INTERPERSONAL RELATIONS AND GROUP PROCESSES
On the Importance of Cognitive Evaluation as a Determinant of
R. Matthew Montoya
University of North Carolina at Chapel Hill
Robert S. Horton
Three studies examined a model of attraction in which the cognitive evaluation of the target individual
was the primary determinant of interpersonal attraction. In Study 1, the cognitive evaluation of the target
individual mediated the influence of attitude similarity on interpersonal attraction. In Study 2, a path
analysis revealed significant indirect effects of (a) similarity on cognitive evaluation via the valence of
information implied by attitudes and (b) the valence of information implied by attitudes on attraction via
cognitive evaluation of the target. Study 3 provided empirical and theoretical support for the uniqueness
of interpersonal attraction from cognitive evaluation. The implications of these data for existing attraction
theory are discussed, and a new model of interpersonal attraction is described.
One of the most robust phenomena in social psychology is the
similarity–attraction effect (Byrne, 1997). Across a variety of
populations and for countless different manipulations of similarity,
increased similarity to a target is associated with interpersonal
attraction to the target (hereafter we refer to this phenomenon as
the similarity effect). The similarity effect has been observed with
schoolchildren (Byrne & Griffitt, 1966; Gaynor, 1971; Tan &
Singh, 1995), undergraduates (Clore & Baldridge, 1968; Pilking-
ton & Lydon, 1997), and married couples (Murstein & Beck, 1972;
Rogers, 1999; Russell & Wells, 1991) and is robust for a number
of different types of information including, but not limited to,
personality traits (Carli, Ganley, & Pierce-Otay, 1991; Singh,
1973; Steele & McGlynn, 1979), attitudes (Bond, Byrne, & Dia-
mond, 1968; Byrne & Blaylock, 1963; Byrne & Clore, 1970),
demographic characteristics (Gillis & Avis, 1980; Spuhler, 1968;
Susanne & Lepage, 1988), and physical attractiveness (Hill, Ru-
bin, & Peplau, 1976; Peterson & Miller, 1980; Stevens, Owens, &
Schaefer, 1990; for a review, see Montoya, Horton, & Kirchner,
Despite the apparent power of similarity, there are situations in
which the similarity effect is diminished or nonexistent. For in-
stance, the similarity effect is weaker for personality traits than it
is for attitudes (Montoya et al., 2004), for peripheral attitudes
compared with central attitudes (Byrne, London, & Griffitt, 1968;
Clore & Baldridge, 1968; Horton & Montoya, 2003), and in field
studies compared with laboratory studies (Montoya et al., 2004;
Sunnafrank, 1992). Moreover, similarity does not produce inter-
personal attraction for negative traits (e.g., Ajzen, 1974; Novak &
Lerner, 1968). The dominant explanation of the similarity effect,
the reinforcement–affect model of attraction (Byrne, 1971; Byrne
& Clore, 1967), has difficulty explaining the inconsistent effects of
similarity on attraction (Ajzen, 1974; Montoya et al., 2004; Sun-
nafrank, 1992). In this article, we use the similarity effect para-
digm as an anchor for investigation to describe and test empirical
predictions of a theoretical model for interpersonal attraction that
may account better for the inconsistencies in the existing literature.
Explanations of the Similarity Effect
In an effort to explain the similarity effect, Byrne and Clore
(1967) suggested that a target who possesses similar attitudes is
reinforcing because the target’s attitudes confirm the legitimacy
and accuracy of one’s own attitudes (i.e., provide consensual
validation). This consensual validation, in turn, nurtures the effec-
tance motive, the basic need for a consistent, logical, and accurate
interpretation of the world. In contrast, dissimilar others offer no
such consensual validation and thus create need-threatening incon-
sistency and uncertainty regarding the self. In short, Byrne and
Clore argued that similar people make us feel good, and dissimilar
people make us feel bad, both about ourselves and about our
world. Interpersonal attraction flows directly from these affective
states. Unfortunately, the effectance motive does not explain ad-
R. Matthew Montoya, Department of Psychology, University of North
Carolina at Chapel Hill; Robert S. Horton, Department of Psychology,
We are grateful to Candy Cecil Ducheneaux, Jeffrey Kirchner, Brad
Pinter, and the SCAR Group for their assistance with this research and to
Sarah Huie, Anne Lentz, Jennifer Mayer, Amit Sharma, Rebecca Denton,
and Tara Taylor for their assistance with data collection.
Correspondence concerning this article should be addressed to R. Mat-
thew Montoya, Department of Psychology, University of North Carolina at
Chapel Hill, Chapel Hill, NC 27514. E-mail: email@example.com
Journal of Personality and Social Psychology, 2004, Vol. 86, No. 5, 696–712
Copyright 2004 by the American Psychological Association 0022-3514/04/$12.00 DOI: 10.1037/0022-3522.214.171.1246
equately the inconsistencies observed in the similarity effect. Al-
though the effectance motive addresses the issue of central versus
peripheral attitude similarity (validation of one’s central attitudes
satisfies the effectance motive to a greater extent than does vali-
dation of one’s peripheral attitudes and thus creates more positive
affect), the effectance motive has difficulty generalizing to other
findings related to similarity. As an example, consider similarity
between the self and another on negative qualities. According to
the reinforcement–affect model, similarity on negative qualities
should satisfy the effectance motive just as effectively as does
similarity on positive qualities and thus should lead to attraction.
