ArticlePDF Available

Sexually aggressive men's perception of women's communication: Testing three explanations

Abstract

The authors tested three explanations of findings that sexually aggressive men perceive women's communications differently than less aggressive men. The first suggests that aggressors are incompetent in decoding women's negative emotions. The second posits that they fail to make subtle distinctions between women's friendliness and seductiveness and between assertiveness and hostility. The third explanation contends that sexual aggressors use a suspicious schema and therefore discount the veridicality of women's communications. These explanations were tested using videotaped scenarios in which a woman's responses to a man's advances were systematically varied. The data were most supportive of the suspicious schema explanation. These findings are integrated with other research on the characteristics of sexual aggressors and on the perceptions of aggressive children and of maritally violent men. Implications for interventions are also discussed.
Journal of Personality and Social Psychology
1994,
Vol. 67, No. 4,699-712
Copyright 1994 by the American Psychological Association,
Inc.
Sexually Aggressive Men's Perceptions of Women's Communications:
Testing Three Explanations
Neil
M. Malamuth and Lisa M. Brown
The authors tested three explanations
of
findings
that sexually aggressive men perceive women's
communications differently than
less aggressive
men. The
first
suggests
that aggressors are incompe-
tent in decoding women's negative
emotions.
The second
posits
that they fail to make subtle distinc-
tions between women's friendliness and seductiveness and between assertiveness and hostility. The
third explanation contends that sexual aggressors use
a
suspicious schema and therefore discount
the veridicality of women's communications. These explanations were tested using videotaped sce-
narios in which a woman's responses
to
a man's advances
were
systematically varied. The data were
most supportive
of
the suspicious schema explanation. These findings
are
integrated with other
research on the characteristics of sexual aggressors and on the perceptions of aggressive children and
of maritally violent men. Implications for interventions are also discussed.
The present study
is
part of research designed
to
identify the
characteristics
of
men
in the
general population
who
sexually
aggress against women.
We
have hypothesized
and
provided
supportive data showing that several factors distinguish between
more sexually aggressive men and their less aggressive counter-
parts.
Examples of such factors include attitudes supporting ag-
gression against women, hostile emotions,
and
relatively high
sexual arousal
to
aggression (e.g., Malamuth, 1986). These
fac-
tors may
be
described
as
part
of
an interrelated network (Ber-
kowitz, 1993) of converging factors that lead some men
to
com-
mit sexual aggression (e.g., Malamuth, Heavey,
&
Linz, 1993).
In this study, we explored men's perceptions
of
women's com-
munications. Perceptions are the first element
in
the interactive
process
and are
likely
to
play
a
primary role
in
activating
and
shaping other aspects of such
a
network (Fiske & Taylor, 1993).
We anticipated that such research will add to our understanding
of the interrelated factors contributing
to
the development
and
the maintenance of sexually aggressive characteristics.
Neil M. Malamuth
and
Lisa M. Brown (now
at
the Department
of
Psychology, University
of
Florida), Department
of
Communication,
University of Michigan, Ann Arbor.
We thank Bill Murphy
for
permission
to
use the stimuli he and
his
associates developed
and
Daniel Linz
and
Lucia
O'Sullivan
for
very
helpful comments
on
earlier versions of this
article.
The preparation of
this article
was
facilitated by National Institute of Mental Health Grant
MH31618
to
Neil M. Malamuth. An earlier version of this article was
rated
by
the Interpersonal Communication Division
as the
"Top Paper"
presented at the 1994 Annual Meetings of the International Communi-
cation Association, Sydney, Australia.
Correspondence concerning this article should be addressed
to
Neil
M.
Malamuth,
who is now
at Communication
Studies,
334
Kinsey Hall,
University of California,
Los
Angeles,
California 90024-1538.
Previous Studies Linking Social Perceptions and Sexual
Aggression
Although earlier work (Check & Malamuth, 1983, 1985;
Malamuth & Check, 1980, 1985; Malamuth, Haber, & Fesh-
bach, 1980) consistently found that sexually aggressive charac-
teristics were associated with more positive perceptions of a
rape victim's experience, more recent work examined whether
perceptual differences also extend to stimuli depicting noncoer-
cive interactions. On the basis of their clinical experience with
rapists, Murphy, Coleman, and Haynes (1986) hypothesized
that more sexually aggressive men may have "deficits in their
ability to separate seductive from friendly behavior or hostile
from assertive behavior" (p. 260). To test this hypothesis, they
measured men's perceptions of a woman in videotaped interac-
tions responding to a man's advances in one of four
ways:
with
hostile rejection, assertive rejection, seduction, or friendliness.
The investigators computed two accuracy indices: The
seduc-
tion
discrimination
index
was
computed by subtracting ratings
of the woman's seductiveness in scenarios in which she behaved
seductively from those in which she behaved in a friendly man-
ner. The hostility
discrimination
index was computed by sub-
tracting ratings of the woman's hostility in scenarios in which
she behaved in an assertively rejecting manner from those in
which she behaved in a hostile rejecting manner. Poorer dis-
crimination
(i.e.,
smaller differences in perceptions) on both the
hostility and seduction discrimination indices were associated
with more rape-supportive attitudes. The hostility discrimina-
tion index, but not the seduction index, also correlated with
sexually aggressive behavior. A regression equation designed to
predict sexual aggression that included personality, sexual
arousal, and attitudinal variables
as
well
as the
hostility discrim-
ination index showed that additional variance
was
explained by
the inclusion of
this
index. However, the total amount of vari-
699
700NEIL M. MALAMUTH AND LISA M. BROWN
ance explained by this equation was considerably less than that
found in some other research (e.g., Malamuth, 1986), which
leaves open the possibility that the added variance predicted by
the perceptions index may not be replicated when a relatively
large amount of the variance has been explained by other
variables.
Although
we
designed the present study primarily in light of
Murphy et al.'s (1986) study and we used their methodology,
two other studies using different methods' than that of Murphy
et al. are directly relevant to the present research. Upton,
McDonel, and McFall (1987) administered in federal prison a
measure of heterosocial cue reading to groups of rapists, violent
nonrapists, and nonviolent nonrapists. Participants responded
to
a
measure consisting of a
series
of videotaped
vignettes,
some
depicting a couple on a first date and others depicting more in-
timate couples. Rapists were less accurate than the other two
groups in reading women's cues in first date interactions. No
differences were found, however, in the intimate situation. Al-
though the differences for the first date interactions were re-
ported as an overall effect, the authors also reported that step-
wise
multiple regression analysis indicated that the negative-cue
items accounted for the greatest portion of unique variance
(where rapists were more likely to perceive the negative re-
sponses
as
relatively positive reactions).
