PDF

Sex Ro les, Vo l. 40, Nos. 7/8, 1999
How Masculin e Ou ght I Be? Men’s Masculinity
and A ggre ssion 1
Max We isb uch, D an ie l B e al, an d E dgar C. O’Neal 2
Tulane Uni versity
Male undergrad uates com pleted the B em Sex Role In ventory (B SRI) as they
are (actu al), as others though t they should be (ough t), as they thou gh t they
should be id eally (id eal), an d then rated the im portan ce of each item . D iscrepan cy scores w ere d eri ved by subtractin g actu al from either ough t (ough t
discrepan cy) or from id eal (id eal discrepan cy) an d w eigh tin g scores by the
im portan ce of each item . B SRI m ascu lin e item s pro vid ed the basis for m ascu lin ity discrepan cies, an d ® ller item s, for gen eral discrepan cies. With only tw o
or three exceptio ns, particip an ts w ere Cau casian . E ach m an com peted again st
a bogu s com petitor on a com puter versio n of the Taylo r reactio n-tim e aggressio n parad igm that yield ed a m easu re of both o vert (in ten sity of th e noise
blast putati vely deli vered to the opponent) an d co vert (noise blast duratio n)
aggressio n. Men w ith high m ascu lin e ``ough t’ ’ discrepan cies engaged in m ore
co vertÐ an d not m ore o vertÐ aggressio n than did lo w s, an effect not m oderated by pro vocatio n le vel. Those w ith high m ascu lin ity scores w ere m ore
o vertly aggressi ve than w ere lo w m ascu lin ity m en.
It has bee n ofte n stated that whe reve r the re is a proble m with control of
aggre ssion it is usually in the male . There is no doubt that multiple linkage s
exist at horm onal, physiologica l, de ve lopme ntal, and pe rsonality leve ls betwe e n degre e of ``male ne ss,’ ’ or masculinity, and aggre ssion (e .g., Kogut,
Langle y & O ’ Neal, 1992; E agly & Ste ffen, 1986; Brain & Sussm an, 1997) .
Without de nying the pote ntial importanc e of such links, this study addre sse s
a le ss direct, but pote ntially important conne ction betwe en the two. That
is, it focuse s on negative discrepancie s betwe e n how masculine me n think
1
The authors are grateful to Janet B . Rusche r and to an anonymous revie wer for valuable
sugge stions on an e arlier version of this article.
2
To whom corre spondence should be addre sse d at Departme nt of Psychology, 2007 Stern
Hall, Tulane Unive rsity, New O rle ans, LA 70118-5698; e -mail: edgar.one [email protected].
583
0360-0025/99/0400-0583$ 16.00/0 Ó
1999 Ple num Publishing Corporation
584
Weisb uch et al.
the y are and how masculine the y fe e l the y ought to be, and how such
discrepancie s might in¯ uence aggre ssion.
The the ore tical pre dictions re st on se lf discre pancy the ory (Higgins,
1987) . A ccording to the the ory, when one ’ s vie w of se lf is lowe r than an
ide al of self, negative e motions re sult. Spe ci® cally, the the ory posits that
discrepancie s be twe en actual self and what one thinks one could be le ads
to e motions like de pre ssion, whe re as whe n the discrepancy is be twe en
actual self and what one thinks othe rs e xpe ct him to be more agitate d
negative e motions, like anxie ty, are the outcome . It is quite possible that
negative emotions such as those resulting from self discre pancie s would
increase an individual’ s prope nsity toward aggre ssion. A ccording to B e rkowitz’ (1998) cognitive neoassociation ist theory, negative emotions can
increase responsive ness to situational factors ante cedent to aggre ssion. The
the ory also assum es that this te nde ncy would apply more strongly to e motions like anxie ty than to de pre ssive e motions be cause of lowe r activation
involve d in the latte r. This study addre sse s the possible aggre ssive conse que nce s re sulting whe n one ’ s vie w of self falls short of what one would
wish se lf to be (ne gative discrepancy) .