However, similarity to another on negative qualities does not lead
to attraction (e.g., Ajzen, 1974; Novak & Lerner, 1968). Similarly,
the reinforcing ability of attitudes does not predict attraction in
field studies (Montoya et al., 2004).
Given this mixed support, it
is important to note that other models have attempted to explain
these diverse findings.
Information Integration Model of Attraction
The information integration approach (e.g., Kaplan & Anderson,
1973) critiques the reinforcement model by arguing that similarity
is only reinforcing when the individual is aware of the relationship
between reward and similarity (e.g., DeNike & Leibovitz, 1969;
Insko & Cialdini, 1969; Sachs & Byrne, 1970; Uleman, 1971). For
example, pairing a reinforcing photograph (e.g., of a friend) with
a nonsense syllable only increases attraction to the nonsense syl-
lable when the participant is aware of the relationship between the
photograph and the stimulus (Lott & Lott, 1969). Given that the
influence of affective processes is profound with and without
conscious awareness, the failure to find reinforcement effects in
the absence of awareness suggests that other, possibly cognitive,
processes must be operating when individuals express attraction to
similar targets (Fishbein & Ajzen, 1972).
The information integration model’s approach to the similarity
effect is dependent on the notion that individuals hold attitudes and
personality traits that they believe to be good and correct. Indeed,
there is theoretical and empirical evidence that people regard more
favorably personality traits and attitudes that are self-descriptive
than those that are not self-descriptive (Alicke, 1985; Stallings,
1970; Wetzel & Insko, 1982). As a result of the inherent positivity
with which individuals regard themselves and their attitudes,
Ajzen (1974) reasoned that an evaluator would infer favorable
information about a target from the target’s similar attitudes.
Consequently, the similarity effect was a result of the compilation
of this evaluative information and not of the reinforcement value
of similarity. Consistent with this idea, a number of researchers
have found that similarity does not predict attraction once one
controls for the positivity of the stimuli attributed to a target
(Lydon, Jamieson, & Zanna, 1988; McLaughlin, 1970). Moreover,
Ajzen manipulated both the valence and similarity of personality
traits attributed to a target and found that only the valence of the
personality traits predicted attraction to the target. These findings
suggest that the positivity of the information one infers about a
target plays a crucial role in the similarity effect.
As an extension of this idea, Kaplan and Anderson (1973) noted
that individuals also infer from attitudes and personality traits
additional information about a target. In Kaplan and Anderson’s
words, “When we are told that X has similar attitudes, we like him
not because that information acts as an unconditioned stimulus, but
because it leads us to expect that he has various positive aspects to
his personality” (p. 304). In short, others who hold attitudes that
are similar to ours are believed to possess positive personality
qualities that those who disagree with us do not. In turn, we will
feel more affection for similar, rather than dissimilar, others. In
support of this emphasis on the information implied by similar and
dissimilar attitudes, Horton and Montoya (2003) found that a target
who was described with positive and informative traits (traits that
were regarded favorably and that conveyed additional information
about a person’s character, e.g., cooperative, independent) was
more attractive than a target who was described with equally
positive, but uninformative traits (e.g., broadminded, prompt). An
emphasis on the information implied by an attitude—information
that may take the form of specific traits held by the target other
(e.g., caring, compassionate), other attitudes (e.g., attitude toward
the death penalty, abortion), or values (e.g., life values)—provides
a parsimonious resolution to discrepancies in the similarity liter-
ature. For instance, because of differences in the amount of im-
plied information, central attitude similarity (e.g., attitude toward
abortion, capital punishment, religion) should lead to more inter-
personal attraction than peripheral attitude similarity (e.g., attitude
toward reading, discotheques), and attitude similarity should lead
to more attraction than personality trait similarity (e.g., compas-
sionate, caring). Negative trait similarity should not lead to inter-
personal attraction because the valence of the information implied
by negative traits is unfavorable. The first goal of the current
project was to investigate the possibility that the similarity effect
results from the positive information that one infers about a target
from the target’s similar attitudes.
An additional goal of this project was to investigate the proxi-
mal determinant of interpersonal attraction within the similarity–
attraction paradigm. As discussed above, information integration
models (Anderson, 1971; Kaplan & Anderson, 1973) suggest that
attraction results from a weighted composition of the information
one compiles regarding a target (whether the information takes the
form of attitudes, personality traits, or inferences derived from
attitudes and personality traits). However, we argue that an assess-
ment of the cognitive quality of the individual precedes the judg-
ment of attraction. That is, individuals create cognitively a com-
posite of the information implied by the attitudes or personality
traits attributed to a target. This composite guides a cognitive
evaluation of the quality of the target, which in turn is the basis for
This proposed model, which includes the addition of a cognitive
composite of the available information, differs from information
integration and reinforcement models in subtle but distinct ways.
Compared with information integration theories (e.g., Anderson,
1971; Kaplan & Anderson, 1973), this proposed model of attrac-
tion shares the emphasis on information inferred from target char-
acteristics (attitudes or personality traits) but differs in its sugges-
Byrne (1992) presented a series of arguments for the reasons why the
similarity effect does not translate well into the field. However, most of the
identified factors (situational factors that contribute to the interpersonal
attraction process but that are unrelated to the reinforcing ability of atti-
tudes) relate to the reduction of increased affect from prolonged initial
interactions (Montoya et al., 2004).