McDonel and McFall (1991) found similar
results in a
follow-
up study using male college students. In this study, they in-
cluded only the first date interactions from the perceptions in-
strument used by Upton et al. (1987). They also used a second
measure that consisted of three dating scenarios wherein par-
ticipants
were
asked to indicate the extent to which the man was
justified in continuing his advances in the face of
the
woman's
negative responses. As the authors noted, this latter measure
appears to be both a perception measure and a decision task
that involves evaluative components (i.e., what a man should
do).
The investigators reported that on both measures college
men who were
less
accurate in decoding women's negative cues
were higher in rape-supportive attitudes and scored higher on a
self-reported measure of likelihood of committing rape. Accu-
racy in decoding women's positive cues and men's cues in gen-
eral,
which were assessed only with the
first
instrument, was not
related to these outcomes.
Explanations for Misperceptions
The three studies discussed above have made important con-
tributions
by
showing that the associations between sexually ag-
gressive characteristics and perceptions of
women
are not lim-
ited to coercive interactions and by developing methodologies
to assess such perceptions. We designed the present study to
advance this area of research further by systematically testing
the predictions of different explanations of the
findings
and by
better integrating this work with the larger body of research on
the characteristics of sexually aggressive men.
In examining this perceptions research, we present three ex-
planations (overperception, negativeness blindness, and suspi-
cious schema) that can provide promising directions for the de-
velopment of theoretical models. We explicate these explana-
tions and test their competing predictions. These explanations
may be
differentiated on the basis of the conditions under which
perception differences are expected to be associated with men's
sexually aggressive characteristics.
Overperception of Hostility or Seductiveness
Nature of
explanation.
The first explanation is based on
Murphy et al.'s (1986) findings that sexual aggressors appear
to be less able to make subtle perceptual distinctions between
friendliness and seductiveness and between assertiveness and
hostility.
Although Murphy et
al.
did not state
so
explicitly, they
implied that this
may be
due to aggressive men's overperception
of sexuality and of hostility. The overperception of sexual intent
is an extension of research showing that men generally perceive
more sexual intent than do women (e.g., Abbey, 1982). As ap-
plied here, it suggests that some men are more inclined than
others to overperceive such intent. Similarly, Abbey and her as-
sociates (Harnish, Abbey, & DeBono, 1990) also suggested this
hypothesis when they argued that "males may misperceive fe-
males'
friendliness as seduction, an error which can create un-
pleasant interactions and sometimes lead to date rape" (p.
1341).2
Conditions under which differences are
predicted.
With re-
spect to the type of scenarios used in this line of research, the
overperception explanation predicts differences between ag-
gressive
and nonaggressive men only in those scenarios in which
the woman responds with relatively mild or benign friendliness
or assertiveness. Therefore, when a woman responds in a
friendly manner that is not sexual or seductive, aggressors are
expected to overperceive sexual cues and thereby label her be-
havior as relatively seductive. Similarly, when she responds in a
rejecting but relatively nonhostile (i.e., assertive) manner, ag-
gressors are expected to overperceive hostility. However, when a
woman responds in a highly hostile or highly seductive manner,
there is no opportunity to read more into the behavior and
therefore no differences are predicted between aggressive and
nonaggressive men.
Negativeness Blindness
Nature of
explanation.
McDonel and McFall (1991) sug-
gested that there is a specific cue-reading impairment in sexu-
ally aggressive men's ability to decode a woman's negative cues.
This results in failing to recognize her negative reactions, which
1 Although Murphy et al. (1986) and the present research used ratings
scales, the two studies by McFall and associates used a test of reading
affective cues measure that required participants to "guess which of five
affective cues—romantic, positive, neutral, negative or bad mood—was
being conveyed by each party in each interaction " (Lipton et al., 1987,
p.
18). It is therefore not feasible to make direct comparisons between
the studies using the different methodologies.
2 Unfortunately, the use of difference scores by Murphy et al. (1986)
clouds interpretation of the findings. These discrimination scores would
have yielded the same correlations even if a psychological process oper-
ating in the opposite way had been operating. More sexually aggressive
men may have perceived less, or underperceived, hostility in the hostile
scenarios and underperceived seductiveness in the seductive scenarios
rather than overperceiving in the assertive and the friendly scenarios.
Difference scores do not permit us to distinguish between the two
possibilities.
PERCEPTIONS OF WOMEN701
could result in persistence in making advances
(e.g.,
sexual ag-
gression). These investigators emphasized that this deficit is
limited to accuracy in decoding negative cues and stated that
"misinterpretation of
positive
cues as negative should not logi-
cally be a contributory factor leading to sexual assault"
(p.
20).
Conditions under which differences are
predicted.
The neg-
ativeness blindness explanation predicts differences associated
with men's aggressivity only when the woman communicates
negative cues. Therefore, when the woman rejects the man by
communicating responses such as hostility, nonaggressors are
expected to read these cues clearly but aggressors are hypothe-
sized to fail to appropriately detect and interpret these cues.
Suspicious Schema
Nature of explanation. A third explanation emphasizes the
role of cognitive schemas3 (Neisser, 1976) as applied to sexually
aggressive men's perceptions (Malamuth, 1983;McFall, 1990).
This approach suggests that perceptual differences are due to
different decision rules
or
judgmental heuristics underlying so-
cial information processing.
Recent research on sexually aggressive men (e.g., Malamuth,
Sockloskie, Koss,
&
Tanaka,
1991;
Malamuth et
al.,
1993)4 has
suggested that the schema likely to guide their perceptions of
women
is
a suspicion schema: Women's communications about
romantic or sexual interest cannot be trusted as veridical (i.e.,
Women don't tell the truth when it comes to
sex).
This schema
is hypothesized to form expectancies that underlie perceptions
of women generally
(e.g.,
greater typicality of hostile behaviors)
and to be applied to initial perceptions of individual women
(e.g., suspecting the veridicality of
her
sexual communication).
In the terminology used by Jones
(1990),
these constitute
cate-
gory-based (women generally) and target-based (particular
woman) expectancies.
Conditions under which differences are
predicted.
A wom-
an's attempt to communicate (in sexually related areas) by her
overt behavior would
be
processed suspiciously by more
aggres-
sive men and as veridical by less aggressive men. Therefore, ag-
gressors would perceive such communication as having the op-
posite meaning of that intended. There are
some
conditions that
might be expected to increase differences between the percep-
tions of aggressive and nonaggressive men, such
as when
women
communicate intense
reactions.
