physical aggression as a function of perceived fighting ability among male and female prisoners

11
AGGRESSIVE BEHAVIOR Volume 33, pages 563–573 (2007) Physical Aggression as a Function of Perceived Fighting Ability Among Male and Female Prisoners John Archer Department of Psychology, University of Central Lancashire, Preston, United Kingdom : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : Study 1 used two scenarios depicting provocations to investigate whether prisoners’ willingness to engage in physical aggression would vary according to the perceived fighting ability (Resource Holding Power [RHP]) of the opponent. RHP was operationalized in terms of three levels of the size, number of allies, and reputation. The sample (both sexes; N 5 1,253) was taken from a range of British prisons. Unexpectedly, the prisoners said that they would be less likely to attack a weaker opponent than an equally matched one. The predicted decline from equally matched to more formidable opponents was found only under some conditions. A further sample of male prisoners (N 5 260), completed a similar provocation scenario, with response options covering fear, several forms of aggression, the importance of the provocation for reputation, and that the provocation was unimportant. Fear increased with RHP as predicted. Lower physical aggression to a low RHP opponent was again found, and it was associated with a lesser impact on reputation of an insult from this source. There was also a lower likelihood of responding with physical aggression to a high RHP opponent, as originally predicted. Implications of the findings for motivational theories of aggression are discussed, and also the applicability to humans of concepts from game theory models of the evolution of fighting strategies. Aggr. Behav. 33:563–573, 2007. r 2007 Wiley-Liss, Inc. : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : Keywords: resource holding power; physical aggression; reputation; sex differences; evolution INTRODUCTION The two studies described in this article address two main questions. The first arose from previous findings that young people respond to scenarios depicting provocation in accordance with the perceived fighting ability of the opponent. Study 1 investigated whether this also occurred in a large sample of prisoners. The second question arose from the unexpected finding in this study that the prison- ers said that they would be less rather than more likely to attack a weaker opponent than an equally matched one. In Study 2, another sample of prison- ers completed the same scenarios but with a wider range of response options that assessed possible reasons for this unexpected finding: in particular, whether responding to provocation by a weaker opponent was viewed as being of lesser importance to the person’s reputation, was assessed. Subsidiary issues investigated were whether men would show more overall readiness than women to report that they would use physical aggression in the scenarios; whether a scenario involving an insult to friends and family would be more provoking than one involving an insult to the person; and whether those designated as bullies would show more overall readiness to say that they would use physical aggression than those showing no bullying. The starting point for these studies was the concept of Resource Holding Power (RHP), which originated from a specific game theory model of the evolution of fighting strategies in animals. Parker [1974] showed that if an animal can assess an opponent’s fighting ability (RHP), and use this to decide whether or not to escalate a dispute, this strategy will be stable under a wide range of evolutionary conditions. Consistent with this, throughout the animal kingdom, cues to fighting ability—such as size, weapons, and numbers of allies—have been shown to predict which of two potential protagonists will withdraw from the conflict first and which one will win [Archer, 1988; Published online 1 August 2007 in Wiley InterScience (www. interscience.wiley.com). DOI: 10.1002/ab.20221 Received 10 July 2006; Revised 2 January 2007; Accepted 15 January 2007 Correspondence to: John Archer, University of Central Lancashire, Preston, UK. E-mail: [email protected] r 2007 Wiley-Liss, Inc.

Upload: john-archer

Post on 06-Jun-2016

222 views

Category:

Documents


8 download

TRANSCRIPT

Page 1: Physical aggression as a function of perceived fighting ability among male and female prisoners

AGGRESSIVE BEHAVIOR

Volume 33, pages 563–573 (2007)

Physical Aggression as a Function of Perceived FightingAbility Among Male and Female PrisonersJohn Archer�

Department of Psychology, University of Central Lancashire, Preston, United Kingdom

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

Study 1 used two scenarios depicting provocations to investigate whether prisoners’ willingness to engage in physical aggressionwould vary according to the perceived fighting ability (Resource Holding Power [RHP]) of the opponent. RHP was operationalizedin terms of three levels of the size, number of allies, and reputation. The sample (both sexes; N5 1,253) was taken from a range ofBritish prisons. Unexpectedly, the prisoners said that they would be less likely to attack a weaker opponent than an equallymatched one. The predicted decline from equally matched to more formidable opponents was found only under some conditions.A further sample of male prisoners (N5 260), completed a similar provocation scenario, with response options covering fear,several forms of aggression, the importance of the provocation for reputation, and that the provocation was unimportant. Fearincreased with RHP as predicted. Lower physical aggression to a low RHP opponent was again found, and it was associated with alesser impact on reputation of an insult from this source. There was also a lower likelihood of responding with physical aggressionto a high RHP opponent, as originally predicted. Implications of the findings for motivational theories of aggression are discussed,and also the applicability to humans of concepts from game theory models of the evolution of fighting strategies. Aggr. Behav.33:563–573, 2007. r 2007 Wiley-Liss, Inc.

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

Keywords: resource holding power; physical aggression; reputation; sex differences; evolution

INTRODUCTION

The two studies described in this article addresstwo main questions. The first arose from previousfindings that young people respond to scenariosdepicting provocation in accordance with theperceived fighting ability of the opponent. Study 1investigated whether this also occurred in a largesample of prisoners. The second question arose fromthe unexpected finding in this study that the prison-ers said that they would be less rather than morelikely to attack a weaker opponent than an equallymatched one. In Study 2, another sample of prison-ers completed the same scenarios but with a widerrange of response options that assessed possiblereasons for this unexpected finding: in particular,whether responding to provocation by a weakeropponent was viewed as being of lesser importanceto the person’s reputation, was assessed.Subsidiary issues investigated were whether men

would show more overall readiness than women toreport that they would use physical aggression in thescenarios; whether a scenario involving an insult tofriends and family would be more provoking thanone involving an insult to the person; and whether

those designated as bullies would show moreoverall readiness to say that they would use physicalaggression than those showing no bullying.The starting point for these studies was the

concept of Resource Holding Power (RHP), whichoriginated from a specific game theory model ofthe evolution of fighting strategies in animals.Parker [1974] showed that if an animal can assessan opponent’s fighting ability (RHP), and use thisto decide whether or not to escalate a dispute,this strategy will be stable under a wide range ofevolutionary conditions. Consistent with this,throughout the animal kingdom, cues to fightingability—such as size, weapons, and numbers ofallies—have been shown to predict which of twopotential protagonists will withdraw from theconflict first and which one will win [Archer, 1988;

Published online 1 August 2007 in Wiley InterScience (www.

interscience.wiley.com). DOI: 10.1002/ab.20221

Received 10 July 2006; Revised 2 January 2007; Accepted 15 January

2007

�Correspondence to: John Archer, University of Central Lancashire,

Preston, UK. E-mail: [email protected]

r 2007 Wiley-Liss, Inc.