COGNITIVE EVALUATION AND ATTRACTION
tion of an intermediate evaluative step: The cognitive evaluation
results from the valenced information acquisition and precedes,
and guides, interpersonal attraction. We argue that a cognitive
composition is generated before the assessment of interpersonal
attraction, whereas the information integration models make no
such assumption. With respect to the reinforcement model, the
proposed model differs in that the reinforcement model suggests
either that a cognitive evaluation and the experience of interper-
sonal attraction occur simultaneously (Byrne, Rasche, & Kelley,
1974) or that the experience of interpersonal attraction precedes
our cognitive evaluation (e.g., Byrne & Clore, 1970; Clore &
This conceptualization of the cognitive process and the experi-
ence of attraction can be conceptualized via the classic model of
attitudes (see Eagly & Chaiken, 1993, 1998). The tripartite model
of attitudes (e.g., Katz & Stotland, 1959; Rosenberg, Hovland,
McGuire, Abelson, & Brehm, 1960) postulates that an attitude, an
evaluation of a particular object, includes three components: af-
fective (emotional response toward the attitude object), cognitive
(thoughts and beliefs about the attitude object), and behavioral
(tendency to act in a particular way toward the attitude object).
Although interpersonal attraction is sometimes perceived as an
attitude toward another person, interpersonal attraction is most
commonly defined conceptually as an affective evaluation of an-
other person (e.g., Ajzen, 1974, Byrne, 1971; Fishbein & Ajzen,
1972; Foa & Foa, 1974). It is important to note, however, that
attraction is most frequently measured using both behavioral and
affective items (Byrne, 1971; Fishbein & Ajzen, 1972). There
seem to be two reasons for this: First, there is a near-perfect
correlation between the affective and behavioral components of
attraction (r .85; Byrne, 1971), and second, there is traditionally
a stronger correlation between the affective and behavioral com-
ponents of attraction than between the cognitive and either of the
other two components (Breckler, 1984; Breckler & Wiggins, 1989;
Woodmansee & Cook, 1967). Whether attraction is defined con-
ceptually or empirically, we suggest that attraction is principally
determined by a cognitive evaluation of another person.
Of course, the notion that a cognitive evaluation precedes and
predicts other evaluative responses is not new. Devine (1989)
demonstrated that the amount of prejudice one feels for an out-
group (one’s affective response toward a group) is dependent on
one’s stereotype of the group (one’s cognitive component). Other
researchers have spoken to the importance of distinguishing be-
tween the cognitive and other components of attitudes (see Miller
& Tesser, 1986), yet such distinctions have yet to be investigated
rigorously in the attraction literature.
As an example of our proposed model, take the hypothetical
persons Chris and Mary. When Mary expresses to Chris a pro-life
attitude, Chris’s attitude regarding abortion will determine (a) his
evaluation of the attitude and (b) the positivity or negativity of the
information that this attitude implies about Mary. If Chris is
pro-life, he might regard the pro-life attitude positively, and he
may infer that Mary holds other conservative attitudes, which he
regards positively (e.g., being for the death penalty or for increased
defense spending), and that Mary possesses certain positive per-
sonality characteristics, such as being conservative, religious, and
compassionate. From this information, Chris will arrive at a cog-
nitive evaluation of Mary (a positive one in this case) and will, in
turn, feel attracted to her. Thus, Chris is not attracted to Mary
because she shares his pro-life beliefs, per se. He is attracted to her
because (a) Mary’s pro-life attitude suggests to Chris that she
possesses other positive qualities, and (b) such positive inferences
about Mary guide a positive cognitive evaluation of her.
Purpose of the Studies
The goal of this research was to investigate the proposed model
of interpersonal attraction. The model, as illustrated in Figure 1,
suggests that the similarity effect proceeds in two stages: (a)
Attitudes imply information, which guides the cognitive evaluation
of the quality of the target person, and (b) the cognitive evaluation
of the target person guides attraction to the target. In other words,
we argue that cognitive evaluation of the quality of the target
mediates the impact of similarity on attraction and that the infor-
mation implied by similarity mediates the impact of similarity on
We conducted three studies to assess the model. Study 1 as-
sessed whether or not cognitive evaluation mediates the similarity
effect. Study 2 replicated Study 1 and tested the notion that the
similarity effect is doubly mediated by (a) the valence of the
information implied by attitudes attributed to a target and (b) one’s
cognitive evaluation of the quality of the target. Study 3 investi-
gated whether or not interpersonal attraction and cognitive evalu-
ation are unique and distinguishable.
Study 1 examined the possibility that the cognitive evaluation of
a target person mediates the similarity effect. We used Byrne’s
(1971) phantom-other technique to manipulate attitude similarity
and to assess cognitive evaluation of and interpersonal attraction to
the target person.
The principal objectives of this study were (a) to replicate the
similarity effect such that attitude similarity increases interper-
sonal attraction to a target, (b) to investigate temporal relations
between interpersonal attraction and cognitive evaluation, and (c)
to investigate the potential mediation of this similarity effect by
cognitive evaluation of the target. The second objective warrants
The proposed model suggests that the influence of cognitive
evaluation on interpersonal attraction is particularly pronounced
when the assessment of cognitive evaluation, which precedes
attraction, is made salient. In this case, cognitive evaluation of the
target influences interpersonal attraction most directly. We manip-
Figure 1. Hypothesized model: Similarity predicts information about the
individual, which then determines cognitive evaluation, which predicts
attraction toward the target other.