The re are two re asons howeve r, that might lead one to e xpe ct that
such actual/ought masculine discre pancie s are more strongly re late d to
aggre ssivity than would be the case with a more gene ralize d discrepancy
or with an actual/ide al masculine discre pancy. First of all, according to selfdiscrepancy the ory (Higgins, 1987) the emotions e licite d by actual/ought
discrepancie s are more like ly than actual/ide al discre pancie s to be ``agitation’ ’ re late d e motions. That is, the arousal compone nt of the affe ctive
reaction would be more like ly to make possible e xcitation transfe r (Z illmann, 1979) . A ccording to excitation transfe r theory arousal from any
source has the pote ntial through misattribution proce sse s to e nhance the
expe rie nce of ange r and pote ntiate aggre ssive be havior (Z illmann, 1979,
p. 337) . Such a process would make an individual with a large masculine
actual/ought discre pancy more re active whe n in an aggre ssive e xchange
than would be the case with an individual with a discre pancy associate d
with lowe r amounts of e motional arousal. This mechanism could well have
the e ffect of making a so-affe cted individual respond very early in an
aggre ssive e xchange , before provocation has re ache d the leve l to which
persons not pre viously arouse d would re spond.
Se condly, discre pancie s speci® c to masculinity should be particularly
strong in the ir e ffects on aggre ssion be cause aggre ssion is associate d with
the masculine ge nde r role (e .g., E agly & Steffe n, 1986; Grossman & Wood,
1992) , and be having more aggre ssive ly could be one way to re duce the
discrepancy (B abl, 1979; Moshe r & Sirkin, 1984) . This would be a much
more like ly outcom e whe n the masculine discre pancy is important for the
individual involve d (Wood, Christe nson, He bl, & Rothge rbe r, 1997) .
Ho w Mascu line Ought I B e?
585
In inve stigating the re lationship of masculine ought discre pancie s to
aggre ssion, aggre ssion both in a more ove rt and a more covert form was
of inte re st. A s aggre ssion be come s more cove rt it see ks to disguise the
ide ntity and/or harmful inte ntions of the aggre ssor (B joÈ rkqvist, O ste rman &
Lage rspetz, 1994) . More cove rt ( some time s refe rre d to as more ``indire ct’ ’ )
aggre ssion is like ly to be engage d in whe n an individual is fearful of retaliation (B e al O ’ Neal, O ng, & Ruscher, 1998) . It is re cognize d that the cove rtove rt dime nsion is a continuum , but for e ase of e xpre ssion in this study,
when two aggre ssion measure s are compare d, the more ove rt will be calle d
``ove rt’ ’ aggre ssion, and the more covert measure , ``cove rt.’ ’
Pre diction of the re lationship be twe e n ``ought’ ’ masculine discre pancie s and aggre ssion is more straight-forw ard for covert than for ove rt aggre ssion. It was pre dicte d that me n high in ``ought’ ’ masculine discrepancy
would expre ss more aggre ssion cove rtly than would men lowe r in ``ought’ ’
masculine discrepancy because such a discre pancy carrie s with it vulne rability and fe ar of re taliation. Cove rt aggre ssion would be less like ly to be
dete cted by one ’ s adve rsary.
It could be argue d that men high in ``ought’ ’ masculine discrepancy
might be more aggre ssive be cause this discrepancy involve s a conce rn for
othe rs’ e xpe ctations and ove rt aggre ssion would involve display of masculine -re levant be havior in a way of which othe rs would be come aware .
Howe ve r, this inclination would be more than counte rbalance d by the ir
prefe re nce for avoiding the possibility of sustaining retaliative attacks by
the ir partne r. For the se individuals ove rt aggre ssion would provide an ope n
invitation for one ’ s adve rsarie s to get e ve n, and the ir discrepancy would
make them fee l particularly disadvantage d in such an e xchange . So it was
predicte d that highe r ``ought’ ’ masculine discre pancy male s would be less
ove rtly aggre ssive than would the othe r men.