Page 2: Physical aggression as a function of perceived fighting ability among male and female prisoners

Archer and Huntingford, 1994]. These cues alsopredict the duration of aggressive encounters, thosebetween equally matched opponents lasting longerthan those between opponents who are unequallymatched.The concept of RHP assessment has only recently

been transferred to the study of human aggression.Sell et al. [2005] found that lifting strength was posi-tively associated with a history of fights, and with agreater proneness to anger, among two samples ofyoung men. Other studies have found an associationbetween physical size and dominance or aggression,in young children [Pellegrini et al., 2007], adolescentboys [Tremblay et al., 1998], and adult men [Felson,1996], and between an index of upper body strengthand physical aggression in a sample of young men[Archer and Thanzami, 2007].Following an ethnographic study of young men in

the context of a ‘‘night out’’ in bars and clubs[Benson and Archer, 2002], Archer and Benson[2007] carried out three studies in which RHP wasmanipulated in scenarios involving a young personin a bar with friends. Teenage men, around 18 yearsof age, reported less likelihood of respondingwith physical aggression when insulted, as threelevels of three RHP cues (size, allies, and reputation)increased, effects that were accentuated by combina-tions of high RHP cues. Similar results were foundin another sample of young men, using two levels ofprovocation and a wider range of response strate-gies. These findings were essentially replicated instudent samples of men and women in their earlytwenties, although the effects were weaker at lowlevels of RHP. In this sample, the lesser likelihood ofresponding with physical aggression to an opponentwith higher RHP was stronger than the greaterlikelihood of responding with physical aggression toone with lower RHP. Overall, men showed a greaterreadiness than women to say that they wouldrespond with physical aggression, but the variationaccording to RHP was very similar in the two sexes.

STUDY 1

We would expect the perception of RHP found inthe previous studies to extend to other samples andother circumstances. In this study, RHP wasinvestigated among prisoners, where it might bemore marked than among non-incarcerated samplesfor several reasons: first, they are in an all-maleenvironment, away from the civilizing influences ofwomen [Courtwright, 1996]; second, they are un-related strangers who have no kinship ties orcommon interests [Daly and Wilson, 1988]; third,

they are in an environment that, although in somerespects closely supervised, has its own code of notinforming on (‘‘grassing up’’) other prisoners’actions [e.g., Ireland, 2002]. These would all providereasons for prisoners valuing the principle that aprovocation or insult must be avenged by a violentact, so as to establish a reputation as someone whois not to be provoked. This principle has beendiscussed in connection with the culture of honor[Nisbett and Cohen, 1996], which is based on theneed to establish a reputation as someone who isdangerous to challenge, under circumstances wherethere is no consensual moral code or effective rule oflaw [Daly and Wilson, 1988].Both men and women prisoners were sampled, so

that the previous finding [Archer and Benson, 2007]that both sexes responded to RHP cues in essentiallythe same way, could be assessed. In addition, thefinding that men were more likely than women tostate that they would escalate to physical aggression,over all RHP levels, could be assessed. This would beconsistent with considerable evidence of a sexdifference in proneness to physical aggression [Archer,2004; Eagly and Steffen, 1986; Hyde, 1986], and withother measures from the present prison sample[Archer et al., 2007]. Whether prisoners identified asbullies would generally be more ready to respond toprovocation with physical aggression irrespective ofthe RHP of the opponent was also assessed: thiswould follow if those identified as bullies are generallyaggressive individuals, as identified by conventionalmeasures of aggression [Ireland and Archer, 2004;Ireland et al., 2007]. The sample used in this study wastaken from a large-scale investigation of bullying inBritish prisons, whose main findings concerned theincidence of bullying behavior and how this is relatedto a variety of other measures [Archer et al., 2007;Ireland et al., 2007].Two scenarios were used, one involving an insult

to the prisoner at a mealtime, and the other an insultabout friends and family who had just visited thatprisoner. Both involved insults in a place wherethere are other prisoners, but the first was directed atthe person and the second to their significant others.In the previous study [Archer and Benson, 2007],insults to a partner were consistently rated as moreprovoking than insults to the self. It is thereforeexpected that the second scenario will overall bemore provoking than the first one.

Method

RHP questionnaires. Two scenarios followedby 27-item RHP questionnaires were used, with a

564 Archer

Aggr. Behav. DOI 10.1002/ab

Page 3: Physical aggression as a function of perceived fighting ability among male and female prisoners

within-subjects factorial design, factors being thesize of the opponent relative to the participant, thenumber of potential allies with them, and theopponent’s reputation. There were three levels foreach factor, high, medium, or low. The scenariosplaced the respondent in two contexts. The firststated: ‘‘You are at the servery and another prisonerstarts to call you abusive names in front of every-one’’. The second stated: ‘‘You are on your wayback from a visit and another prisoner startspointing at you and making insulting remarks aboutyour friends and family’’. Respondents were askedto rate their likelihood of retaliating with physicalaggression on a scale from 1 to 5 (15 not at alllikely; 25 unlikely; 35 possibly; 45 likely; 55 verylikely), for each of 27 descriptions of the potentialopponent. For example, the highest level of RHP oneach factor was ‘‘He (or she) looks bigger than you,has more friends with him (or her) than you do, andhas a reputation for being tough’’. The lowest levelof RHP on each feature was ‘‘He (she) is smallerthan you, has fewer friends with him (her) than youdo, and has a reputation for being all talk’’. Valuesare expressed as mean scores, for example for malesand females and the two scenarios, and for eachlevel of RHP.

The direct and indirect prisoner behaviorchecklist. This is a method of gathering informa-tion on bullying behavior by assessing the presenceor absence of a number of discrete behaviors[Ireland, 1999]. It has been used extensively withmen, women, young, and adult prisoners [Ireland,2002]. The checklist does not use the term bullying,in view of problems with applying this a term toprisoners [Ireland, 2002]. Instead, it involves cate-gories of behavior that prisoners regard as indicativeof bullying [Ireland and Archer, 1996]: the ques-tionnaire asks about behavior indicative of bullyingothers and of being bullied. It assesses directbullying, including physical (e.g., ‘‘I have hit orkicked another offender’’), theft-related, psycholo-gical or verbal (e.g., ‘‘I have called someone namesabout their offence or charge’’), sexual bullying, andalso indirect bullying, including gossiping, spreadingrumors, and ostracizing others (e.g., ‘‘I havedeliberately ignored someone’’). In total, 65 itemsindicative of bullying are included, 48 involvingdirect forms and 17 involving indirect forms.Prisoners were asked to identify which behaviorhad occurred to them in the previous week andwhich they had engaged in. For the purpose of thepresent study, prisoners were classified according towhether they had endorsed at least one bullying item(‘‘bully’’) as victim and/or perpetrator, or had not

done so. Further details of this part of the study canbe found in Ireland et al. [2007].