MONTOYA AND HORTON
ulated cognitive evaluation salience by having half of the partici-
pants assess the overall quality of the other prior to the interper-
sonal attraction assessment (evaluation salient) and half of them
assess the quality of the other following the interpersonal attraction
assessment (evaluation not salient). As a reminder, we believe that
similarity influences attraction via one’s cognitive evaluation of
the target. Thus, we predicted an interaction between similarity
and evaluation salience, such that interpersonal attraction to a
similar other should be greater when evaluation is assessed prior
to, rather than immediately following, interpersonal attraction.
Eighty-one participants (34 men and 47 women) at the University of
North Carolina at Chapel Hill participated in this study in partial fulfill-
ment of an introductory psychology class requirement. Participants were
assigned randomly to one of six experimental conditions created by a 2
(evaluation salience: salient vs. not salient) 3 (similarity condition:
similar vs. dissimilar vs. control) factorial design.
Cognitive Evaluation Questionnaire. Participants completed 7 items
(e.g., “In general, how good a person do you think your partner is?”), each
on a 9-point scale, to express their cognitive evaluation of the target person.
The 7 items were embedded within 12 filler items and were averaged to
form an index of cognitive evaluation (
.84). A complete list of items
is displayed in the Appendix.
Interpersonal Attraction Questionnaire. Participants completed nine
items, each on a 9-point scale, to express their attraction to the target
person. Adapted from Byrne and Wong’s (1962) Interpersonal Judgment
Scale, this measure included three affective attraction items and six be-
havioral attraction items. Ratings on the nine items (e.g., “How much do
you think you will like your partner?”) were averaged to form an index of
interpersonal attraction (
.93). For the complete list of items, see the
Participants reported for the experimental session in groups of between
3 and 6 persons. As participants arrived, they were seated separately in
private rooms. Care was taken to avoid verbal and visual contact between
participants before the experiment. The experimenter initiated a cover story
by informing participants that the study would take place in two sessions.
According to this cover story, during the first session participants would (a)
complete an attitude assessment, (b) obtain information about another
participant (their partner for the second session), and (c) form an impres-
sion of their partner. The second session was to involve the participant
interacting with the partner on a problem-solving task.
In the first experimental session, participants completed a brief attitude
questionnaire. The survey included 11 attitude items adapted from Byrne’s
(1971) list. Items included primarily central attitudes: belief in God and
attitudes toward smoking, acting on impulse, premarital sex, money, the
“one true religion,” novels, strict discipline of children, the nuclear arms
race, dancing, and men’s adjustment to stress. Each attitude was rated on
a 7-point scale ranging from I strongly disagree to I strongly agree. For
each item, participants indicated their attitude by checking the statement
that most closely described their own attitude. On completion, the exper-
imenter collected the attitude surveys, and participants completed a booklet
of filler tasks while the experimenter prepared the manipulated attitude
We followed Byrne’s (1971) protocol for manipulating similarity. For
both the similar and dissimilar condition, the experimenter created a bogus
attitude survey by marking attitudes that differed from the participant’s
attitudes by varying degrees (e.g., I strongly agree differs from I agree by
one degree). In the similar condition, the experimenter marked three
attitudes that were the same response that the participant had marked, three
attitudes that differed from the participant’s attitudes by one degree, two
attitudes that differed by two degrees, and one attitude that differed by
three degrees. In the dissimilar condition, the experimenter marked three
attitudes that differed from the participant’s by three degrees, three atti-
tudes that differed by two degrees, two attitudes that differed by one
degree, and one attitude that was the same as that which the participant had
marked. In the control condition, participants received no feedback about
their future partner, and thus, no surveys were prepared.
After preparing the manipulated surveys, the experimenter showed the
survey to the participant with the explanation that it was the survey that had
been completed by his or her future partner. One third of participants saw
a survey that included attitudes similar to their own, one third saw a survey
that included attitudes dissimilar to their own, and one third saw no survey
(control participants). After reviewing the manipulated surveys (or in the
control condition, immediately after completing the filler task), participants
completed the cognitive evaluation of their partner and the interpersonal
attraction assessment. To manipulate evaluation salience, half of the par-
ticipants completed the Cognitive Evaluation Questionnaire prior to the
Interpersonal Attraction Questionnaire; the remaining half completed the
Interpersonal Attraction Questionnaire prior to the Cognitive Evaluation
Questionnaire. After participants completed the two outcome measures, the
experiment concluded with a full debriefing.
We had three primary goals in our analysis of these data: (a) to
replicate the similarity effect, (b) to investigate an interaction
between evaluation salience and similarity, and (c) to assess the
mediational role of cognitive evaluation in the similarity effect.
Cognitive Evaluation Versus Interpersonal Attraction
To assess the overlap between our outcome measures, we sub-
jected all of the items of the Cognitive Evaluation Questionnaire
and the Interpersonal Attraction Questionnaire to a maximum-
likelihood principal-components analysis with a varimax rotation.
This analysis resulted in a two-factor solution that accounted for
63.7% of the variance. Important to this analysis, each of the
affective attraction and behavioral attraction items loaded on the
interpersonal attraction factor, whereas five of the seven of the
cognitive evaluation items loaded on the second factor (see Table
1 for factor loadings). These loadings and an investigation of the
scree plot suggest that cognitive evaluation and interpersonal at-
traction are distinct but correlated (r .68) factors.
To further assess the overlap of affective and behavioral com-
ponents of attraction, we measured the correlation between an
affective attraction composite (the average of the affective items in
our attraction assessment) and a behavioral attraction composite
(the average of the behavioral items in the assessment). As ex-
Across the three studies, there were no significant gender differences
for interpersonal attraction or cognitive evaluation. Moreover, for both
interpersonal attraction and cognitive evaluation, there were no significant
interactive effects of gender and any other independent variable.