ME THOD
O verview an d D esign
The participants were 86 male unde rgraduate s e nrolle d in be ginning
Psychology or Sociology courses who participate d for course credit. A ll of
the me n exce pt two or thre e were Caucasian. E xpe rime nte rs were male ;
one a graduate stude nt, and four, advance d unde rgraduate s. Participants
® rst comple ted four ve rsions of the B em Sex-Role Inve ntory (short form)
which provide d the basis for the discre pancy score s, and then participate d
in a compute r ve rsion of the Taylor re action-tim e contest aggre ssion paradigm in which e ach was e xpose d to thre e le vels (low, mode rate and high)
586
Weisb uch et al.
of provocation . A fte r data were colle cted, the men were divide d by median
split into high and low discre pancy conditions. The basic design, the re fore
was a 2 ( low or high discre pancy) 3 3 (low, mode rate , and high provocation)
with the latte r factor be ing within subje ct.
D iscrepan cy Measu rement
A fte r re porting to the laboratory participants were informe d by an
expe rime nter that the y would participate in two unre late d studie s, the ® rst
to test some que stionnaire s to be use d in late r e xpe riments, and the se cond
a compute r reaction-tim e e xpe rime nt. The two-e xpe rime nt ploy was inte nde d to obscure the relationship betwe e n the masculinity discre pancy
measure ment and the aggre ssion measure ment proce dure .
The B e m Se x-Role Inve ntory (B SRI-S; B e m, 1981) was use d to measure discrepancy. The B SRI-S contains thirty personality attribute s Ð te n
more socially desirable for males than fe males (masculine ite ms), ten more
desirable for fe male s than male s (fe minine items), and ten e qually de sirable
for males and for fe male s (``® lle r’ ’ ite ms). Responde nts rate e ach ite m on
a seve n-point scale in te rms of to what de gree the ite m applie s to the m
(B em, 1981) . Response s to the masculine ite ms were use d as a basis for
the masculine discre pancy score s, and those to the ``® ller’ ’ ite ms were
used for ge ne ral discre pancy score s. Since the primary inte re st was in the
masculine dime nsion, the B e m was use d in lie u of othe r, multidim ensional
measure s of gende r role (e .g. A shmore, De l B oca, & B ilde r, 1995) .
The ® rst time e ach participant comple te d the B SRI-S he did so in a
way that described himse lf (the usual application of the B SRI-S) . producing
a actual se lf score . The ne xt two ways he comple te d it were in alte rnating
orde r. The ought me asure was obtaine d by having the participant comple te
the B SRI-S as othe r people think he should be. The participant comple ted
the Inve ntory as he would like to be ide ally, producing the basis of ide al
measure . (The orde r in which the ``ought’ ’ adm inistration and the ``ideal’ ’
administration of the BSRI-S was found to have no e ffect.) Finally, the
participant rate d (again, on the 7 point scale ) how important e ach item
was for him, providing the basis of e ach ite m’ s importance score (Wood,
et al., 1997) .
Pro vocatio n an d B eha vio ral A ggressio n
A second e xpe rime nter, who re maine d blind to discrepancy score s,
introduce d the participant to the compute r-base d re action time provocation
Ho w Mascu line Ought I B e?
587
and aggre ssion measure ment proce dure . A Hype rcard application running
on a Macintosh Pe rforma 6116 compute r was use d to colle ct data on be havioral aggre ssion; the application was adapte d from one devise d by B ushman
(B ushman & B aume ister, in press) .3 The proce dure was base d on what has
come to be calle d the Taylor (1967) reaction-tim e me asure of aggre ssion.
The program measure s noise inte nsitie s and durations set by participants
to be re ceive d as punishme nt by a bogus oppone nt on those trials in a
reaction time contest which the oppone nt lost. The oppone nt was supposedly in anothe r room , and the participant’ s ``losse s’ ’ were programm e d to
occur on half of the trials, in randomize d orde r. The noise inte nsity that
the compe titor set for e ach trial was displaye d on the scree n on e ach trial,
win or loss, as was the inte nsity se t by the participant for the oppone nt.