Participants and procedure. A total of 1,253adult prisoners (21 years of age and over) weresampled from 11 UK prisons. Of these, 728 weremen (58.1%) and 525 women (41.9%). The responserate was 40%. The sample was drawn from closed,medium, and medium-high security settings. Themean age was 32.1 years (SD 9.9); men: 32.7 years(SD 10.7); women: 31.4 years (SD 8.7): 88.8% wereof white ethnic origin, 3.8% Afro-Caribbean, and2.8% from the Asian subcontinent. The meansentence length was 43.6 months and the mean totallength of time served in penal institutions through-out their lives was 50 months; 32.6% were servingfor violent offences, 26.7% for acquisitive offences,18.5% for drug-related offences, 12.5% for otheroffences (e.g., motoring offences), and 9.6% for sexoffences.The sample included all prisoners based on the

prison wing/house at the time of the study. In mostprisons, participants completed the questionnairepack, which contained the RHP scenarios and thedirect and indirect prisoner behavior checklist(DIPC), on their own in their cells. Owing topractical issues at two prisons, participants wereapproached during education or workshop classes.Those recruited in this way sat separately and weresupervised as they completed the questionnaires. Itwas stressed that names and prison numbers werenot required and that the questionnaire onlyinvolved basic information (e.g., age, sentencelength, and total time served, offence type andethnic origin). All were informed that they hadeither 1 or 2 hr (depending on the prison regime) tocomplete the questionnaires and that these would becollected at the end of that time. The questionnaireswere placed in an unmarked envelope and sealed.

Data analysis. The main analysis involved amixed design analysis of variance (ANOVA), invol-ving 2 (sex)� 2 (scenario 1 or 2)� 3 (size)� 3(friends)� 3 (reputation). Significant main effectsof the RHP factors were followed up by contrastsbetween the various levels. Significant two or three-way analyses were followed up by simpler ANO-VAs, and if appropriate, by contrasts between RHPlevels. The impact of bully and victim status wasanalyzed separately in a 2� 2 ANOVA.

Results

Overview of the main analysis. First, themain effects of the factorial ANOVA are reported.These concern sex differences, difference between

565Physical Aggression and Perceived Fighting Ability in Prisoners

Aggr. Behav. DOI 10.1002/ab

Page 4: Physical aggression as a function of perceived fighting ability among male and female prisoners

scenarios, and within-subjects differences for thethree RHP manipulations. Second, interactions areconsidered, first between sex and any one of thewithin-subjects factors; second between scenario andthe three types of RHP; and third, between the threetypes of RHP. Finally, the analysis of bully andvictim status is presented.

Main effectsSex. Consistent with the prediction that men

would be generally more likely than women toreport that they would respond to an insult withphysical aggression, the between-subjects factor(‘‘sex’’) showed a significant main effect(F(1,996)5 10.16; Po.001), with men showing high-er scores than women (M5 2.82 vs. 2.57 per item;d5 .21), a difference comparable with that found fora student sample [Archer and Benson, 2007].

Scenario. There was a significant and largemain effect of scenario (F (1,996)5 72.7; Po.0001;Z2 5 .068). The second scenario (insulting remarksabout friends and family) produced higher scoresthan the first (insult to self at the servery): M5 2.79and 2.60 per item.

RHP manipulations. As shown in Table I, therewere significant main effects for size (Z2 5 .021),number of friends (Z2 5 .012), and reputation(Z2 5 .044). For size and reputation, level 1 (thelowest) showed a significantly smaller value thanlevels 2 and 3, which did not differ significantly. Forfriends, levels 1 and 3 showed significantly lowervalues than level 2, and levels 1 and 3 did not differsignificantly. The differences between levels 1 and 2are the opposite of the predicted direction, and thatfound previously [Archer and Benson, 2007], in thatlow RHP levels attracted a lesser likelihood ofphysical aggression than the two higher levels. Forfriends, the highest RHP level did show the expectedlower value than the medium level.

Interactions involving sex of participantSex � scenario. This interaction was signi-

ficant (F(1,996)5 18.85; Po.0001; Z2 5 .019): thesex difference was larger (d5 .26; t(2,1004)5 4.14;Po.0001) in scenario 2 (insulting remarks aboutfriends and family) than in scenario 1 (d5 .12;t(2,1020)5 1.96; P5 .05). Thus, the more provokingscenario (see above) also produced the larger sexdifference (in the male direction).

Sex and reputation. Sex also interacted signifi-cantly with reputation (F(2,1992)5 5.6; P5 .004;Z2 5 .006). Men showed a large difference acrossthe three levels of reputation (F(2,1154)5 44.81;Po.0001; Z2 5 .072): level 1 was significantly lower(Po.0001; d5 .27) than levels 2 and 3, which didnot differ (d5 .02). Although women also differed

across the three levels (F(2,838)5 9.99; Po.0001),this was a smaller effect (Z2 5 .023): again, level 1was significantly lower than levels 2 and 3 (Po.001and .002; d5 .15), which did not differ (d5 .02).

Two-way interactions between scenario andRHP factors. Each RHP factor showed a signi-ficant interaction with scenario (size: F(2,1992)5

7.60; Po.001; Z2 5 .008; friends: F(2,1992)5 6.66;Po.001; Z2 5 .007; reputation: F(2,1992)5 6.03;P5 .002; Z2 5 .006). These interactions werefollowed up by conducting separate 3� 3� 3ANOVAs for the two scenarios. Table I (lines 4–9)shows the findings: there were larger differencesbetween the levels of each RHP factor in the firstscenario (which was overall less provoking) than inthe second scenario, for all three RHP cues.

Two-way interactions between RHP factors.Only one of the interactions between the three RHPfactors was significant, and this only marginally so:it involved size and reputation (F (4,3984)5

2.39; P5 .049; Z2 5 .002). The third section ofTable I shows the results of one-way ANOVAscarried out at each level of size for reputation. Atall three levels of size, there was a large overalldifference between reputation levels, as a result oflower scores at the lowest level of reputation. Atlevels 1 and 2 of size, levels 2 and 3 of reputationdid not differ significantly, but at the highest levelof size, a significant reduction was found fromlevels 2 to 3 of reputation, albeit of a smallermagnitude than the difference between level 1 andthe other two levels (F(1,997)5 5.35; P5 .02;Z2 5 .005). Thus, when the scenario involved a largeopponent, a reputation for being tough did result ina lowered score, as predicted.

Three-way interactions between scenarioand RHP factors. There were two significantthree-way interactions between scenario and one ofthe RHP factors. The first involved size and friends(F(4,3984)5 3.33; P5 .01; Z2 5 .003). The secondinvolved size and reputation (F(4,3984)5 7.17;Po.001; Z2 5 .007).

Scenario � size � friends. A follow-up ana-lysis, involving separate one-way ANOVAs at eachlevel of scenario and size, showed a different patternof results for the two scenarios that can besummarized as follows (shown in the fourth sectionof Table I). In the first (less provoking) scenario,with the smallest size, there were no differencesbetween the numbers of friends: all these valueswere low, suggesting that it was the small sizethat influenced the scores. With both the equaland largest size, scores were highest with the samenumber of friends, and were equally low when there

566 Archer

Aggr. Behav. DOI 10.1002/ab

Page 5: Physical aggression as a function of perceived fighting ability among male and female prisoners

were more or fewer friends. Again, this showedevidence of a reduction at the highest RHP level.There was a different pattern for the second (moreprovoking) scenario. With the smallest size, therewas no difference in scores for fewer and the samenumbers of friends, but the value for more friendswas significantly reduced, again as predicted. Withthe equal size, there were lower values for fewerfriends than for an equal number or more, whichdid not differ. With the largest size, there wereno differences according to the numbers of friends.