This prediction may
be
derived
from the principle of distinctiveness (Kelley, 1971) whereby
unique events are perceived as more informative. When the
woman's communications are intense
(e.g.,
she
is
highly hostile
or highly seductive), less aggressive men are apt to believe that
it is easy to clearly read her communication and to accept it as
honestly describing how she feels. More sexually aggressive
men, in contrast, are prone in such circumstances to discount
the veridicality of women's communications and to question
why her reactions are so strong (e.g., "She protests too much"
or "Why is she being
so
nice?").
woman rejects a man's advances, in a high hostile fashion or a
low hostile
(i.e.,
assertive) manner, and two in which the woman
reacts to the man's advances favorably, either seductively or in a
friendly manner (i.e., nonseductive). Because we used some of
the stimuli developed by Murphy et al. (1986), we first at-
tempted to replicate their
findings
using procedures and analy-
ses similar to those they used. Such replication not only estab-
lishes the reliability of the phenomenon but provides a firmer
basis for suggesting that any differing conclusions that we may
reach using other types of analyses would apply to the data they
had gathered. We then tested the ability of the overperception,
negativeness blindness, and suspicious schema explanations to
account for participants' interpretations of
the
four scenarios.
We
also attempted to directly test the explanation that sexually
aggressive characteristics are associated with a hostile, suspi-
cious schema affecting category-based and target-based expec-
tancies by asking participants questions pertaining to their be-
liefs about typicality and
honesty.
Next,
we
attempted to extend
Murphy et al.'s (1986)
finding
that the addition of information
about social perceptions enables better prediction of sexual ag-
gression
(i.e.,
in a regression equation). Because
we
used a set of
variables that has been shown to be particularly successful in
such prediction, this is a rather stringent test of the potential
utility of such perceptions.
Explanations' Predictions for Four Situations
The differences in the predictions made for the four scenarios
(used in the present research) by the overperception, negative-
ness blindness, and suspicious schema explanations are shown
in Table 1. Below, we summarize these predictions for each of
these explanations. These predictions are made for a measure
that assesses perceptions of how negative are the woman's reac-
tions to a man's advances. Such an overall index of negativity
was
indeed computed in this
study,
as
described later.
Overperception
explanation.
To
reiterate the basis for mak-
ing predictions for this explanation, it should be recalled that
the overperception explanation predicts differences only in
those scenarios in which sexual aggressors are likely to overper-
ceive the woman's benign friendly or assertive reactions as re-
vealing more seductiveness or hostility than she intends. How-
The Present Study
Overview
In the present article we report on data collected from men
who responded to four relevant situations: Two in which a
3 Fiske and Taylor (1984) defined a schema as "a cognitive structure
that contains knowledge about the attributes of a concept and the rela-
tionships among those attributes" (p. 8). Schemas "form a relatively
cohesive and persistent body of knowledge capable of guiding subse-
quent perception and appraisals" (Segal, 1988, p. 147). Schemas have
been shown to influence varied aspects of information processing in-
cluding the encoding of new information (Markus & Zajonc, 1985).
They are often activated automatically (Bargh, 1989) and frequent
thought about a domain is likely to lead to automatized schemas (e.g.,
Andersen, Spielman,
&
Bargh,
1992;
Hayes-Roth, 1977).
4
This
work indicates that most of the factors identified
in
earlier work
as predictors of
sexual
aggression (e.g., Malamuth, 1986) are part of
a
constellation of variables that may be organized within a statistical path
labeled hostile
masculinity.
Men high on such characteristics show a
desire to be in control, to be dominating, particularly in relations with
women, coupled with an insecure, defensive, and suspicious or adver-
sarial orientation to them.
702NEIL M. MALAMUTH AND LISA M. BROWN
Table 1
Three
Explanations'
Predictions
of
Correlations Between
Men's Sexually
Aggressive Characteristics
and Their
Perceptions
of
Women's
Negativity
Across the Four Scenarios
Explanation
Overperception
Negativeness blindness
Suspicious schema
Hostile
no correlation
inverse correlation
strong inverse correlation
Scenario
Assertive
positive correlation
weak inverse correlation
weak inverse correlation
Seductive
no correlation
no correlation
strong positive correlation
Friendly
inverse correlation
no correlation
weak positive correlation
ever, when she communicates a high level of seductiveness or
hostility, overperception is not likely to occur.
In Scenario
1
(see Table 1) the woman is communicating a
high level of hostility. Therefore, overperception (of hostility or
negativeness) would not have the opportunity to occur. Sim-
ilarly, in Scenario 3, the woman is communicating a high level
of seductiveness; overperception (of seductiveness or positivity)
would not occur here either. Consequently, in these two scenar-
ios no correlation
is
predicted by this explanation.
In contrast, in Scenario
2,
when the woman reacts in a reject-
ing but low hostile manner (i.e., assertively), more sexually ag-
gressive men are predicted to overperceive hostility
(i.e.,
to per-
ceive
her reactions
as
more
negative
than they actually
are).
This
should result in a positive correlation between men's aggressive
characteristics and perceptions of her
negativeness.
Similarly, in
Scenario 4, in which the woman reacts in a friendly but nonse-
ductive manner, the overperception explanation suggests that
more aggressive men will overperceive positive reactions, par-
ticularly seductiveness. Consequently, the prediction here is of
an inverse correlation between aggressive characteristics and
perceptions of
the
woman's overall negativeness (i.e., they will
perceive more positive reactions).
Negativeness
blindness.
The negativeness blindness expla-
nation (which suggests that aggressive men fail to detect wom-
en's negative cues) predicts that when the woman communi-
cates high hostility (Scenario 1), the more aggressive men will
fail to perceive her negative reactions accurately (i.e., they will
perceive less negativity). Therefore this explanation predicts an
inverse correlation between men's aggressive characteristics
and their perceptions of the woman's negative or hostile reac-
tions for this scenario. In Scenario 2, she is assertively rejecting
the man. Here, too, it may be that some weak inverse correla-
tion would be expected, although the opportunity to miss nega-
tive cues is far less. In neither of the other conditions is the
woman communicating negative
cues,
and therefore this expla-
nation would not predict any correlations between perceptions
of her reactions and men's aggressive characteristics.
Suspicious
schema.
This explanation predicts that
more
ag-
gressive men will discount the veridicality of the messages com-
municated by the woman. Differences between relatively high-
and low-aggressive men are particularly likely to emerge when
the woman communicates intense, ostensibly clear messages.