The noise setting oste nsibly set on each trial by the ``oppone nt’ ’ constitute d the within-subje ct manipulation of provocation across 24 trials. Possible se ttings varie d from 0 (no noise ) to 10, represe nting the highe st inte nsity.
In the ® rst block of e ight trials (low provocation) the noise inte nsity se ttings
putative ly se t for the participant to re ceive should the participant ``lose ’ ’
on the next trial randomly varie d be twe e n 1 and 4. For those four trials
on which a noise blast was actually de live re d to the participant, the de cibe l
inte nsity correspondingly varie d from 60 to 75, with an ave rage of 67.5.
For the ne xt block of trials (mode rate provocation ) the se ttings ave rage d
5.5, and the decibe l, 82.5. A nd on the ® nal block of trial (high provocation ),
the se ttings ave rage d 8.5, with de cibe l ave raging 97.5.
The inte nsity measure was take n to be the more ove rt, in that the
inte nsity value s were displaye d on the scree n and the participant was led
to belie ve that they appe ared on the partne r’ s scree n as well. The measure
was the re fore continuousl y salie nt to the participant and thought to be
monitore d by the oppone nt as well. Conve rsely the duration me asure was
conside red the more covert, since its value s were not include d on the display
scree n, and it was le ss salie nt and easily monitore d.
In initial instruction s the participant was inform ed how he could vary
the duration of the noise to be re ceived by the oppone nt. In addition, the
participant was told that the oppone nt’ s compute r did not have the ability
to regulate duration be cause of a te mporary compute r malfunction . The
noise e xpe rie nce d by the participant was of a standard 1 se c. duration.
The re fore , the participant was aware of the duration re sponse dime nsion
in his own re sponse options, but acknowle dgme nt of variation in the opponent’ s duration was not possible , minim izing the like lihood of re ciprocity
(cf. McDanie l, O ’ Neal & Fox, 1971) . The se feature s of the duration measure
we re inte nde d to bring it in line with the construct as described by B joÈ rkqvist
(B joÈ rkqvist, O ste rman, & Lage rspetz 1994) .
3
We thank Dr. B rad B ushman for providing a copy of the software adapte d for this study.
588
Weisb uch et al.
R E SU LTS
O f the 86 me n comple ting the proce dure , e le ve n we re e xclude d from
analyse s; se ve n, for being suspicious about the ``oppone nt,’ ’ two, be cause
the y did not unde rstand ope ration of the mouse in ge ne rating the duration
measure , and two, be cause the y did not comply with instructions. O f the
remaining 75 me n, 23 were e xclude d from analyse s of masculine ought
discrepancie s be cause the y had positive discre pancy (actual highe r than
ought) score s.4 For analyse s using ide al masculine discre pancie s, e ight were
exclude d for the same reason, and positive discre pancie s exclude d thre e
participants from analyse s of ge neral discrepancie s.
D iscrepan cies
The inte rnal consiste ncy of the actual ve rsion of the B SRI-S
(alpha 5 .84) was acce ptable and its factor structure conform e d to the oretical
expe ctations re garding the inde pe nde nce of the masculine scale from the
othe r scales.5 The fe mininity ite ms were not use d in the analyse s.
Discre pancy score s were calculate d by subtracting the ``actual’ ’ score
from the ``ide al’ ’ or ``ought’ ’ score for each ite m and multiplying the re mainder by the importance score accorde d that ite m by each participant. The se
products were ave rage d within category to produce for e ach participant a
masculine /ought discre pancy score, and score s for masculine /ideal discre pancy, ge ne ral/ought discre pancy, and ge ne ral/ide al discre pancy. The re sulting score s the re fore re ¯ e ct discre pancie s which take into account the
single importance rating give n to e ach B SRI trait by each subje ct. For each
analysis, men were assigne d to a discrepancy condition base d on whe ther
the y were above ( high discre pancy) or below (low discrepancy) the median
on the discre pancy involve d.