Scenario � size � reputation. Follow-upANOVAs (shown in the final section of Table I)indicated consistently lower values for level 1reputation than for the other two levels, across eachlevel of scenario and size, consistent with the maineffect of reputation and the follow-up contrastsshown in section 1 of Table I. What is specific to thisanalysis is that for scenario 1 (where RHP effectswere stronger than those in scenario 2), for sizes 2and 3, there was also a significant reduction from

medium to high levels of reputation. In these twoinstances, therefore, the responses were consistentwith the prediction of a decrease from medium tohigh RHP levels. This occurred particularly at thehighest level of size, again suggesting that for alarger sized opponent, a reputation for being toughreduced the likelihood of responding with physicalaggression.

Overall bully-category differences. A 2 (bullyvs. non-bully) � 2 (victim vs. non-victim) ANOVAwas computed for the overall scores on the RHPitems. Those showing one or more acts of bullyingbehavior on the DIPC had significantly (F (1,972)5

34.25; Po.0001; Z25 .034) higher values per item(M5 3.01) than did non-bullies (M5 2.49) for thecombined scores (both scenarios). Those receivingone or more acts of bullying behavior on the DIPCdid not differ significantly (F(1,972)5 1.63; NS[nonsignificant]) from those who did not (M5 2.84vs. 2.56). The interaction between these two factorswas nonsignificant (F(1,972)5 .15; NS), indicating

TABLE I. Main Effects and Follow-up Tests for the Male and Female Participants in Study 1 (Situation 1: Verbal Abuse at

Servery; Situation 2: Insults to Family and Friends)

Factor (df) F Level 1 Level 2 Level 3 Contrasts

(1) Main effects of size, friends, and reputation in the main analysis

Size [2, 1992] 21.18�� 2.67 2.71 2.71 1 vs. 2��; 1 vs. 3��

Friends [2, 1992] 11.90�� 2.69 2.71 2.69 1 vs. 2��; 2 vs. 3��

Reputation [2, 1992] 46.12�� 2.63 2.73 2.73 1 vs. 2��; 1 vs. 3��

(2) 3 (size)� 3 (friends)� 3 (reputation) ANOVAs for each scenario

Scenario 1 Size [2, 2042] 28.65�� 2.56 2.62 2.62 1 vs. 2��; 1 vs. 3��

Friends [2, 2042] 15.45�� 2.58 2.62 2.59 1 vs. 2��; 2 vs. 3��

Reputation [2, 2042] 55.42�� 2.51 2.65 2.64 1 vs. 2��; 1 vs. 3��

Scenario 2 Size [2, 2010] 6.37��� 2.78 2.80 2.80 1 vs. 2��; 1 vs. 3���

Friends [2, 2010] .90 2.79 2.80 2.79

Reputation [2, 2010] 29.83�� 2.74 2.82 2.82 1 vs. 2��; 1 vs. 3��

(3) One-way ANOVAs for reputation at each level of size

Size 1 [2, 1994] 36.83�� 2.61 2.72 2.72 1 vs. 2��; 1 vs. 3��

Size 2 [2, 1994] 47.80�� 2.65 2.77 2.78 1 vs. 2��; 1 vs. 3��

Size 3 [2, 1994] 36.36�� 2.66 2.77 2.75 1 vs. 2��; 1 vs. 3��; 2 vs. 3���

(4) One-way ANOVAs for friends at each level of size, for scenarios 1 and 2

Scenario 1 Size 1 [2, 2044] 2.10 2.55 2.58 2.57

Size 2 [2, 2044] 9.30�� 2.62 2.67 2.63 1 vs. 2��; 2 vs. 3�

Size 3 [2, 2044] 9.47�� 2.61 2.67 2.62 1 vs. 2��; 2 vs. 3�

Scenario 2 Size 1 [2, 2010] 4.77� 2.82 2.81 2.79 1 vs. 3�; 2 vs. 3�

Size 2 [2, 2044] 3.42��� 2.82 2.85 2.84 1 vs. 2�

Size 3 [2, 2044] 1.04 2.83 2.82 2.84

(5) One-way ANOVAs for reputation at each level of size, for scenarios 1 and 2

Scenario 1 Size 1 [2, 2044] 38.18�� 2.47 2.61 2.62 1 vs. 2��; 1 vs. 3��

Size 2 [2, 2044] 41.94�� 2.55 2.67 2.70 1 vs. 2��; 1 vs. 3��; 2 vs. 3���

Size 3 [2, 2044] 36.64�� 2.55 2.70 2.64 1 vs. 2��; 1 vs. 3��; 2 vs. 3��

Scenario 2 Size 1 [2, 2010] 15.23�� 2.75 2.84 2.82 1 vs. 2��; 1 vs. 3��

Size 2 [2, 2010] 28.93�� 2.76 2.87 2.87 1 vs. 2��; 1 vs. 3��

Size 3 [2, 2010] 19.31�� 2.77 2.85 2.87 1 vs. 2��; 1 vs. 3��

�Po.01; ��Po.001; ���Po.05.F values for main effects or follow-up ANOVAs for the interactions; mean values for the three levels of each factor; contrasts (LSD), for the threelevels (low, medium, and high) within the factors.ANOVA, analysis of variance.

567Physical Aggression and Perceived Fighting Ability in Prisoners

Aggr. Behav. DOI 10.1002/ab

Page 6: Physical aggression as a function of perceived fighting ability among male and female prisoners

that those who were both bullies and victims did nothave higher values than those who were only bullies.

Summary of the main analysis. Overall, forall three RHP manipulations, level 1 (the opponentwith the least retaliatory power) was less likely toattract physical aggression than was level 2 (theopponent with equal retaliatory power). For theRHP manipulations, size and reputation, levels 2and 3 (the opponent with greater retaliatory power)did not differ overall. Only in the case of numbers offriends was there a lower level of physical aggressionat the highest RHP level than at the medium level.Follow-up analyses to the significant interactionsshowed that this effect occurred for sizes 2 and 3 inthe first scenario (insult to the person) and for size 1in the second scenario (insult to friends and family).The same pattern was also found under someconditions for reputation. At the highest level ofsize, a reputation for being tough reduced thelikelihood of responding with physical aggression,and this was most pronounced in the first scenario,an insult at a mealtime.In relation to the subsidiary hypotheses, men

showed a greater overall tendency to respond withphysical aggression, but this difference was signifi-cantly higher in the second scenario (insult to friendsand family) than in the first (insult to self). Aspredicted, the second scenario (insult to friends andfamily) produced higher scores than the first (insultto self). However, the RHP differences were morepronounced in the first, less provoking, situation, forall three factors. Men were more affected by thedifferent levels of the opponent’s reputation thanwere women, i.e. there was a greater reduction foran opponent who was ‘‘all talk’’ compared with onewho could ‘‘look after himself’’. Bullies showed alarger overall tendency to respond with physicalaggression, whatever the level of RHP or provoca-tion, than those who did not admit acts of bullying.Victim status did not affect these overall scenarioresponses.