For the high hostility scenario (Scenario 1), therefore, a rela-
tively strong inverse correlation is predicted between men's ag-
gressive characteristics and their perceptions that the woman
is being truly negative or hostile (i.e., high-aggressive men will
perceive less negativeness on the part of the woman). A similar
but weaker relationship
is
predicted when the woman rejects the
man in a manner relatively
low
in hostility (Scenario
2),
because
the discounting process is less likely to show as strong a differ-
ence between aggressors and nonaggressors when the woman
reacts in a less distinct way (see earlier discussion of the princi-
ple of distinctiveness). Furthermore, when the woman commu-
nicates a high level of seductiveness, it is predicted that more
aggressive men will again perceive her
as less
truthful. They will
therefore perceive her
as
actually less responsive (or more nega-
tive)
to the man's
advances.
Consequently,
a
positive correlation
is predicted here between men's aggressive characteristics and
their perceptions of the woman's negativeness.
A
similar corre-
lation is predicted when the woman reacts in a friendly, low-
seductive manner (Scenario 4) but again the correlation is ex-
pected to be weaker in the context of this
less
distinct reaction.
Perceptions
of
man's
behavior.
Finally, we examined per-
ceptions of the man's behavior, which was not systematically
manipulated across the scenarios but was supposed to remain
constant. The schema explanation, in the context of other re-
search on the characteristics of sexually aggressive men, sug-
gests predictions regarding how the man would be perceived.
Research indicates that not only do schemas affect social per-
ceptions but experiencing emotions that
may be
associated with
such schemas may independently affect such perceptions (Kel-
tner, Ellsworth,
&
Edwards,
1993).
The
hostile masculinity
con-
struct suggests that men with more sexually aggressive charac-
teristics have emotions
(e.g.,
anger toward women) and schemas
(e.g., male-female relations are adversarial) that would led
them to perceive the man in the scenarios as hostile toward the
woman. Neither of the other two explanations discussed in this
article appear to predict any associations between sexual ag-
gression and perceptions of
the
man, although the overpercep-
tion one might predict higher perceptions of both hostility and
seductiveness for the man
as
well as for the woman.
Method
Overview
of Design
The research
was
conducted over a
2-year
period.
All
of the measures
used in the
first
year
were
also used in the second
year,
but an additional
instrument—the self-report measure of
sexual
aggression—was added
in the second
year.
Data are therefore available for this measure for par-
ticipants from the second year only. When the same measures
were
used
in both years, the data were combined. The present sample that in-
cluded the measure of sexual aggression and social perceptions came
from participants in a study reported in an earlier article (Malamuth,
1986).
As
described
below,
this provided the opportunity to anatyzeXhe
PERCEPTIONS OF WOMEN703
contribution of social perceptions to predicting sexual aggression in the
context of other factors shown to relate quite strongly to such behavior.
We conducted the research in three phases. During the first session,
all of
the
questionnaire measures (except for the social perceptions as-
sessment) were administered. These included various paper and pencil
measures assessing attitudes and sexual aggression (in the second year
only) as well as the
other questionnaire measures used for the regression
analyses reported
below.
In the second phase (conducted several days or
weeks
later),
we
measured sexual arousal using self-reports and physio-
logical measures (see Malamuth, 1986).
The third phase was composed of
two
assessments. First, we mea-
sured laboratory aggression
as
described by Malamuth (1988). Second,
in the
final
stage of the research,
we
moved participants to another loca-
tion and assessed social perceptions. This final assessment, which was
presented to participants as an attempt to obtain perception ratings to
be used in future independent research, is the primary focus of this
article.
Participants
One hundred and seventy-four men participated
in the
study.
Missing
data on the attitudes or the perception ratings permitted analyses on
161 participants. For 90 men, data were also available from the
self-
report sexual aggression measure. As noted below, analyses using the
penile tumescence index (of
sexual
arousal to rape vs. consenting sex)
have somewhat reduced sample sizes because not all of the men chose
to participate in that phase of the research. However, systematic com-
parisons of volunteers with nonvolunteers in similar research using the
penile tumescence measure have revealed no differences in various
measures pertaining to sexual aggression, although volunteers scored
higher on a measure of general sensation seeking (Malamuth
&
Check,
1983).
We recruited participants from several sources, including university
courses, a sign-up list displayed in the psychology department, a city
summer employment center, and newspaper
ads.
Only applicants over
the age of
18
participated in the study. They were paid $7 per hour for
their participation.
Participants signed up for a general subject
pool.
Experimenters then
selected participants from this general list and invited them to partici-
pate in specific experiments. When contacted by the individual experi-
menters conducting each of the three phases (presented to participants
as
independent experiments), potential participants received general
de-
scriptions of
the
procedures and measures
used.
The experimenters em-
phasized in each phase that participants could leave at any time and
that there would be no penalty nor would any explanation be required.
Participants were paid on arrival at each study and were told that they
could keep the money regardless of whether they completed the experi-
ment. As an additional safeguard, a law professor
was
hired to serve as
ombudsman for the project. Participants received his name and phone
number on signing up for the experiment. They were told that this per-
son
was
completely independent of the staff conducting the research and
that they could direct any complaints to him. None were made. At the
end of
the
second and third phases, participants received relevant de-
briefings (Malamuth
&
Check, 1984).
Although participants' names were not used for identification, back-
ground information
(e.g.,
date of birth) enabled exact matching of
ques-
tionnaires across the separate phases of the study. Questions asked at
the end of
the
third phase verified that none of the participants recog-
nized any association between this study and the earlier phases of the
research.
Primary
Materials and Procedure
Assessment
of social perceptions.
We
obtained videotapes of four
of
the scenarios developed by Murphy et al. (1986). These portrayed 30-s
scenes simulating interactions in a bar between a woman and a man in
which the man attempts
to
initiate a
social
interaction with
the
woman.
The woman responds to the man's advances in one of four
ways.
These
scenarios
were
viewed in the following
order:
friendly, assertively reject-
ing, seductive, and high hostile. To facilitate presentation, we
first
dis-
cuss the
two
scenarios in which the woman rejects the man (with hostil-
ity and assertively) and then the two scenarios in which she responds
favorably (with high seductiveness and in a friendly manner).
After viewing each interaction, participants indicated their percep-
tions on several
9-point
scales,
ranging from -4
(strongly disagree)
to 4
(strongly
agree),
with 0
(neither agree
nor
disagree)
as the midpoint.
