heat
dents and I have relied on a well-
oped in the U.S. South, and that hypothesis. For example, there are
worn philosophical approach this cultural difference has been about 2.6% more murders and as-
know as triangulation. This in- passed on to present-day inhabit- saults in the United States during
volves examining competing ex- ants. Reasons given for this cul- the summer than other seasons of
planations of the heat effect from tural development differ among the year; hot summers produce a
multiple perspectives. Because the scholars; analyses of who settled bigger increase in violence than
weaknesses of one particular meth- the South, the institution of slav- cooler summers; and violence rates
odology differ from those of other ery, and the effects of being a fron- are higher in hotter years than in
methodologies, an explanation of tier or a herding economy have all cooler years even when various
observed heat effects that works been offered. Nonetheless, claims statistical controls are used (Ander-
across different methodologies is that Southern culture accounts for son et al., 2000). Other time-period
less likely to be invalid than expla- the observed high violent-crime studies provide consistent results.
nations that work only for one or rate in hotter regions of the United Aggression—as measured by as-
two methods. For example, States are contradicted by recent sault rates, spontaneous riots,
changes in routine activities may analyses of violent-crime rates in spouse batterings, and batters be-
be able to explain summer in- 260 U.S. cities (Anderson, Ander- ing hit by pitched baseballs—is
creases in violent crime, but cannot son, Dorr, DeNeve, & Flanagan, higher during hotter days, months,
account for the finding that base- 2000). Latent variable statistical seasons, and years.
ball pitchers are more likely to hit techniques were used to estimate Several studies have examined
batters with a pitched ball on hot the effect of temperature on vio- the heat hypothesis with time peri-
days than on cool days (Reifman, lent-crime rate, while statistically ods even shorter than days. Some
Larrick, & Fein, 1991). The parsi- controlling for the Southernness, have found increases in assaults,
monious explanation is that heat- population size, and socioeco- rapes, and domestic violence at
induced discomfort increases ag- nomic status of the cities. This hotter temperatures (Anderson et
gressive inclinations on the same analysis estimated the effect al., 2000). Studies that have mea-
baseball field and in other natural- of Southernness on violent-crime sured temperature at the exact time
istic settings. rate, while statistically controlling that aggressive behaviors occurred

Published by Blackwell Publishers Inc.

Downloaded from cdp.sagepub.com at WESLEYAN UNIV on August 4, 2013
 CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 35

results. The negative-affect escape

model (Anderson, 1989; Anderson
et al., 2000; Baron & Bell, 1976) pos-
tulates that excessive heat increases
aggression when the total amount
of negative affect a person experi-
ences is in the low to moderate
range (the fight response), but that
excessive heat decreases aggres-
sion when total negative affect gets
too high (the flight response). In
other words, if other aspects of a
particular situation (such as being
insulted) also produce negative af-
fect, then further increases in nega-
tive affect caused by hot tempera-
ture should (according to this
model) lead to escape behavior in-
stead of aggressive behavior. In
Fig. 1. Latent variable model of effects of temperature and Southernness on violent
crime, controlling for population and socioeconomic status (SES). Positive path co-
brief, hot temperatures should pro-
efficients (e.g., the .43 above the line connecting “Temperature” to “Violent Crime”) duce a decline in aggression in
indicate a positive relation between the variables linked by that path. Negative path situations that have other negative-
coefficients would indicate a negative relation between the linked variables. A path affect-producing factors present. A
coefficient of zero would indicate that the two variables are totally unrelated. Solid meta-analysis (i.e., an analysis
lines linking two variables indicate that the associated path coefficient is reliably
different from zero (i.e., is statistically significant). The dashed line indicates that the
combining results across all rel-
link between the two variables is not reliably different from zero. Adapted from evant studies) yielded some sup-
Anderson, Anderson, Dorr, DeNeve, and Flanagan (2000). port for the standard heat effect
(hot temperatures increased ag-
have also yielded the standard heat tempered by the possibility that gression) in lab settings that had
effect. Kenrick and MacFarlane’s unknown extraneous variables few extraneous negative-affect-
(1984) classic study in Phoenix, caused a spurious relation between producing factors present. How-
Arizona, found that aggressive heat and aggression. However, the ever, there was little support for
horn honking increased at hotter two major challenges to the heat the predicted decrease in aggres-
temperatures, but only for drivers hypothesis—changes in routine ac- sion when extraneous negative-
without air-conditioned cars. More tivities and Southern culture—do affect-producing factors were
recently, Vrij, van der Steen, and not fare well from a broad perspec- present (Anderson et al., 2000).
Koppelaar (1994) conducted a field tive. Each can account for a few Many early lab studies, especially
experiment in which Dutch police findings, but neither can account those that used kerosene heaters,
officers performed in a simulated for the broad array of heat effects. suffered from potential suspicion
burglary scenario under hot or The consistency of findings across problems. That is, some partici-
comfortable conditions. Hot offic- many settings and methods pro- pants in those studies may have be-
ers reported more aggressive and vides strong support for the causal come suspicious about the “true”
threatening impressions of the sus- version of the heat hypothesis, purpose of the study, and may
pect, and were more likely to draw even from correlational studies. therefore have behaved in an arti-
their weapon and shoot the suspect Furthermore, the few experimental ficial way. Boyanowsky (1999) re-
(with laser training weapons), rela- and quasi-experimental field stud- cently discussed other method-
tive to officers in the cool condi- ies lend considerable support to ological problems with early lab
tion. the causal interpretation. studies and provided experimental
evidence that when people’s atten-
Summary of Field Studies Laboratory Studies and tion is not focused on temperature
Field studies consistently find Aggressive Behavior (as in most naturalistic settings),
positive associations between un- hot temperatures increase aggres-
comfortable heat and aggression. Mixed Results sion even when additional nega-
Most field studies are correlational, Lab studies of the heat hypoth- tive-affect-producing factors (such
so causal interpretation must be esis have yielded somewhat mixed as being insulted) are present.