Discussion

The main hypothesis was that there would be lesswillingness to use physical aggression in a provokingsituation as the RHP of the potential opponentincreased. Although in most cases, medium and highlevels of RHP did not differ, there were signs of theexpected decrease in scores from levels 2 (equal tothe participant) to 3 (more formidable), but onlyunder some of the conditions: for the number offriends, in particular scenario 1 (insult to the person)at the two larger opponent sizes. A reduction was

also found for reputation at the highest level of size,and this was more pronounced in the first scenario.This was the extent to which the findings supportedthe hypothesis that willingness to escalate to physicalaggression would decline with increasing RHP.The most noteworthy finding was contrary to the

main hypothesis: the reported willingness to escalateto physical aggression was lower when the opponenthad less retaliatory power than when he or she had alevel equal to that of the participant. This wasconsistent for all three RHP manipulations (size,number of friends, and reputation) and applied toboth sexes and both scenarios. These findings arecontrary to those found previously with threesamples of non-incarcerated individuals [Archerand Benson, 2007].In seeking an explanation for the low level of

reported physical aggression when RHP levels arelow, in game theory models of animal fighting, asmaller animal may be more likely to attack a largerone than vice versa if there is a payoff asymmetrysuch that the smaller animal stands to gain morefrom the encounter than the larger one [Parker andRubenstein, 1981]. This could apply to humanconflicts involving a perceived insult if the impor-tance of the insult in terms of reputational damage isdiminished when it comes from someone with lowRHP. Thus, an insult from a low RHP individualwould then not constitute a challenge or damage tothe recipient’s reputation. In the Introduction to thisstudy, I mentioned cultural rules such as those thathave been described as the culture of honor, andsuggested that the need to respond with violence to achallenge to the person’s perceived status wouldoperate in a prison setting, because, in the absenceof an effective rule of law, not to respond in this waywould lay the person open to having his resourcesbeing taken by the challenger [Daly and Wilson,1988, 1990; Nisbett and Cohen, 1996]. However, ifthe person making the challenge is seen as not beinga credible threat, their challenge may be regardedas not serious, and can be disregarded, therebyavoiding the costs of a physical encounter. Suchcosts may be incurred if the physical attack takesplace where prison officers were likely to intervene,as would be the case with the present scenarios. Thisreasoning informs Study 2, which used a version ofthe same basic scenario, but with a series of ratingsdesigned to assess: (1) the perceived importance ofthe incidence; and (2) the likelihood of respondingwith direct and with delayed aggression [Archer andBenson, 2007].Before moving on to this second study, some other

findings from the first study are noted. Men were

568 Archer

Aggr. Behav. DOI 10.1002/ab

Page 7: Physical aggression as a function of perceived fighting ability among male and female prisoners

more likely than women to report that that theywould show physical aggression following a provo-cation. This was consistent with previous findingsusing the same scenario method [Archer andBenson, 2007], and with many previous studiesusing a variety of self-report measures [Archer,2004]. The second scenario, which involved an insultto friends and family, produced higher scores thanthe first one, which involved an insult to the personhim or her self. These findings are consistent withthose in the sample of young men studied previously[Archer and Benson, 2007] showing that insults tothe man’s girlfriend was a more potent provocationthan a personal insult.RHP differences were more pronounced in the

first, less provoking, situation, for all three factors,indicating that such differences were less pro-nounced when the provocation was stronger. Menwere also more affected by the different levels of theopponent’s reputation than were women. Thus, formen, there was a greater reduction for an opponentwho was ‘‘all talk’’ compared with one who could‘‘look after himself’’. This may be related to thegreater emphasis on a physically based reputationamong men than women [Archer, 1994; Daly andWilson, 1988, 1990; Nisbett and Cohen, 1996],although in this case it would mean that whencertain men make a challenge to another man, theycould be disregarded.Those who were perpetrators of bullying reported

more overall willingness to escalate to physicalaggression than those who were not perpetrators.This is consistent with other findings from the samesample showing that prisoners who report perpe-trating bullying also report higher values forphysical aggression in a scenario involving provoca-tion [Archer et al., 2007], and with other findingsusing a standard trait measure of aggression [Irelandand Archer, 2004].

STUDY 2

The second study involved a conceptual replica-tion of the first with a sample of male prisoners,using three levels of RHP (low, equal, and high),derived from those used in Study 1. A wider range ofresponses was provided, to assess the potency of theRHP cues to induce fear, other aggressive responses,the need to be ‘‘get even’’, and that the provocationwas seen to be unimportant. The study thereforeinvolved: (1) the overall effectiveness of the RHPmanipulations to induce fear of the opponent: thedegree of fear was predicted to increase withincreasing RHP level; (2) direct and delayed physical

aggression [the second occurred when both RHPand provocation were high: Archer and Benson,2007]: these were included to assess whether theyfollowed physical aggression in being reduced at lowRHP levels; (3) items indicating how important itwas to be seen to get even, for example ‘‘think it isimportant for your reputation that you do some-thing to them’’ and ‘‘feel that you must be seen toget even’’: it was predicted that these would be lowat the low RHP level; (4) items indicating that theprovocation was unimportant, such as ‘‘think thatthey weren’t worth bothering with’’ and ‘‘laugh itoff’’: again, these were predicted to be low at the lowRHP level.

Method

Response to provocation questionnaire.There were three versions of this questionnaire,involving three levels of RHP manipulation: (1)smaller than you and has a reputation for beingall talk (RHP level 1); (2) about your size andcan usually stand up for himself (RHP level 2);and (3) bigger than you and has a reputation forbeing tough (RHP level 3). Because Study 1 showedsimilar findings for the two provoking situations,but a greater effect of RHP manipulation for thefirst scenario, it was this one that was incorporatedinto the response to provocation questionnaire inthe second study (‘‘You are at the servery whenanother prisonery starts to call you abusivenames’’). The RHP questionnaire used in Study 1involved only one response item at each level ofRHP. In Study 2, there were 16 alternatives, asoutlined in Introduction, involving the fear inducedby the opponent, physical and other forms of directaggression, and the importance of the provocation.Three versions of the questionnaire involving the

three RHP levels were used in a within-subjectsfactorial design. Respondents were asked to ratetheir likelihood of responding on each of 16 differentitems on a scale from 1 to 5 (15 not at alllikely; 25unlikely; 35possibly; 45 likely; 55 verylikely), for each of three descriptions of the potentialopponent.