(For some of the analyses reported below, the scales were transformed
to range from
1
to 9.) Participants rated
the
woman and man separately,
indicating the extent to which they agreed in each interaction that the
person was friendly, assertive, rejecting, seductive, and hostile. These
items were embedded within a larger questionnaire containing other
adjectives designed
to
disguise
the
specific focus of the study's scenarios.
Participants also indicated the extent to which they thought that the
woman's behavior accurately reflected her feelings.
Attitudes
supporting
aggression.
We used four scales developed by
Burt (1980) to create a composite-attitudes score (see her article for
information regarding reliability and validity). These were the Rape
Myth Acceptance (19 items), the Acceptance of Interpersonal Violence
(6 items), the Adversarial Sexual Beliefs (9 items), and the Sex Role
Stereotyping
(9
items) scales. Burt has theorized that the scales she de-
veloped measure certain attitudes that are widely accepted in Western
culture but are particularly held
by
rapists and potential
rapists.
She
has
argued that such attitudes play an important role in contributing to
sexual aggression. Each man's score for each scale was converted to z
scores before being added together.
Sexually aggressive behavior. Sexually aggressive behavior
was
mea-
sured
by
the self-report instrument developed
by
Koss and Oros (1982).
It assesses a continuum of sexual aggression including psychological
pressure, physical coercion, attempted rape, and rape. Participants re-
spond by using a true-false format to
10
descriptions of different levels
of sexual coercion. An example of an item is "I have had sexual inter-
course with a woman when she didn't want to because I used some de-
gree of physical force (twisting her arm, holding her
down,
etc.)." Koss
and Oros (1982) and Koss, Gidycz, and Wisniewski (1987) presented
data regarding the reliability and validity of this scale.
Secondary Measures
We used
five
additional measures in the regression analyses. We de-
scribe these measures briefly below. A more detailed description is
found in Malamuth (1986).
Dominance
motive.
This
8-item
measure focused on power (i.e.,
dominance)
as
a motive in sexual relations (Nelson, 1979).
Hostility toward
women.
We used the Hostility Toward Women
(HTW) scale (30 items; Check, 1984; Check, Malamuth, Elias, & Bar-
ton, 1985). Data concerning its reliability and validity were presented
by Check (1984).
Psychoticism. We used the Psychoticism scale of
the
Eysenck Per-
sonality Questionnaire (Eysenck, 1978). It includes 20 items and is de-
signed to assess a dimension associated with general antisocial person-
ality characteristics.
Sexual
experience.
We used the Sexual Behavior Inventory (Bent-
ler, 1968) to assess sexual experience in conventional heterosexual acts.
It includes 21 items that inquire whether the subject has engaged in
various sexual behaviors including such acts as fondling breasts, inter-
course, and oral sex.
Sexual
arousal to
rape.
We
computed an index of sexual arousal to
rape (with arousal measured by a
gauge
directly assessing penile tumes-
cence) for each participant by dividing maximum arousal to a rape de-
704NEIL M. MALAMUTH AND LISA M. BROWN
Table 2
Mean Perceptions Ratings
of
Videotaped Interactions
Scenario
Hostile
Assertive
Seductive
Friendly
Friendly
-3.37"
-2.75"
3.15C
2.39'
Assertive
1.80"
1.86b
2.14"
-0.40"
Ratings of woman
Seductive
-3.35"
-3.18"
2.89C
-1.59"
Hostile
2.91"
1.37b
-3.40c
-3.01°
Rejecting
2.59b
2.34"
-2.98C
-2.02"
Note. Perceptions rated on scales ranging from -4
(strongly disagree)
to +4
(strongly
agree),
with 0
(nei-
ther agree nor disagree) as
the midpoint. For each characteristic within each column, means with common
superscript do not differ from each other at the .05 level.
piction by maximum arousal to a depiction portraying consenting sex
(Abel, Barlow, Blanchard,
&
Guild 1977). As noted earlier, this was as-
sessed in a separate session, with about two thirds of
the
participants
choosing to participate in this assessment.
Results
Manipulation Check
Although Murphy et
al.
(1986) provided data for the validity
of their scenarios by relying on expert raters, we also evaluated
the specific scenarios we selected from those of Murphy et al.
We assessed whether the intended behaviors were perceived as
intended by the sample as a whole. We conducted analyses of
variance on each of five ratings (friendly, assertive, seductive,
hostile, and rejecting). The analyses reported here used the data
obtained from all participants, although it might be contended
that only participants who do not show aggressive characteris-
tics accurately encode the woman's communications. We
judged it more appropriate to use the entire sample for several
reasons. First, it would be difficult to decide what criterion to
use for exclusion. If
we
excluded any participant who scored
above the lowest point on the composite measure of attitudes
supporting violence against women, then the vast majority of
participants would be excluded (98%), because most scored
above that lowest level. Using aggressive behavior only as the
criterion for exclusion seems inappropriate because
we
also ex-
pected attitudes to be related to perceptions. Second, compari-
sons between scenarios (in terms of validating the intended ma-
nipulations) can be meaningfully made because even highly ag-
gressive men are expected to recognize these differences.
Finally, inclusion of all participants would be expected to re-
duce support for the intended manipulations. Consequently, we
concluded that including all participants was the more conser-
vative approach and that, if such comparisons showed support,
considerable confidence could be placed in such assessment.
The expected effects
were
obtained on all of these variables
(p
< .001). Follow-up comparisons strongly supported the success
of the intended variations
in the
woman's behavior
(see Table
2).
The scenarios in which the woman rejects the man's advances
differed primarily on perceptions of hostility, although both
were perceived as showing some degree of hostility (1.37 vs.
2.91). Correspondingly, in the high hostility scenario, the
woman was perceived as somewhat less friendly than in the as-
sertive scenario
(-3.37
vs. -2.75). Interestingly, these two sce-
narios did not differ in perceptions of the woman's assertiveness
or rejection.
In the scenario in which the woman was instructed to re-
spond in a friendly or pleasant manner, she was in fact rated
as showing a high degree of friendliness and a low degree of
seductiveness. She was also perceived as low on the negative re-
actions of hostility and rejection. The seductive scenario was
also rated as intended: The clearest difference between the
friendly and seductive scenarios was on perceptions of seduc-
tiveness
(-1.59
vs.
2.89), but expected differences
were
also ob-
served on perceptions of friendliness, assertiveness, hostility,
and rejection. Overall, the scenarios
we
selected from Murphy
et al. (1986) seem to have been very successful in yielding the
intended manipulation differences.