A ggressio n
O f primary inte re st was the re lationship be twe e n the masculine
``ought’ ’ discre pancy and aggre ssion. A 2 (low or high masculine discre pan4
While positive se lf-discrepancie s are potentially a very intere sting topic, their de terminants
and conseque nce s are tangential to the pre sent study.
5
A factor analysis with varimax rotation was performe d on the actual form of the B SRI-S.
Thre e factors emerged that we re comme nsurate with the masculinity, fe mininity, and ® ller
subscales. In order to e xplore the structure of the two main constructs of the scale e ve n
further, a factor analysis with oblique rotation was pe rformed for the masculine and fe minine
items only; masculine and fe minine subscale s we re uncorre lated (r 5 2 .0039) .
Ho w Mascu line Ought I B e?
589
cie s) 3 3 (low, mode rate , or high provocation) A NOV A on noise duration
(cove rt aggre ssion) scores re vealed two main e ffe cts. High masculine ought
discrepancy men were more ( F(1,50) 5 4.57, p , .05, d 5 .53) covertly
aggre ssive than we re their low discre pancy counte rparts (se e Table I). A s
was true in the othe r analyse s, in this analysis the re were the expe cted
main effe cts ( F(1.66,83.13) 5 22.16, p , .05) for provocation le ve l with less
covert aggre ssion from subje cts at lowe r provocation le vels than at highe r
le vels. There was no inte raction be twee n O ught Discrepancy and Provocation Le vel, indicating that high discrepancy men’ s gre ate r aggre ssion was
not mode rate d by provocation .6 A similar analysis of ``ought’ ’ masculine
discrepancy on noise inte nsity (ove rt aggre ssion) yie lde d only the expe cted
main e ffe cts for Provocation , F(1.42,70.99) 5 48.14, p , .05.
The re were no e ffects of an analysis of masculine actual/ide al discrepancy on e ithe r noise inte nsity or duration e xce pt for the usual main e ffe cts
in the expe cted direction for Provocation . In te rms of ge neral discrepancie s
tappe d by the ® lle r items, no re sults othe r than main e ffe cts of Provocation
we re obtaine d for noise duration or noise inte nsity.
O ne additional se t of analyse s were unde rtake n, due to the large
amount of the oretical and e mpirical work linking masculinity and aggre ssion. These focuse d on the re lationship be twe e n simple (not discrepancy)
Tab le I. Means and Standard De viations for Cove rt and Overt A ggression by Le ve l of
Masculine Discre pancy and Provocation
Low Masculine
Discre pancy
Mean
High Masculine
Discre pancy
SD
Me an
SD
Cove rt aggression
Low
provocation
Moderate
provocation
High
provocation
671.38
393.97
1063.30
898.72
1032.13
860.73
1546.67
1304.54
1207.10
867.01
1922.51
1408.72
Overt aggre ssion
Low
provocation
Moderate
provocation
High
provocation
6
4.41
1.55
4.81
1.67
5.71
1.59
5.54
1.27
6.89
1.75
7.03
1.53
Give n that there was no interaction with provocation, it is possible to e xamine masculine
ought discre pancie s as a continuous variable using re gre ssion. This analysis prove d to be
similar to the main e ffect reporte d in the A NO V A , F (1,50) 5 3.81, p 5 .057.
590
Weisb uch et al.
BSRI-S masculinity score and aggre ssion. In 2 (high and low masculinity)
3 3 (low, mode rate and high provocation) A NOV A s high masculine men
delive re d highe r, F(1,73) 5 4.45, p , .05, d 5 .38, noise inte nsitie s than
did low masculine men (se e Table II). B e side the usual main effe ct for
provocation, the re we re no othe r e ffe cts of Masculinity on e ithe r ove rt or
covert aggre ssion.7
D ISCU SSION
The re were two important outcome s of this study, the greate r covert
aggre ssive ne ss of the men high in important masculine actual/ought discrepancie s, and the gre ate r ove rt aggre ssion in the men high in masculinity . It
is inte re sting that in ne ithe r case was the effe ct mode rated by degre e of
provocation, and that the effe cts were not obtaine d for masculine ide al
discrepancy or gene ral discre pancy.