Participants and procedure. A total of 260adult male prisoners (21 years of age and over) weresampled from three UK prisons. The overallresponse rate was difficult to calculate in view ofthe distribution method (see below). The sample wasdrawn from medium-high security settings. Themean age was 33.03 years (SD 23.0): 90% were ofwhite ethnic origin, 3% Afro-Caribbean, and 2%from the Asian subcontinent. The mean sentence

569Physical Aggression and Perceived Fighting Ability in Prisoners

Aggr. Behav. DOI 10.1002/ab

Page 8: Physical aggression as a function of perceived fighting ability among male and female prisoners

length was 31 months and the mean length of timeserved in penal institutions throughout their lives was44 months; 38% were serving for violent offences,28% for acquisitive offences, 19% for drug-relatedoffences, 12% for other indictable offences (e.g.,motoring offences), and 3% for sex offences.The questionnaires were distributed by prison

staff, and as a consequence it was difficult to find outhow many had been distributed. Participants com-pleted the questionnaires, the response to provoca-tion questionnaire (described above) and also theDIPC-SCALED [Ireland, 2004], which was used inanother analysis, on their own in their cells. It wasstressed on the front cover of the questionnaires thatnames and prison numbers were not required andthat only basic information was needed (e.g., age,sentence length, time served, offence type, andethnic origin). Participants were informed that theyhad a minimum of 1 hr to complete the question-naires, which were then to be placed in anunmarked, self-seal envelope, which was put into acollecting box.

Data analysis. There were 16 response itemsand three RHP levels. Initially, separate factoranalyses of the three different RHP levels werecomputed, with a view to constructing subscales toreduce the 16 items. However, there were twoproblems with this approach. The first was thatalthough the first of the three factors produced wasconsistent for most items, the second and third oneswere inconsistent across the three versions of thequestionnaire. Second, the main factor consisted ofitems that included three of the conceptual group-ings described above: direct aggression, delayedaggression, and the importance of the incident.Because it was important for testing the hypothesesthat these groupings were distinguished in theanalysis, separate analyses on each of the 16 itemswere used instead of subscales based on the factoranalyses. Correlations between them are reportedinitially. The items are then grouped togetherconceptually under the following headings: (1) fear(‘‘feel scared’’); (2) physical aggression and otherforms of direct confrontation; (3) delayed aggressiveresponses (e.g., ‘‘get even later’’); (4) the importanceof the provocation, for example being seen to geteven; (5) that the provocation was unimportant,such as ignoring the person concerned and laughingit off.

Results

Correlations. Correlations between the 16 re-sponse items for the combined scores across the

three scenarios indicated that the item that corre-sponded most with the single response used in Study1, ‘‘hit him there and then’’, showed high positivecorrelations with the following: (1) other directaggression items (r5 .73 for ‘‘kick them’’; r5 .60 for‘‘insult them back’’; r5 .59 for ‘‘shout at them’’;r5 .59 for ‘‘make fun of them and make them looksmall’’); (2) items indicating a delayed aggressiveresponse (r5 .74 for ‘‘hit them when no one islooking’’; r5 .68 for ‘‘get even later’’); and (3) withreputational items (r5 .62 for ‘‘think it importantfor your reputation that you do something to them’’and r5 .71 for ‘‘feel that you must be seen to‘get even’’’; but only r5 .26 with ‘‘feel that you’dbeen shown up in front of people’’). The first tworeputational items were strongly correlated (r5 .85)and to a lesser extent with the third one (r5 .42 and.40). The fear item, ‘‘feel scared’’, showed only lowcorrelations with other items, the highest being with‘‘feel that you’d been showed up in front of people’’(r5 .39).The correlation between the item ‘‘think that it is

important for your reputation that you do some-thing to them’’ and physical aggression was con-sistent at all three RHP levels (r5 .62, .56, and .54).Similarly, the lack of association between ‘‘feelscared’’ and physical aggression was consistent at allthree RHP levels (r5 .10, .03, and .06).The items indicating that the provocation was

unimportant were intercorrelated: for example‘‘ignore them’’ was correlated with ‘‘laugh it off’’(r5 .48) and ‘‘think that they weren’t worth bother-ing with’’ (r5 .57). ‘‘Tell them they’re out of orderbut leave it at that’’, which might be considereda form of direct confrontation, was more closelyrelated to ‘‘laugh it off’’ (r5 .50) and ‘‘think thatthey weren’t worth bothering with’’ (r5 .43), and istherefore included in the conceptual group indicat-ing the unimportance of the provocation. Thesecorrelations show that groupings based on factoranalyses would combine items from different con-ceptual groupings, and therefore would not havebeen appropriate for testing the hypotheses.

Differences between RHP conditions. Inthe following sub-sections, the results of repeated-measures ANOVAs for the three RHP conditionsare presented for the 16 items, organized in termsof the conceptual groupings outlined above andin Methods. The findings are summarized inTable II.

Fear. The single item ‘‘feel scared’’ showeddifferences between all three levels of RHP. Lowestscores occurred at low RHP, medium scores atmedium RHP, and higher levels at high RHP

570 Archer

Aggr. Behav. DOI 10.1002/ab

Page 9: Physical aggression as a function of perceived fighting ability among male and female prisoners

(Table II), as predicted. Thus, there were consider-ably lower levels of reported fear for a provocationfrom a low RHP individual than for an equallymatched person, and more fear when provocationarose from a more formidable person.

Direct aggression and confrontation. As shownin Table II, the item that corresponded with thedependent variable used in Study 1, ‘‘hit them thereand then’’, showed a lower value at the lower thanthe medium RHP level, in accordance with Study 1.There was also a decline from the medium to highRHP levels, so that both low and high levels weresignificantly different from the medium level. Twoother direct confrontational responses, ‘‘demand anapology’’ and ‘‘shout at them’’ showed a differentpattern, no difference between low and mediumlevels of RHP but a reduction at higher levels. Otheritems in this group showed no overall differences.There is, therefore, for three items, a decline frommedium to high levels of RHP, which would bepredicted by the original RHP model. Low RHPshowed lower values for physical aggression, whichwould be predicted by the hypotheses that aprovocation from such individuals was unimpor-tant. However, the two verbal items were notlowered for a low RHP opponent.

Delayed aggressive responses. Only one ofthe two delayed aggressive responses (‘‘hit themwhen no one is looking’’) showed a significantoverall difference (Table II). It was lower for lowRHP than for the other two levels, but there wasno decline at high RHP levels.

The need to be seen to ‘‘get even’’. For allthree of these items (shown in Table II), the lowerlevel of RHP showed lower values than the mediumlevel, and in two cases, medium and high levels ofRHP did not differ. In the third case (‘‘think it isimportant for your reputation that you do some-thing to them’’) there was a decline from the mediumto high level, indicating that reputation was mostimportant for opponents equally matched to therespondent, whereas being shown up or feeling thatone had to get even were similar for equally matchedand more formidable opponents.