Perceptions
of
Negativity
Index
We
computed a composite score that reflects the overall judg-
ment of the woman's responsivity. It combined the four ratings
with a clear valence: hostility, rejection, seductiveness, and
friendliness. The composite was created by adding these four
variables together (after each was converted to a z score), with
the friendliness and seductiveness ratings weighted negatively
and the hostility and rejection ratings weighted positively. We
did not include the assertiveness ratings in the composite score
because they did not have a clear valence rating in terms of the
degree of negativeness of the woman's
reactions.
We
labeled this
measure the
perceptions
of
negativity
index,
reflecting how dis-
couraging or negative rather than encouraging or positive the
woman was perceived to have reacted to the man's advances.
Higher
ratings
on this index reflected more negative perceptions
of the woman's responsiveness
(i.e.,
more rejection and hostility
and less seductiveness and friendliness). The alpha coefficients
for this index were .62 for the hostile scenario, .50 for the asser-
tive scenario, .60 for the seductive scenario, and .42 for the
friendly scenario. Except for the last of these, these coefficients
are certainly satisfactory for
a
four-item assessment. Further
ex-
amination of the friendly scenario revealed that all of the items
except ratings of the woman's seductiveness showed relatively
high intercorrelations. Eliminating the seductiveness item for
this scenario results in an alpha of
.61,
which
is
high for
a
three-
item assessment. This might be because relatively few men are
expected to rate the woman's friendly behavior as seductive in
this scenario. However, for the sake of consistency across see-
PERCEPTIONS OF WOMEN705
narios, we computed the composite in the same way across all
of the scenarios. The alpha coefficient for an analysis using all
of the four perceptions of negativity indices
was
.66.
Replicating Murphy et
al.
's
Findings
As indicated earlier, we considered it critical to first replicate
Murphy et al.'s (1986)
findings
so
that
we
may be confident that
we measured the same phenomenon as them before we test al-
ternative explanations for these findings.
Perceptions
of
negativity
index.
We
first
attempted to repli-
cate with the perceptions index the type of
analyses
conducted
by Murphy et
al.
(1986). As noted earlier, they found that rape-
supportive attitudes correlated significantly with both the hos-
tility discrimination and seduction discrimination indices,
whereas sexual aggression correlated only with the former mea-
sure.
We calculated for each participant the hostility discrimi-
nation index by subtracting the
scores
on the composite percep-
tions of negativity index for the assertive scenario from the
scores for
this same
index for
the
hostile
scenario.
We
calculated
the seduction discrimination index by subtracting the compos-
ite scores for the friendly scenario from the seductive scenario.
We then correlated scores on these indices with attitudes sup-
porting aggression.
Our results generally replicated Murphy et al.'s (1986). The
hostility discrimination index was correlated with attitudes
K161) =
-.26,
p
< .001.
An effect approaching significance was
found for the seduction discrimination index, r{\6\)
=
.\5,p<
.10.
The analyses also replicated Murphy et al.'s findings for
sexual aggression. The hostility discrimination index was sig-
nificantly related to sexual aggression, r(90) =
-.30,
p
< .01.
As
in Murphy et al.'s
findings,
the seduction discrimination index
was not related to sexual aggression, r(90) =
—.13,
ns.
Individual
ratings.
To
more precisely follow Murphy et al.'s
(1986) analyses, we subtracted (a) the individual woman's hos-
tility ratings (rather than using the composite of
the
four vari-
ables) for the assertive rejecting scenario from the woman's hos-
tility ratings of the hostile rejecting scenario and (b) the wom-
an's seductiveness
ratings
of the friendly scenario from woman's
seductiveness ratings of the seductive scenario. We found that
rape-supportive attitudes correlated as predicted with both the
hostility discrimination index, r(161) =
—.20,
p < .02, and se-
duction discrimination index, K161) =
—.32,
p < .001. How-
ever, neither index was correlated with sexual aggression, ri90)
= -.09, ns, K90) =
—.09, ns,
respectively.
Findings from the present study appear to mostly replicate
Murphy et al.'s (1986) results. The overall perceptions of nega-
tivity index and the individual item ratings correlated with atti-
tudes supportive of aggression against
women.
However, for
sex-
ual aggression, Murphy et
al.'s
results
were
replicated only when
the overall index was used.
Examining Alternative Explanations for Perceptual
Differences
The analyses reported immediately below focus on the four
scenarios: Woman rejects man's advances in a hostile fashion;
woman rejects man in an assertive manner; woman responds
seductively; or woman responds in a pleasant or friendly man-
Table 3
Correlations Between
Men's Sexually Aggressive
Characteristics
and
Their Perceptions
of Women
Across
the
Four Scenarios
Measure
Perceptions of
negativity index
Attitudes
(n =
161)
Aggression (n = 90)
Ratings of woman's
hostility
Attitudes (n= 161)
Aggression
(n =
90)
Ratings of woman's
seductiveness
Attitudes
(n
= 161)
Aggression (n = 90)
Hostile
-.34***
-.44***
-.21**
-.24**
-.31***
-.42***
Scenario
Assertive
-.13*
-.15
.04
-.12
-.33***
-.19*
Seductive
.27**
.18*
.35***
.29**
.29***
.14
Friendly
.14*
.32**
.24**
.30**
-.14*
.01
Note. To facilitate comparison with the explanations' predictions
listed in Table 1, the seductive item was reversed so that higher scores
indicate lower seductiveness. Data are reported using perceptions of
negativity index
as
well
as ratings of woman's hostility and seductiveness
(see
Table
1
for comparison with explanations' predictions).
*p<.10.
**p<.01.
***p<.001.
ner. We tested the ability of the overperception, negativeness
blindness, and suspicious schema explanations to account for
participants' responses to the four
scenarios.
As
shown in Table
3,
we report the
findings
for (a) the overall composite index as-
sessing perceptions of
the
woman's negativity and (b) the indi-
vidual woman's hostility and woman's seductiveness items.
These two items were focused on by Murphy et al. (1986) and
are the ones most pertinent to the three explanations described.
For the purposes of presentation in Table 3, the ratings for the
seductiveness item were reversed (i.e., higher ratings indicate
lower seductiveness) so that they are in the same direction as
the negativity composite and the hostility ratings. The analyses
focus on both attitudes and sexual aggression because these are
the two categories that showed relationships to social percep-
tions in the three studies described above.
High hostility
scenario.