COGNITIVE EVALUATION AND ATTRACTION
pected, these components were strongly correlated, r(78) .82,
We assessed the influence of similarity by subjecting cognitive
evaluation and interpersonal attraction to separate 2 (evaluation
salience: salient vs. not salient) 3 (similarity: similar vs. dis-
similar vs. control) analyses of variance (ANOVAs).
Interpersonal attraction. Consistent with previous research,
the main effect for similarity was significant, F(2, 75) 7.10, p
The similar-versus-dissimilar contrast (hereafter
referred to as the critical contrast) was of primary theoretical
interest; however, we included the second contrast (control vs.
similar/dissimilar) in the analysis in order to fully represent the
three-level factor. Preplanned orthogonal contrasts revealed that
participants were more attracted to similar, rather than dissimilar,
partners, t(78) 4.38, p .05. The combination of the similarity
and dissimilarity condition did not differ from the control, t(78)
0.83, p .40. There was no main effect for evaluation salience,
F(1, 75) 0.12, p .72,
The critical Evaluation Salience Similarity interaction was
significant, F(2, 75) 3.25, p .05,
.08. The interaction
suggests that the effect for similarity was larger when the evalu-
ation was salient rather than not salient. Table 2 displays the mean
interpersonal attraction ratings as a function of evaluation salience
and the three similarity conditions. When interpersonal attraction
was assessed before the cognitive evaluation, similarity did not
exert a significant influence on attraction, F(2, 33) 0.96, p
.02. In contrast, the similarity effect was significant
when the cognitive evaluation preceded the assessment of inter-
personal attraction, F(2, 44) 12.34, p .05,
Cognitive evaluation. With respect to cognitive evaluation, the
expected main effect for similarity was significant, F(2, 75)
12.49, p .05,
.25. Table 2 displays the mean cognitive
evaluation ratings as a function of the three similarity conditions
and evaluation salience. The data pattern for this effect was iden-
tical to that reported for interpersonal attraction: Participants eval-
uated similar targets significantly more favorably than they did
dissimilar targets, t(78) 5.21, p .05, whereas a combination of
the similarity and dissimilarity conditions did not differ from the
control condition, t(78) 1.43, p .14. No other effects, includ-
ing the Evaluation Salience Similarity interaction, F(2, 75)
0.78, p .45,
.02, reached significance.
To represent accurately the three-level similarity factor, both of
the orthogonal contrasts were included as predictors in these
analyses, even though only the first contrast (similarity vs. dissim-
ilarity) is relevant to our mediation expectations. We assessed the
potential mediation of the similarity effect by cognitive evaluation
using Baron and Kenny’s (1986) four-step model. According to
this model, mediation is established when (a) the independent
variable (IV) influences significantly the dependent variable (DV);
All p values are based on two-tailed tests; effect sizes are reported
We replicated the basic analysis for cognitive evaluation after remov
ing the two items that loaded heavily on the interpersonal attraction factor.
Consistent with original analysis, the main effect for similarity was sig-
nificant, F(2, 75) 8.69, p .05. Neither the main effect for evaluation
salience, F(1, 75) 0.24, p .62, nor the Evaluation Salience
Similarity interaction, F(2, 75) 0.41, p .66, reached significance.
Factor Pattern for Interpersonal Attraction and Cognitive Evaluation Items: Varimax Rotation
Interpersonal attraction items
I would like to meet my future interaction partner. .806 .234
To what extent do you want to work on the upcoming task with your partner? .843 .234
I would probably dislike talking with my future interaction partner at a party. .667 .324
I would enjoy discussing controversial topics with my future interaction partner. .567 .220
My future interaction partner would probably not make a good friend to me. .654 .245
How much do you think you will like your partner? .766 .402
I would like to get to know this person better. .853 .236
I think I would enjoy my future interaction partner’s company. .840 .309
To what extent are you looking forward to meeting your partner? .898 .187
Cognitive evaluation items
My future interaction partner is probably well-respected. .402 .645
My future interaction partner is probably good at everything that s/he does. .133 .790
In general, how good a person do you think your partner is? .417 .703
My future interaction partner will probably be successful in life. .305 .759
My future interaction partner probably achieves all of his/her goals. .190 .745
My future interaction partner could help me accomplish my goals. .520 .331
I think that my future interaction partner would make a good leader. .656 .434
Note. Boldface type indicates loadings greater than .50.
MONTOYA AND HORTON
(b) the IV influences significantly the proposed mediator; (c) the
proposed mediator predicts significantly the DV; and (d) including
the proposed mediator and the IV as predictors of the DV, the
influence of the proposed mediator remains significant, whereas
the influence of the IV is reduced significantly. With respect to
conducting the mediational analysis for Study 1, the significant
Evaluation Salience Similarity interaction for interpersonal at-
traction stresses the importance of the sequence of the mediator
and outcome variables. As a result, the mediational analysis in-
cludes only those participants who completed the cognitive eval-
uation prior to the attraction assessment.
The results of the mediational analysis are presented in Figure 2.
The first three mediation conditions were assessed and satisfied
(cf. Baron & Kenny, 1986; Judd & Kenny, 1981): Participants
were more attracted to similar, rather than dissimilar, partners
.54, t(48) 4.38, p .05; participants evaluated
similar partners more favorably than dissimilar partners (Step 2),
.59, t(48) 5.04, p .05; and cognitive evaluation predicted
interpersonal attraction (Step 3),
.71, t(48) 6.91, p .05.