Provocation unimportant. None of these fouritems (shown in Table II) differed between RHPlevels. Thus, ‘‘ignore them’’, ‘‘laugh it off’’, ‘‘thinkthat they weren’t worth bothering with’’, and ‘‘tellthem they’re out of order but leave it at that’’ allseemed to be a function of individual differences inresponse to provocation rather than how formidablethe opponent was.

Discussion

The item indicating fear (‘‘feel scared’’) was theonly one that showed a difference between all threeRHP levels, increasing from low to medium to high.This confirmed that as an opponent’s RHP increa-sed he was perceived as more frightening, and byimplication more formidable. Yet, there was not acomparable decline in direct aggression, measuredby ‘‘hit them there and then’’, as would be predictedby the original RHP hypothesis. Values for this item

TABLE II. Main Effects and Follow-up Tests for Study 2: F values for Main Effects of RHP

Item F Level 1 Level 2 Level 3 Contrasts

[1] Feel scared 19.80�� 1.78 2.05 2.16 1 vs. 2��; 1 vs. 3��; 2 vs. 3���

[2] Hit them there and then 5.10� 1.68 1.83 1.70 1 vs. 2�; 2 vs. 3�

Kick them. .96 1.46 1.51 1.50

Insult them back .10 2.18 2.21 2.19

Shout at them 3.95��� 1.99 2.05 1.87 1 vs. 3���; 2 vs. 3�

Make fun of them and make them look small .07 1.96 1.97 1.95

Demand an apology 7.00�� 1.87 1.85 1.68 1 vs. 3��; 2 vs. 3�

[3] Hit them when no one is looking 8.61�� 1.73 1.91 1.93 1 vs. 2��; 1 vs. 3��

Get even later 2.67 1.89 1.97 2.05

[4] Feel that you must be seen to ‘‘get even’’ 5.90� 1.72 1.87 1.88 1 vs. 2�; 1 vs. 3�

Feel that you’d been shown up in front of people 6.33� 2.46 2.68 2.72 1 vs. 2�; 1 vs. 3��

Think it is important for your reputation that you do something

to them

3.76��� 1.84 1.98 1.86 1 vs. 2���; 2 vs. 3���

[5] Ignore them .41 3.05 3.01 3.08

Laugh it off .24 2.76 2.71 2.74

Think that they weren’t worth bothering with 2.57 3.08 2.89 2.92

Tell them they’re out of order but leave it at that 1.84 2.56 2.55 2.44

�Po.01; ��Po.001; ���Po.05.Means and contrasts (LSD), for the three levels (low, medium and high). The numbers in bold indicate the conceptual groupings: [1] Fear; [2]Direct aggression or confrontation; [3] Delayed retaliation; [4] The need to be seen to ‘‘get even’’; and [5] Provocation unimportant.

571Physical Aggression and Perceived Fighting Ability in Prisoners

Aggr. Behav. DOI 10.1002/ab

Page 10: Physical aggression as a function of perceived fighting ability among male and female prisoners

showed no relation to those for ‘‘feel scared’’. Therewas a decline from medium to high RHP, as wasfound under some circumstances in Study 1. Thefinding in Study 1 that a low RHP individual wasless likely to attract a violent response than someonewho was more equally matched was replicated.One of the two delayed aggression responses, ‘‘hitthem when no one is looking’’, which was highlycorrelated with ‘‘hit them there and then’’, alsoshowed lower values at the low RHP level, but nodecline from medium to high levels. Two directverbally aggressive items, ‘‘demand an apology’’ and‘‘shout at them’’, showed a decline from medium tohigh RHP. In these cases, low and medium levelswere similar, indicating that the lower scores forphysical aggression at the low RHP level was notaccompanied by lower levels of verbal aggression.There is, therefore, evidence from two items for

a reluctance to aggress physically to a low RHPindividual who has acted in an insulting way.For the more direct of these items, there is alsoa reluctance to aggress to a high RHP individualin the same situation. For two verbally aggressiveitems, there was no lowered aggression to a lowRHP individual but there was evidence of a reluc-tance to aggress toward a high RHP protagonist.All three items indicating the need to be seen to

‘‘get even’’, ‘‘feel that you must be seen to ‘get even’’’,‘‘feel that you’d been shown up in front of people’’,and ‘‘think it is important for your reputation thatyou do something to them’’ were lower at the lowRHP level. This is consistent with the hypothesis thatthe lower levels of physical aggression to a low RHPopponent occurred because an insult from such asource is regarded as being of lesser importance. Thehigh correlations between ‘‘hit them there and then’’and two of these items, is consistent with thisinterpretation. The first two of these items showedno difference between medium and high RHP levels,but the third one showed a decline at the high RHPlevel. Thus, although there was a feeling that theparticipant had been shown up and should get evenfollowing an insult from a high RHP individual, theirreputation did not depend on it as much as it didwhen the opponent was more evenly matched. Thefour items indicating that the provocation wasunimportant, such as ‘‘ignore them’’ and ‘‘laugh itoff’’ showed no differences across the RHP condi-tions, suggesting that they were more related toindividual differences in reactions to provocationthan to characteristics of the opponents.Taken together, these findings suggest that the

pattern of lower physically aggressive responses toboth low and to high RHP opponents than to equally

matched ones arise from two influences: first, that aninsult from a low RHP individual is generally judgedto have fewer reputational consequences if it is notavenged by violent action, but it may neverthelessattract a verbally aggressive response. Second, whenthe insult arises from a high RHP source, there isan inhibition of both a violent response and directverbally aggressive responses. Although ‘‘feelingscared’’ was generally higher as RHP increased, itwas not correlated with physical aggression, at theany RHP level. In contrast, reputational importancewas highly and positively correlated with physicalaggression at all three levels: thus those who doregard an insult from a low RHP individual asimportant for reputation will be the ones most likelyto physically aggress, even though both importancefor reputation and physical aggression are generallyless when the opponent is low in RHP.These findings suggest that the mechanism under-

lying a response to a provocation involves the degreeto which the provocation matters, and this depends,in the prison context, not only on the nature of theprovocation itself but also the credibility (in this caseRHP level) of the provoking agent. Although the levelof fear increased in relation to the level ofthe opponent’s RHP, fear was found to be unrelatedto direct aggression measures within individuals. Inmost existing motivational theories of aggression,there is a separation of the instigation to respondfrom the decision process governing the choice of theparticular response. This is apparent in two-processmodels of animal aggression [Archer, 1976, 1988], andin Berkowitz’s [1989] neoassociationist model ofhuman aggression, which involves first the generationof negative affect, and second the choice of directaggression or an alternative response. The presentfindings, however, suggest that the both the degree ofprovocation and the likelihood of aggressing, areinfluenced by the properties of the opponent and enterinto a single cost–benefit calculation. This conclusionis consistent with the findings of Sell [2005] that upperbody strength is associated with both a history ofphysical aggression and higher self-ratings of anger ina sample of young men. Because anger forms part ofthe initial stage of two-process theories of aggression,from this perspective we would not expect it to berelated to RHP. However, the present findings alsosuggest that the opponent’s RHP does combine withthe nature of the provocation to directly influence theoutcome through a single process.The predictions derived from RHP models are

similar to those that would arise from ‘‘rationalchoice’’ theories of aggressive actions, such as Tedeschiand Felson’s [1994] social interactionist theory of