As
indicated in Table
3,
the data for
this scenario were consistent with the suspicious schema and
negativeness blindness explanations. Strong associations were
found between characteristics associated with sexual aggression
(attitudes and behavior) and more positive perceptions of the
woman's responsiveness when she behaved in a highly hostile
way. This pattern was found for the overall perceptions index
and for the individual ratings of the woman's hostility and se-
ductiveness. The results for seductiveness are more easily ac-
counted for by the suspicious schema than by the negativeness
blindness explanations. The latter explanation predicts that
participants will fail to detect the woman's negative cues but
does not suggest that they will be more likely to perceive such
responses as seductive.
Assertive
scenario.
Using the composite negativity index,
the data showed
weak,
nonsignificant correlations
in a
direction
supportive of the suspicious schema and negativeness blindness
explanations (see Table 3). Analyses using the individual items
706NEIL M. MALAMUTH AND LISA M. BROWN
showed no effects with perceptions of the woman's hostility,
whereas both the men's attitudes and aggression revealed some
correlation with perceptions of her seductiveness
(i.e.,
more ag-
gressive characteristics associated with higher perceptions of
woman's seductiveness; see Table 3). Again, these data appear
more consistent with the suspicious schema than the negative-
ness blindness explanation. To reiterate, the latter explanation
suggests that it is the failure to detect negative responses that
differentiates aggressors from nonaggressors. Therefore, differ-
ences should occur primarily on items directly assessing nega-
tiveness (i.e., hostility), which were not found here. Although
supporters of this explanation could argue that perceptions of
positive responses could be indirectly affected by the failure to
detect negative cues, it is less clear that such a failure would
necessarily result in greater perceptions of cues such as seduc-
tiveness. On the other hand, the suspicious schema explanation
would suggest that aggressors may be more likely to perceive
women's "no" responses as seductive teasing. Consequently, it
follows more clearly from this explanation that an assertive re-
jection might be interpreted as somewhat seductive by those
with an adversarial orientation to women.
Seductive
scenario.
In the scenario in which the woman re-
sponds in a seductive way, the data were supportive only of the
suspicion explanation—that
is,
men with more sexually aggres-
sive characteristics perceived her in a relatively more negative,
hostile way. Using the negativity composite index, the results
reached statistical significance for attitudes and approached sig-
nificance for aggression. For the individual hostility item, they
were significant on both attitudes and aggression, but for seduc-
tiveness they were significant on attitudes only (see Table 3).
Friendly
scenario.
For the scenario in which the woman re-
sponds in a friendly manner, the data using the negativity com-
posite index were again consistent with the suspicious schema
explanation, reaching statistical significance with aggressive be-
havior and approaching significance with attitudes (see Table
3).
Examining the individual items, we also found that on per-
ceptions of the woman's hostility, rape-supportive attitudes and
aggression
were
correlated in the direction predicted by the sus-
picious schema explanation, although the strength of the rela-
tionship here actually appears to exceed that predicted by this
explanation. However, on ratings of seductiveness, more rape-
supportive attitudes yielded a weak relationship in the direction
predicted by the overperception explanation, although no cor-
relation was found with aggressive behavior (see Table 3).
Perceptions of Man
We tested the prediction of the suspicion explanation that
men with more sexually aggressive characteristics would also
perceive the man in the scenarios
as being
more
hostile.
In keep-
ing with the hostile masculinity construct, we predicted that
there would be a general tendency for men with more sexually
aggressive characteristics to project their own hostility to
women in these types of interactions onto the man in the sce-
narios and thereby to perceive the man as being more hostile
toward the woman.
We computed an overall hostility composite that combined
all of the ratings in the four scenarios (after first converting each
to
a
7
score)
of
perceptions
of the degree of the man's hostility.
As predicted, both attitudes supporting sexual aggression,
r(160) = 31,p<
.001,
and aggressive behavior, r{90)= .28,p<
.008,
were related to this composite.
Examination of the data for each scenario separately showed
the same pattern. In all of
the
scenarios, attitudes significantly
correlated with greater perceptions of the man's hostility, al-
though they appeared somewhat higher in the friendly (r = .27,
p
< .002) and seductive (r =
.33,
p
<
.001) scenarios than in the
assertive (r =
.21,
p < .01) and hostile (r =
.23,
p < .005) sce-
narios. Similarly, sexually aggressive behavior also correlated
significantly in the friendly (r = .22, p < .05) and seductive (r
= .26, p < .03) scenarios, but the correlations did not reach
significance in the assertive and hostile scenarios (both corre-
lations = .16, ns).
Directly Testing the Suspicious Schema Construct
In an attempt to directly test the explanation that sexually
aggressive characteristics are associated with a hostile, suspi-
cious schema affecting category-based and target-based expec-
tancies,
we
selected the scenario in which the woman responded
in a hostile manner. This type of scenario has generally shown
the strongest associations with men's sexually aggressive char-
acteristics, both in the present and in previous research. We
asked participants questions pertaining to their beliefs about
typicality and honesty, with the same
9-point
scales used to as-
sess perceptions. For typicality, we asked participants to indi-
cate for each scenario the degree to which the type of reaction
exhibited by the woman was the way most women would be-
have;
for honesty, we asked the degree to which the woman ex-
pressed her feelings honestly.
We
reasoned that schema suspiciousness would be a function
of
(a)
the belief that most women behave in such
a
highly hostile,
rejecting way and (b) the extent to which the particular woman
was perceived as being honest. A low-suspicion schema would
include the belief that such hostile behavior was atypical and
that the particular woman in the scenario was honest. The op-
posite pattern would reflect high suspicion: The more suspi-
cious schematic male would believe that hostile behavior was
typical and that this individual could not be trusted. We rea-
soned that a mixture of
the
particular and general belief would
reflect intermediate degrees of schema suspicion. However, per-
ceptions of typicality of women generally might take precedence
over
a
belief in this specific woman's honesty.
To operationalize these notions of typicality and specificity
we divided our sample at the median for the two items. Typi-
cality and honesty ratings that were above the median of the
distribution for all participants
were
judged to show high typi-
cality or honesty, whereas those at or below the median were
judged
as
being
low
on these
dimensions.
Therefore, the pattern
rated as showing the least suspicious schema, which was evi-
denced by 30% of the participants, consisted of a below-the-
median rating on the typicality of the woman's hostile rejection
coupled with an above-the-median rating of
the
belief that she
is being honest in communicating her feelings. The most suspi-
cious pattern was the opposite on both dimensions, which was
exhibited by
16%
of the participants. In the second least suspi-
cious pattern, exhibited by
40%
of the participants, the partici-
pant did not believe that this hostility is typical of women in
PERCEPTIONS OF WOMEN707
Table 4
Frequency
Analysis
of
Degree
of
Suspicious Schema
as
a
Function
of
Men's Aggressive Characteristics
Men's aggressive
characteristics
Low
High
1
23
1
Degree of schema suspicion
2
26
12
3
9
7
4
7
5
Note. x2(3) =
10.1,
p < .02. 1 = low suspicion, 4 = high suspicion.
general, but did not judge this particular woman's hostility to
be veridical. The opposite pattern (i.e., hostility is typical of
women, but believed this woman) was exhibited by
15%
of the
participants and
was
rated
as
being the third level of suspicion.