Copyright © 2001 American Psychological Society

Downloaded from cdp.sagepub.com at WESLEYAN UNIV on August 4, 2013
36 VOLUME 10, NUMBER 1, FEBRUARY 2001

One recent experiment using and beliefs looks, at first glance, ior. Specifically, GAAM highlights
more subtle ways to manipulate like a cognitive priming effect, au- the fact that any social interaction
temperature than kerosene heaters tomatically increasing the accessi- involves at least two people. Fur-
succeeded in creating conditions bility of aggressive thoughts. How- thermore, aggression can come
that yielded both heat-induced in- ever, hot temperatures do not about through fairly automatic
creases and decreases in aggres- automatically prime aggressive processes (i.e., impulsively) as well
sion. Two factors were involved: an thoughts, at least not in the same as through careful planning. My
ambiguous provocation, followed way that viewing pictures of guns colleagues and I believe that most
by multiple opportunities to retali- does (Anderson, Anderson, & heat-induced increases in aggres-
ate. Under these conditions, my Deuser, 1996). Thus, the effects of sion, including the most violent be-
colleagues and I found an initial heat on attitudes and beliefs are in- haviors, result from distortion of
heat-induced increase in aggres- direct, most likely mediated by the social interaction process in a
sive retaliation, followed by a de- more direct effects of heat on hos- hostile direction. Heat-induced dis-
crease (Anderson et al., 2000). One tile affect. Uncomfortably warm comfort makes people cranky. It in-
explanation of this pattern involves temperatures also produce biases creases hostile affect (e.g., feelings
two separate processes. The initial in the interpretation of observed of anger), which in turn primes ag-
outburst of aggression may have social interactions. Specifically, gressive thoughts, attitudes, prepa-
been the result of heat-induced in- heat seems to increase the likeli- ratory behaviors (e.g., fist clench-
creases in aggressive inclinations hood that ambiguous social inter- ing), and behavioral scripts (such
(via hostile affect and cognition). actions will be interpreted as hav- as “retaliation” scripts). A minor
The later decrease may have been ing aggressive components provocation can quickly escalate,
the result of a social justice norm; (Anderson et al., 2000). Finally, especially if both participants are
hot participants may have decided heat stress decreases performance affectively and cognitively primed
that the initial retaliation was suf- on many cognitive tasks. for hostility by their heightened
ficient. Of course, in most natural level of discomfort. A mild insult is
settings, the initial outburst will it- more likely to provoke a severe in-
self provoke an aggressive re- PSYCHOLOGICAL sult in response when people are
sponse from the victim, initiating PROCESSES UNDERLYING hot than when they are more com-
an escalating cycle of retaliatory THE HEAT EFFECT fortable. This may lead to further
aggression. increases in the aggressiveness of
Numerous fascinating psycho- responses and counterresponses.
Summary of Laboratory logical processes might be in- An accidental bump in a hot and
Studies of Aggressive Behavior volved in the typical effect of high crowded bar can lead to the trad-
In affectively neutral and posi- temperatures on aggression and ing of insults, punches, and (even-
tive circumstances, hot tempera- violence. The simplest and most tually) bullets.
tures cause increases in aggression. powerful ones all revolve around
Recent lab studies show that even the “crankiness” notion. Being un-
in affectively negative circum- comfortable colors the way people NEW RESEARCH
stances, heat causes increases in see things. Minor insults may be DIRECTIONS
initial retaliatory aggression. perceived as major ones, inviting
(even demanding) retaliation. This Many of the basic pieces of this
notion is compatible with several puzzle have been found, but sev-
Laboratory Studies and well-established theories in social eral are still missing. Though re-
Aggression-Related Variables psychology, including Berkowitz’s search on the heat hypothesis has
(1984) cognitive neo-association been carried out for many years,
Heat effects on affective, cogni- theory and Zillmann’s (1983) exci- my colleagues and I believe that
tive, and arousal variables have tation transfer theory. Our own the hardest work lies ahead and
proven quite consistent. Exposure General Affective Aggression that the missing pieces are likely to
to hot temperatures increases heart Model (GAAM; e.g., Anderson et be found in future laboratory stud-
rate, endorsement of aggressive at- al., 2000) explicitly incorporates the ies. Additional work is needed to
titudes and beliefs, and feelings of key aspects of these earlier models, answer the following key questions.
hostility, all the while decreasing including the crankiness notion.
feelings of arousal and comfort. GAAM also includes social in- 1. Does excessive heat bias percep-
The heat-induced increase in en- teraction processes that play a key tions in ongoing social interac-
dorsement of aggressive attitudes role in the genesis of violent behav- tions?