Most important, when cognitive evaluation and the similarity
contrasts were included as predictors of attraction, cognitive eval-
uation predicted attraction significantly,
.60, t(48) 4.72,
p .05, whereas the critical similarity contrast was reduced to
.18, t(48) 1.42, p .16. The indirect effect
(Kenny, Kashy, & Bolger, 1998) of the similarity contrast on
interpersonal attraction via cognitive evaluation reached signifi-
cance (z 3.71, p .05); the path between similarity and
interpersonal attraction was reduced significantly when cognitive
evaluation was included in the model.
Because we were interested in investigating a temporal relation-
ship between cognitive evaluation and interpersonal attraction, it
was also important to test the plausibility of the alternative model,
that interpersonal attraction mediates the effect of similarity on
one’s cognitive evaluation. Consistent with the reasoning of the
previous analysis, we retained only those participants who as-
sessed interpersonal attraction before cognitive evaluation. This
alternative model also satisfied all of the prerequisites of a medi-
ational analysis. However, an investigation of the indirect effect of
similarity on cognitive evaluation revealed that interpersonal at-
traction did not reduce the effect of similarity reliably (z 1.35,
Two findings supported our prediction of the critical role of
cognitive evaluation in the similarity effect on attraction. First, our
mediation analysis was consistent with the hypothesis that cogni-
tive evaluation mediates the influence of attitude similarity on
attraction. Furthermore, it is important to note that the opposite
mediational pattern (mediation of similarity–evaluation link via
attraction) was not significant. Second, similarity only influenced
attraction when cognitive evaluation was measured prior to assess-
ment of attraction (when the evaluation was salient). This result
establishes a cognitive evaluation as the proximal determinant of
attraction and suggests that attitude similarity/dissimilarity impacts
attraction indirectly via cognitive evaluation of the target.
The results of Study 1 conformed to predictions; however, Study
1 did not assess directly the information implied by similar or
dissimilar attitudes. Without such assessment, a full explanation of
the similarity effect via the proposed model is incomplete. The
principal aim of Study 2 was to assess and investigate the medi-
ational role of the information implied by similar and dissimilar
attitudes. We expected the valence of the information implied by
similar and dissimilar attitudes to mediate the impact of similarity
on cognitive evaluation, which would, in turn, mediate the impact
of similarity on attraction. More specifically, we expected similar
attitudes to imply positive information about a target. This infor-
Mean Ratings of Cognitive Evaluation and Interpersonal Attraction as a Function of Attitude Information
Cognitive evaluation first Interpersonal attraction first Total
Similar Control Dissimilar Similar Control Dissimilar Similar Control Dissimilar
SD 1.08 1.26 1.02 1.18 0.95 1.10 1.16 1.11 1.06
SD 0.81 1.20 0.95 1.16 1.03 0.58 0.95 1.11 0.87
Note. N 81. Means ranged between 1 and 9 for both dependent measures. Higher values corresponded to greater levels of either interpersonal attraction
or cognitive evaluation. In each row, means sharing common subscripts do not differ at p .05.
Figure 2. Cognitive evaluation as a mediator of interpersonal attraction
from similarity in Study 1. The number in parentheses indicates the direct
effect of similarity on attraction prior to the inclusion of cognitive evalu-
ation in the regression equation. *p .05.
COGNITIVE EVALUATION AND ATTRACTION
mation should, in turn, be the foundation for the cognitive evalu-
ation of the target, which guides attraction to that target.
To investigate this additional component to the model, we
conducted Study 2. Study 2 was methodologically similar to Study
1. As in Study 1, participants expressed various attitudes, were
presented with the attitudes of a future partner, and then expressed
their interpersonal attraction to and cognitive evaluation of their
future partner. However, Study 2 also expanded on Study 1 in an
important way: Participants returned 1–2 weeks after the initial
session to complete the final stages of the experiment. At that time,
participants (a) wrote down information that they would infer
about an individual based on the attitudes attributed to their partner
and (b) rated the valence of this information.
Participants and Experimental Design
Sixty-nine introductory psychology students at the University of North
Carolina at Chapel Hill participated in partial fulfillment of a course option.
Participants were assigned randomly to one of three similarity conditions
(similar, dissimilar, or control).
The first session of this study was exactly the same as Study 1. Partic-
ipants completed an 11-item attitude survey, received an attitude survey
that was ostensibly completed by a partner (one third of the participants
saw similar attitudes, one third saw dissimilar attitudes, one third saw no
survey), and then completed assessments of cognitive evaluation and then
interpersonal attraction. Note that we made the cognitive evaluation salient
for all participants in Study 2. On the basis of the results of Study 1, it
seemed clear that the role in the similarity effect of the information implied
by attitudes could be assessed most effectively in the context of a salient
To prepare for Session 2, the experimenter created an attitude question-
naire that was idiosyncratic for each participant. This questionnaire in-
cluded the 11 attitudes that had been attributed in Session 1 to the
participant’s partner as well as 5 filler attitudes irrelevant to the experi-
ment. In Session 2, participants were asked to (a) consider independently
each of the attitudes listed and (b) list information that they would infer
about an individual on the basis of the knowledge that this individual
possessed each attitude. Control participants (who received no information
about their partner in Session 1) were assigned randomly to generate
information about attitudes that were either similar or dissimilar to their
own. After generating information, participants completed three items
assessing the valence of the information they wrote down for each attitude
(e.g., “How would you evaluate the information that you just listed?”;1
very negative, 9 very positive). Session 2 concluded with a full debrief-
ing of experimental procedures.