Using this four-level classification, we found that both atti-
tudes supporting sexual aggression, r(161) =
.23,
p <
.01,
and
aggressive behavior, r(90) = .29, p <
.01,
correlated in the ex-
pected direction, with high suspicion associated with higher at-
titudes and aggression. Because of potential differences of opin-
ion regarding the appropriateness of ranking the two interme-
diate levels in the manner described above, we also calculated
these relationships using a
3-level
classification with both the
intermediate levels being designated as a 2 on this suspicion
dimension. The results were virtually identical, with both atti-
tudes,
r(161)
=
.24,
p
<
.01,
and aggression,
r(90)
=
.28,
p
<
.01,
correlating significantly.
To illustrate the basis for these correlations, we conducted
an analysis that classified participants as having low- or high-
aggressive characteristics using both attitudes and behavior. If a
participant scored in the top 10% of
the
distribution on either
attitudes or
behavior,
he
was
classified as having high-aggressive
characteristics. We then conducted a frequency analysis using
this twofold classification (low
vs.
high) crossed with the degree
of schema suspicion classification described above. This analy-
sis yielded a significant relationship (see Table 4), indicating
that men with aggressive characteristics were more likely to use
schemas with higher suspiciousness.
Predicting Sexual Aggression
As noted earlier, Murphy et al. (1986) found that in a regres-
sion equation (that included several variables), social percep-
tions showed a relatively strong unique contribution to the pre-
diction of sexual aggression. As also indicated earlier, many of
the participants in the present study came from a pool used in
a study reported in an earlier article (Malamuth, 1986). In that
study, a combination of several variables, not including social
perceptions, had a relatively high success in cross-sectional pre-
diction of
sexual
aggression. In the analysis reported below, we
examined whether the addition of information about social per-
ceptions might enable
even
better statistical prediction of sexual
aggression or whether it would be redundant with some of the
other measures already included in that equation.
Malamuth
(1986)
performed analyses on
155
participants
us-
ing all of the variables except for the penile tumescence mea-
sure.
Of these, 95 had also agreed to participate in the research
phase in which the penile tumescence index (of
sexual
arousal
to rape
vs.
consenting
sex) was
assessed.
In the social perceptions phase reported here, 90 men (out of
155 possible) participated. Of these, 58 participants had penile
tumescence data. The analyses reported below focus on this
sample.
We
used two approaches in the analyses reported
below.
The
first used a stepwise approach. It allowed to enter the equation
(a) all of the variables found to be significant contributors in
Malamuth's
(1986)
study and
(b) social
perceptions.
The
second
approach, which was a more stringent test, first forced entered
all of the variables found significant in Malamuth's article and
only then entered the social perceptions variable.
To include social perceptions in the regression analyses, we
created a single index across the four scenarios based on the
discounting predictions of
the
suspicious schema explanation.
The perception of woman's negativity indices for the four sce-
narios were summed together. For the two scenarios in which
the schema explanation predicted that more aggressive men
would perceive the woman's reactions
as
more positive
(i.e.,
the
hostile and assertive scenarios), the woman's negativity indices
were weighted
positively,
whereas for
the two
scenarios in which
the schema explanation predicted that more aggressive men
would perceive her reactions
as
more negative
(i.e.,
the seductive
and friendly scenarios), they were weighted negatively. Also, to
make the analyses most comparable with Malamuth's (1986),
we
used the Acceptance of Interpersonal Violence
(AIV)
against
women attitude scale only, rather than the composite of several
attitude measures.
Regressions without penile
tumescence.
Using the full sam-
ple of 90 participants, four variables entered the equation in the
stepwise analysis, yielding a multiple correlation of
.71,
which
accounted for 50% of the variance. The entering variables (in
the order entered and with the probability values after all vari-
ables entered) were sexual experience (p < .0001), AIV (p <
.0006),
HTW
(p
< .002), and perceptions
(p
< .01).
In the forced entry analysis, after entering sexual experience
(p <
.0001), AIV
(p <
.0003), HTW
(p
<
.02),
and sexual domi-
nance (p < .06), the multiple correlation was .69. The amount
of variance accounted for
was
48%.
The addition of perceptions
yielded an F change that was significant, F(l, 84) = 6.66, p <
.01,
R
=
.72.
The percentage ofvariance accounted for
was
52%.
The additional percentage of variance explained by including
the perceptions measure was therefore 4%. With all variables
entered, all of the variables (including perceptions) made sig-
nificant unique contribution to the equation, except for sexual
dominance, which showed a nearly significant contribution (p
< .07).
Regressions with penile
tumescence.
With the 58 partici-
pants in the phase assessing penile tumescence, we found that
four variables entered the equation in the stepwise analysis,
yielding a multiple correlation of
.70
and accounting for
48%
of
the variance. The variables entering (in the order entered and
with the probability values after all variables entered) were the
perceptions index (p < .0002), rape tumescence index (p <
.002),
sexual experience
(p <
.002),
and HTW
(p
< .01).
In the forced entry analysis, after entering sexual experience
(p <
.001),
the rape tumescence index
(p <
.008),
AIV
(p <
.02),
HTW (p <
.
14),
and sexual dominance (p <
.
19),
the multiple
708NEIL M. MALAMUTH AND LISA M. BROWN
correlation was .69. The percentage of variance accounted for
was
48%.
The addition of the perceptions index yielded an F
change that was significant, F(\, 51) = 8.99, p < .005, R = .75.
The percentage of variance accounted for was
56%.
Therefore,
the increase in the percentage of variance explained with the
addition of perceptions
was
8%.
With all variables entered, only
perceptions
{p
< .002), sex experience (p < .002), and rape tu-
mescence (p < .007) made a significant unique contribution to
the equation, with HTW approaching significance (p < .08).
In all of the analyses, then, the perceptions index consistently
showed a significant contribution to predicting sexual aggres-
sion and appeared to exert
as
consistent an influence (and some-
times the strongest)
as
the most influential variables.5
Discussion
The data across the four scenarios were clearly more consis-
tent with the predictions of the suspicious schema explanation
than the other