Published by Blackwell Publishers Inc.

Downloaded from cdp.sagepub.com at WESLEYAN UNIV on August 4, 2013
 CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 37

2. Do people in hot conditions—

and who are therefore physi-
ologically aroused (i.e., have an
increased heart rate) but psy-
chologically unaroused (i.e., feel
lethargic)—misattribute some of
their heat-based arousal to mi-
nor provoking social events?
3. Do the cognitive effects of heat
stress interfere with normal
mechanisms for inhibiting ag-
gression?
4. How do escape motives influ-
ence the heat effect? The nega-
tive-affect escape model speci-
fies that escape motives should
play a major role. It predicts that
under some circumstances, in- Fig. 2. Estimated effect of global warming on murders and assaults in the United
creases in heat-induced discom- States. The graph shows the estimated increase in the murder-assault rate and in the
fort will increase the desire to number of murders and assaults (for a population of 270 million) based on three
estimates of the relation between temperature and violence. From Anderson, Ander-
escape more than the desire to
son, Dorr, DeNeve, and Flanagan (2000).
retaliate, and therefore will re-
duce aggression if escape is in-
compatible with aggression. trates just how much of an increase
Recommended Reading
However, no research has ex- can be expected, based on several
plicitly pitted escape motives estimates of the true relation be- Anderson, C.A., Anderson, K.B.,
against aggressive motives. tween temperature and U.S. mur- Dorr, N., DeNeve, K.M., & Flana-
der and assault rates, at several es- gan, M. (2000). (See References)
5. Do social justice processes un- timated levels of global warming. Berkowitz, L. (1993). (See Refer-
derlie the finding that excessive ences)
For example, using the best esti- Geen, R.G. (1990). Human aggression.
heat can at first increase and
mate of how much the violent- Pacific Grove, CA: Brooks Cole.
later decrease aggression?
crime rate will increase for each Reifman, A.S., Larrick, R.P., & Fein,
1 °F increase in temperature (i.e., S. (1991). (See References)
4.58), we see that a 2 °F increase in
CURRENT IMPLICATIONS average temperature predicts an
Acknowledgments—I thank Kathryn
increase of about 9 more murders Anderson, Brad Bushman, and Kristina
A broad view of the research— or assaults per 100,000 people, or DeNeve for their helpful comments on
more than 24,000 additional mur- this article.
triangulation—suggests that in
many settings hot temperatures ders and assaults per year in a
cause increases in aggression. population of 270 million.
Note
There are conditions that limit the There are numerous institu-
generality of this conclusion, but tional settings in which aggression 1. Address correspondence to Craig
the overall pattern of data is im- is a problem and in which tempera- A. Anderson, Department of Psychol-
pressive and convincing. ture can be controlled. Schools, ogy, W112 Lagomarcino Hall, Iowa
The implications of this general prisons, and a wide variety of State University, Ames, IA 50011-3180;
e-mail: caa@iastate.edu.
conclusion are many. Consider the workplaces are good targets for in-
finding that hot years produce in- tervention. Research on the effects
creases in violent-crime rates. If of better climate control in such set- References
this heat effect is truly caused by tings might well show that the Anderson, C.A. (1989). Temperature and aggres-
heat-induced increases in aggres- additional costs are outweighed sion: Ubiquitous effects of heat on occurrence
of human violence. Psychological Bulletin, 106,
sive motivation, then increased by the benefits—better learning, 74–96.
violence can be added to the list of lower incarceration costs, less Anderson, C.A., Anderson, K.B., & Deuser, W.E.
(1996). Examining an affective aggression
negative social consequences of property damage, and increased framework: Weapon and temperature effects
global warming. Figure 2 illus- productivity. on aggressive thoughts, affect, and attitudes.