O f the se two outcome s, the more straight-forward is the ® nding in
regard to ge nde r role masculinity. The strong re lationship be twee n
BSRI-S masculinity score s and ove rt aggre ssion is in line with the ore tical
and e mpirical linkage s that have be en made at a numbe r of le ve ls, including
those e nume rated at the outse t of this article . Not only is aggre ssion part
of the masculine stere otype in our society, in ge nde r sche ma theory it is
Tab le II. Me ans and Standard Deviations for Covert and O ve rt A ggression by Le vel of
Masculinity and Provocation
Low Masculinity
Mean
High Masculinity
SD
Me an
SD
Covert aggre ssion
Low
provocation
Moderate
provocation
High
provocation
907.49
804.98
958.84
867.88
1215.72
934.54
1433.54
1356.01
1627.76
1088.15
1647.31
1380.51
O ve rt aggression
Low
provocation
Moderate
provocation
High
provocation
4.80
1.56
5.54
2.19
5.58
1.45
6.63
1.66
6.87
1.77
7.21
1.71
A re gre ssion analysis again reve ale d the same main effe ct of Masculinity, F(1,73) 5
p , .05, that is reporte d in the A NO V A .
7
5.21,
Ho w Mascu line Ought I B e?
591
fe ature d promine ntly as part of the male role, and it does in fact constitute
an ite m on the masculine sex role scale of the B SRI (B em, 1981) . That
the se masculinity scores are re late d to ove rt aggre ssion lends support to
the construct validity of the e xpe rimental proce dure .
The fact that the actual/ought masculine discre pancie s produce d e ffects
on aggre ssion and that ne ithe r the ge ne ral discre pancie s nor the actual/
ide al masculine discre pancy yielde d signi® cant e ffects offe rs discriminative
validity support for the rationale provide d in the Introductio n. A ccording
to discre pancy the ory, actual/ought discrepancie s produce ``agitate d’ ’ emotions (Higgins, 1987) , and toge the r with the masculine conte nt of the discrepancy, make s particularly like ly e xcitation transfe r to aggre ssion (cf.,
Taylor, O ’ Neal, Langle y & B utche r, 1991) . It also was e xpe cted that it
would be important discre pancie s that produce the se re sults (Wood, e t
al., 1997) .8
The re sults show clearly that the me n with high masculinity ``ought’ ’
discrepancie s e xpre ssed the ir aggre ssion, but in a more cove rt form. It
see ms plausible that such a discre pancy carrie s with it a vulne rability and
accompanying fe ar of re taliation. A fter all, the se men fee l some thing lacking
in masculinity and in this paradigm the adve rsary’ s pote ntial for retaliation
is extre mely salie nt. Fear of re taliation may well cause a greate r re liance
on cove rt means of aggre ssion (B e al e t al., 1998) . If so, it is ironic that
inse curitie s about how masculine one ought to be re sults in male s adopting a
patte rn of aggre ssion more common with fe male s in our society (B joÈ rkqvist,
1994) . For male s ge nde r role s pe rtain not only to magnitude of aggre ssion
but manne r of aggre ssion.
R E FE R E NCES
Ashmore, R. D., Del B oca, F. K., & Bilder, S. M. (1995) . Construction and validation of the
Ge nde r A ttitude Inventory, a structured inventory to asse ss multiple dimensions of ge nder
attitudes. Sex Ro les, 32, 753± 785.
Babl, J. D. (1979) . Compensatory masculine re sponding as a function of se x role. Jo urn al of
Con sulting an d Clinical Psycholo gy, 47, 252± 257.