We calculated composite indices for interpersonal attraction
.81) and cognitive evaluation (
.89). With respect to the
information valence index, we first averaged the three items that
assessed the positivity of the information implied by each of the 11
s .90). This procedure left us with 11 subcom-
posite scores that assessed the positivity of the information implied
by each of 11 attitudes. We next formed the information valence
composite by averaging these 11 subcomposites (
We subjected the interpersonal attraction, cognitive evaluation,
and information valence composite indices to one-way (similarity
condition: similar vs. dissimilar vs. control) ANOVAs. We then
performed preplanned orthogonal contrasts comparing (a) similar
condition with dissimilar condition and (b) control condition with
the combination of similar and dissimilar conditions.
We continued by assessing the mediation of the influence of
similarity on attraction via cognitive evaluation. This analysis was
an effort to replicate the critical Study 1 result. Using regression
analyses (and the aforementioned orthogonal contrasts), we fol-
lowed the four-step procedure outlined by Baron and Kenny
(1986) for establishing mediation and computed indirect effects
consistent with each mediational relation.
Finally, we constructed with LISREL 8.50 (Jo¨reskog & So¨rbom,
2001) a path analysis for the proposed model of attraction. We
used the path coefficients from this analysis to compute the indi-
rect effects of (a) similarity on interpersonal attraction via both
information valence and cognitive evaluation, (b) similarity on
cognitive evaluation via information valence, and (c) information
valence on attraction via cognitive evaluation. For both the regres-
sion analyses described above and for the path analyses described
here, we included both of the orthogonal contrasts as predictors in
order to represent accurately the three-level similarity factor.
Univariate Influences of Similarity
There was a significant effect of similarity on attraction, F(2,
66) 6.62, p .05; cognitive evaluation, F(2, 66) 9.23, p
.05; and information valence, F(2, 65) 12.58, p .05. Table 3
displays the means for cognitive evaluation, interpersonal attrac-
tion, and valence of the implied information in each of the three
similarity conditions. Preplanned orthogonal contrasts revealed
that participants were more attracted to similar rather than dissim-
.50, t(65) 3.62, p .05; evaluated similar
partners more favorably than dissimilar partners,
4.25, p .05; and evaluated the information derived from similar
attitudes as more positive than the information derived from dis-
.66, t(64) 5.07, p .05.
Additional Analysis on Information Valence
To investigate the possibility that participants’ Session 2 infor-
mation valence ratings were influenced by receiving information
about their partner in Session 1, we subjected the ratings to a 2
(Session 2 attitudes: similar vs. dissimilar) 2 (Session 1 infor-
mation condition: information vs. control) ANOVA. This analysis
revealed only an effect of Session 2 attitudes, F(1, 64) 19.81,
p .05. Participants rated information implied by similar attitudes
(M 6.48, SD 1.02) as more positive than information implied
by dissimilar attitudes (M 5.14, SD 1.11). Receiving infor-
mation about the partner in Session 1 did not influence, solely or
interactively, information ratings in Session 2: Session 1 informa-
A1–2-week latency between sessions was included to reduce the
likelihood that participants would rate the attitudes as consistent with their
previous rating either (a) to appear consistent to themselves or the exper-
imenter (Davidson & Jaccard, 1979; Norman, 1975; Schwartz, 1978) or (b)
as a result of demand characteristics (Orne, 1962; Orne & Scheibe, 1964).
MONTOYA AND HORTON
tion main effect, F(1, 64) 0.54, p .50; Session 2 Attitude
Session 1 Information interaction, F(1, 64) 2.04, p .30.
Mediation of Similarity–Attraction by Cognitive
In accordance with Baron and Kenny’s (1986) first two medi-
ational requirements, the similarity contrast predicted interpersonal
attraction and cognitive evaluation significantly (see above). Par-
ticipants were more attracted to similar partners and evaluated
them more favorably than dissimilar partners. Addressing the third
mediational requirement, cognitive evaluation predicted signifi-
cantly attraction to the partner,
.64, t(67) 6.80, p .05.
Finally, when including the two similarity contrasts and cognitive
evaluation as predictors of attraction, the predictive power of
cognitive evaluation on attraction remained significant,
t(64) 5.30, p .05, and the influence of similarity was rendered
.18, t(64) 1.35, p .19. The indirect effect
of similarity on attraction via evaluation was significant (z 3.28,
p .05). As in Study 1, our mediation analyses were consistent
with our prediction that cognitive evaluation mediates similarity’s
influence on interpersonal attraction.
We constructed a path analysis to investigate the proposed
model of attraction. We began by predicting attraction, cognitive
evaluation, and information valence with the two orthogonal sim-
ilarity contrasts. As expected, the contrast comparing similar ver-
sus dissimilar condition was a significant predictor of each out-
come measure: attraction,
.76, t(67) 3.80, p .05;
.83, t(67) 4.63, p .05; and infor-
.86, t(67) 5.26, p .05. Next, we specified
paths consistent with the proposed model of attraction: Similarity
contrasts predicted information valence, information valence pre-
dicted cognitive evaluation, and cognitive evaluation predicted
attraction. As expected, the control contrast predicted information
.86, t(67) 5.26, p .05; information
valence predicted cognitive evaluation significantly,
t(67) 4.79, p .05;
and cognitive evaluation predicted attrac-
.68, t(67) 6.68, p .05. Specifying the
proposed model of attraction rendered the similarity–attraction
.26, t(67) 1.32, p .18, and the
indirect effect of similarity on attraction via information valence
and cognitive evaluation was significant,
.32, t(67) 3.13,
p .05. Information valence and cognitive evaluation mediated
the influence of similarity on attraction.