Copyright © 2001 American Psychological Society

Downloaded from cdp.sagepub.com at WESLEYAN UNIV on August 4, 2013
38 VOLUME 10, NUMBER 1, FEBRUARY 2001

Personality and Social Psychology Bulletin, 22, Berkowitz, L. (1993). Aggression: Its causes, conse- Reifman, A.S., Larrick, R.P., & Fein, S. (1991). Tem-
366–376. quences, and control. New York: McGraw Hill. per and temperature on the diamond: The
Anderson, C.A., Anderson, K.B., Dorr, N., DeNeve, Boyanowsky, E. (1999). Violence and aggression in heat-aggression relationship in major league
K.M., & Flanagan, M. (2000). Temperature and the heat of passion and in cold blood. Interna- baseball. Personality and Social Psychology Bul-
aggression. In M. Zanna (Ed.), Advances in ex- tional Journal of Law and Psychiatry, 22, 257–271. letin, 17, 580–585.
perimental social psychology (Vol. 32, pp. 63–
Kenrick, D.T., & MacFarlane, S.W. (1984). Ambient Vrij, A., van der Steen, J., & Koppelaar, L. (1994).
133). New York: Academic Press.
temperature and horn-honking: A field study Aggression of police officers as a function of
Baron, R.A., & Bell, P.A. (1976). Aggression and
of the heat/aggression relationship. Environ- temperature: An experiment with the Fire
heat: The influence of ambient temperature,
ment and Behavior, 18, 179–191. Arms Training System. Journal of Community
negative affect, and a cooling drink on physi-
cal aggression. Journal of Personality and Social Montesquieu, C. (1989). The spirit of the laws (A. and Applied Social Psychology, 4, 365–370.
Psychology, 33, 245–255. Cohler, B. Miller, & H. Stone, Trans.). New Zillmann, D. (1983). Arousal and aggression. In R.
Berkowitz, L. (1984). Some effects of thoughts on York: Cambridge University Press. (Original Geen & E. Donnerstein (Eds.), Aggression:
anti- and prosocial influences of media events: work published 1748) Theoretical and empirical reviews, Vol. 1. Theoret-
A cognitive-neoassociation analysis. Psycho- Nisbett, R.E. (1993). Violence and U.S. regional ical and methodological issues (pp. 75–101). New
logical Bulletin, 95, 410–427. culture. American Psychologist, 48, 441–449. York: Academic Press.

Published by Blackwell Publishers Inc.

Downloaded from cdp.sagepub.com at WESLEYAN UNIV on August 4, 2013