Be al, D. J., O ’ Neal, E . C. O ng, J., & Rusche r, J. B . (1998) . The w ays an d m ean s of interracial
aggressio n: Mod ern racists’ use o f co vert retaliatio n. Unpublished manuscript.
Be m, S. L. (1981) . B em sex-role in ven tory. Palo A lto, CA: Consulting Psychologists Press.
Be rkowitz, L. (1998). A ffective aggre ssion: The role of stre ss, pain, and negative affect. In
R. G. Ge en & E . Donnerste in (E ds.), Hum an aggressio n: Th eo ries, research , and imp lications fo r social p olicy. San Diego, CA: A cade mic Pre ss.
BjoÈrkqvist, K. (1994) . Se x difference s in physical, verbal, and indirect aggre ssion: A review
of re cent rese arch. Sex Roles, 30, 177 ± 188.
BjoÈrkqvist, K., Oste rman, K., & Lagerspe tz, K. M. (1994) . Se x difference s in cove rt aggression
among adults. A ggressi ve Beh avior, 20, 27± 33.
8
As was e xpecte d, analyse s of the discre pancie s unweighte d for importance yie lded no signi® cant effe cts.
592
Weisb uch et al.
Brain, P. F., & Susman, E . J. (1997) . Hormonal aspe cts of aggression and violence . In D. M.
Stoff, J. B reiling, & J. D. Maser (Eds.), Han db ook of antiso cial b eh avior. Ne w York: Wiley.
Bushman, B . J., & Baume ister, R. F. (in pre ss). Thre atened e gotism, narcissism, se lf-este e m,
and direct and displaced aggre ssion: Does self-love or se lf-hate lead to violence? Jo urnal
of Perso nality and So cial Psycho logy.
Eagly, A. H., & Steffen, V . J. (1986) . Ge nder and aggre ssive be havior: A me ta-analytic re view
of the social psychological literature. Psycho lo gical B ulletin, 100, 309 ± 330.
Grossman, M., & Wood, W . (1992) . Se x difference s in e motional intensity: A social role
explanation. Journ al of Perso nality and So cial Psycho logy, 65, 1010 ± 1022.
Higgins, E . T. (1987) . Self-discrepancy: A theory relating se lf and affe ct. Psycho logical Re view ,
94, 319± 340.
Kogut, D., Langley, T., & O ’ Neal, E . (1992). Gende r role masculinity and angry aggression
in women. Sex Ro les, 26, 355± 368.
McDanie l, J., O ’ Neal, E ., & Fox, E . (1971) . Magnitude of re taliation as a function of the
similarity of available re sponse s to those e mploye d by attacke r. Psych on om ic Scien ce,
22, 215± 217.
Mosher, D. L., & Sirkin, M. (1984) . Me asuring a macho personality conste llation. Jo urnal of
Research in Perso nality, 18, 150 ± 163.
OÈ ste rman, K., B joÈ rkqvist, K., & Lagerspe tz, K. M. J., with Kaukiaine n, A., Landau, S. F.,
Fracze k, A ., & Caprara, G. V . (1998) . Cross-cultural e vide nce of fe male indirect aggression. A ggressi ve B eha vio r, 24, 1± 8.
Taylor, S. L., O’ Ne al, E ., Langley, T., & B utcher, A . (1991) . A nge r arousal, deindividuation,
and aggre ssion. A ggressi ve B eh a vior, 17, 193± 206.
Taylor, S. P. (1967) . Aggre ssive behavior and physiological arousal as a function of provocation
and the tende ncy to inhibit aggre ssion. Jou rnal o f Perso nality, 35, 297 ± 309.
Wood, W., Christense n, P. N., Hebl, M. R., & Rothgerbe r, H. (1997) . Conformity to se xtype d norms, affe ct, and the self-conce pt. Jou rnal o f Perso nality and So cial Psycho logy,
73, 423± 535.
Zillmann, D. (1979) . Hostility and aggressio n. Hillsdale, NJ: E rlbaum.