572 Archer

Aggr. Behav. DOI 10.1002/ab

Page 11: Physical aggression as a function of perceived fighting ability among male and female prisoners

coercive power. However, the concept of RHP isderived from mathematically rigorous models ofevolutionary change, rather than the intuitive notionof coercive power, and it can therefore integratestudies of human aggression with a large database onindividual differences in aggressiveness in animals,where size and strength are fundamentally important.However, neither RHP by itself nor rational choicetheories can readily encompass the present findings oflower than expected physical aggression to a lowRHP opponent. The lesser importance attached to aninsult from such an opponent introduces an addi-tional variable considered by neither the originalRHP game theory model of Parker [1974] nor socialinteractionist theories of coercive power. From anevolutionary game theory perspective, in such cases,the payoff of a physically aggressive response wouldbe low (because the person’s social status was notsubstantially threatened by the insult) compared withthe likely costs (of a response by prison officers).Issues that might be of interest in future studies are:(1) whether the lower than expected reaction to aninsult from a low RHP opponent forms part of acoherent set of attitudes about only responding tochallenges from opponents that ‘‘matter’’ in terms ofdominance relations; and (2) whether there arenegative attitudes toward people who retaliatephysically to those clearly weaker than themselves.

ACKNOWLEDGMENTS

I thank Christina L. Power for collecting the datafor both studies, and Aaron Sell for access to hisunpublished work. Study 1 was financed by a grantfrom the British Economic and Social Researchcouncil to John Archer and Jane L. Ireland.

REFERENCES

Archer J. 1976. The organisation of aggression and fear in verte-

brates. In: Bateson PPG, Klopfer P, (eds): ‘‘Perspectives in

Ethology 2.’’ New York: Plenum, pp 231–298.

Archer J. 1988. ‘‘The Behavioural Biology of Aggression.’’ Cam-

bridge, UK: Cambridge University Press.

Archer J. 1994. Violence between men. In: Archer J, (ed): ‘‘Male

Violence.’’ London: Routledge, pp 121–140.

Archer J. 2004. Sex differences in aggression in real-world settings:

A meta-analytic review. Rev Gen Psychol 8:291–322.

Archer J, Benson DA. (2007). Physical aggression as a function of

perceived fighting ability and provocation: An experimental

investigation. Aggress Behav.

Archer J, Huntingford FA. 1994. Game theory models and

escalation of animal fights. In: Potegal M, Knutson JF, (eds):

‘‘The Dynamics of Aggression: Biological and Social Processes.’’

London & Hillsdale, NJ: Lawrence Erlbaum, pp 3–31.

Archer J, Thanzami VL. 2007. The relation between physical

aggression, size and strength, among a sample of young Indian

men. Pers Individ Diff 43:627–633.

Archer J, Ireland JL, Power CL. 2007. Differences between bullies

and victims, and men and women, on aggression-related variables

among prisoners. Br J Soc Psychol 46:299–322.

Benson DA, Archer J. 2002. An ethnographic study of sources of

conflict between young men in the context of the night out.

Psychol Evol Gend 4:3–30.

Berkowitz L. 1989. Frustration-aggression hypothesis: Examination

and reformulation. Psychol Bull 106:59–73.

Courtwright DT. 1996. ‘‘Violent Land: Single Men and Social

Disorder From the Frontier to the Inner City.’’ Cambridge, MA:

Harvard University Press.

Daly M, WilsonM. 1988. ‘‘Homicide.’’ New York: Aldine de Gruyter.

Daly M, Wilson M. 1990. Killing the competition: Female/female

and male/male homicide. Hum Nat 1:81–107.

Eagly A, Steffen VJ. 1986. Gender and aggressive behavior: A meta-

analytic review of the social psychological literature. Psychol Bull

100:309–330.

Felson RB. 1996. Big people hit little people: Sex differences in

physical power and interpersonal violence. Criminology 34:433–452.

Felson RB. 2002. ‘‘Violence and gender reexamined.’’ Washington,

DC: American Psychological Association.

Hyde JS. 1986. Gender differences in aggression. In: Hyde JS,

Linn MC, (eds). ‘‘The psychology of gender: Advances through

meta-analysis.’’ Baltimore, MD: Johns Hopkins University Press,

pp 51–66.

Ireland JL. 1999. Direct and Indirect Prisoner Behaviour Checklist

(DIPC). Department of Psychology, University of Central

Lancashire.

Ireland JL. 2002. ‘‘Bullying Among Prisoners: Evidence, Research

and Intervention Strategies.’’ London: Brunner-Routledge.

Ireland JL. 2004. Direct and Indirect Prisoner Behaviour Checklist-

Scaled version (DIPC-SCALED). Department of Psychology,

University of Central Lancashire.

Ireland JL, Archer J. 1996. Descriptive analysis of bullying in male and

female adult prisoners. J Commun Appl Soc Psychol 6:35–47.

Ireland JL, Archer J. 2004. Association between measures of

aggression and bullying among juvenile and young offenders.

Aggress Behav 30:29–42.

Ireland JL, Archer J, Power CL. 2007. The characteristics of male

and female prisoners involved in bullying behavior. Aggress

Behav 33:220–229.

Nisbett RE, Cohen D. 1996. ‘‘Culture of Honor: The Psychology of

Violence in the South.’’ Boulder, CO: Westview Press.

Parker GA. 1974. Assessment strategy and the evolution of fighting

behavior. J Theor Biol 47:223–243.

Parker GA, Rubenstein DI. 1981. Role assessment, reserve strategy,

and acquisition of information in asymmetric animal conflicts.

Anim Behav 29:221–240.

Pellegrini AD, Roseth C, Mliner S, Bohn CM, van Ryzin M, Vance

N, Cheatham C, Tarullo A. 2007. Social dominance in preschool

classrooms. J Comp Psychol 121:54–64.

Sell A. 2005. Regulating Welfare Tradeoff Ratios: Three Tests of an

Evolutionary-Computational Model of Human Anger. Unpub-

lished doctoral dissertation. Santa Barbara: University of California.

Sell A, Tooby J, Cosmides L. 2005. Physical strength predicts

entitlement, anger, and attitudes about aggression in men. Paper

presented at the Human Behavior and Evolution Society

Conference, Austin, Texas.

Tedeschi JT, Felson RB. 1994. ‘‘Violence, Aggression, and Coercive

Actions.’’ Washington, DC: American Psychological Association.

Tremblay RE, Schaal B, Boulerice B, Arseneault L, Soussignan RG,

Paquette D, Laurent D. 1998. Testosterone, physical aggression,

dominance, and physical development in early adolescence. Int J

Behav Dev 22:753–757.

573Physical Aggression and Perceived Fighting Ability in Prisoners

Aggr. Behav. DOI 10.1002/ab