To investigate further the proposed model of attraction, we
assessed the critical components of the model. More specifically,
we computed the indirect effects of (a) similarity on cognitive
evaluation via information valence (i.e., the mediation of the
similarity–cognitive evaluation by information valence) and (b)
information valence on attraction via cognitive evaluation (i.e., the
mediation of the information valence–attraction link by cognitive
evaluation). As stated above, similarity predicted significantly
both information valence and cognitive evaluation when no link
between the two outcomes was specified. When specifying the
path from information valence to cognitive evaluation, information
valence predicted cognitive evaluation,
.38, t(67) 2.92, p
.05, and the path from similarity to cognitive evaluation,
t(67) 2.50, p .05, was reduced significantly (indirect effect:
z 2.53, p .05). Information valence partially mediated the link
between similarity and cognitive evaluation.
Speaking to the indirect effect of information valence on attrac-
tion via cognitive evaluation, information valence predicted sig-
nificantly cognitive evaluation,
.54, t(67) 4.79, p .05, and
.66, t(67) 5.74, p .05. When specifying the
path from cognitive evaluation to attraction, cognitive evaluation
predicted attraction significantly,
.50, t(67) 4.55, p .05,
and the path from information valence to attraction,
t(67) 3.34, p .05, was reduced significantly (indirect effect:
One might wonder about the overlap of information valence and
cognitive evaluation: The data do not suggest collinearity. The correlation
between the two measures, although significant, was of moderate magni-
tude (Cohen, 1992), r(67) .51, p .001.
An additional analysis was conducted to determine if the mediation of
the similarity–cognition evaluation link by the implied information was the
result of the participant’s attempt to generate responses at Session 2 that
were similar to their Session 1 responses. To investigate this possibility, we
assigned 19 participants to a “no-rate” control condition. These participants
underwent the same procedures as other participants with one exception:
They did not rate their attraction to or their cognitive evaluation of their
partner. If Session 2 ratings are a result of consistency motives activated by
evaluating the partner in Session 1, information ratings by no-rate partic-
ipants should differ significantly from those of rate participants. We
subjected information ratings to a 2 (similarity condition: similar vs.
dissimilar) 2 (information condition: information vs. control) 2 (rating
partner: rate vs. no rate) ANOVA. This analysis revealed only a main effect
of similarity condition, F(1, 79) 26.46, p .001; participants rated
information implied by similar attitudes (M 6.53, SD 1.01) as more
positive than the information implied by dissimilar attitudes (M 5.11,
SD 1.01). No other effects reached significance. Rating the partner in
Session 1 did not influence information ratings in Session 2; the media-
tional influence of information ratings cannot be attributed to consistency
Mean Ratings of Interpersonal Attraction, Information Valence,
and Cognitive Evaluation as a Function of Attitude Information
Dissimilar Control Similar
SD 1.55 1.20 1.21
SD 0.89 1.32 0.98
SD 1.42 1.08 1.11
Note. N 69. Means ranged between 1 and 9 for both dependent
measures. Higher values corresponded to greater levels of either interper-
sonal attraction or perceptions of quality. In each row, means sharing
common subscripts do not differ at p .05.
COGNITIVE EVALUATION AND ATTRACTION
z 3.30, p .05).
Cognitive evaluation partially mediated the
influence of information valence on attraction.
The results of Study 2 are consistent with the proposed model of
attraction. First, we replicated the findings of Study 1 such that the
effect of similarity on interpersonal attraction was mediated by the
cognitive evaluation of the quality of the target person. Addition-
ally, in a path analysis including all components of the proposed
model of attraction, the indirect effects of (a) similarity on cogni-
tive evaluation via information valence and (b) information va-
lence on attraction via cognitive evaluation were both significant,
suggesting partial mediation in both cases. Though these analyses
of indirect effects yielded only partial, not full, mediation, it is
important to note that the influence of similarity on attraction was
rendered nonsignificant by the inclusion of paths consistent with
the proposed model. Such findings provide substantial support for
the model as well as for the notion that it is cognitive evaluation
derived from the information implied by the similar attitudes that
increases interpersonal attraction when one is confronted by a
One might wonder about the possibility that these data were
influenced by demand characteristics (Orne, 1962; Orne &
Scheibe, 1964) or by the operation of consistency motives (Da-
vidson & Jaccard, 1979; Norman, 1975; Schwartz, 1978). If one
rates a target positively, one may be likely to express attraction to
the target and to generate positive information about the attitudes
attributed to that target simply to avoid the appearance of incon-
sistency. Given the legitimate threat of such confounds, we inves-
tigated the possible mediation of the similarity–cognitive evalua-
tion link by interpersonal attraction and included an additional
control group, one that did not rate the target about whom they
We found no evidence of demand or consistency bias. Interper-
sonal attraction did not mediate the similarity–cognitive evalua-
tion link (as a reminder, attraction did not mediate this link in
Study 1, either). Further, participants who rated a target in Session
1 generated information in Session 2 that was of similar valence to
that generated by participants who did not rate a target in Session
1 (and thus were not susceptible to consistency motives or demand
characteristics). If participants were motivated by consistency mo-
tives or demand characteristics, reciprocal mediations or differ-
ences between those who rated and did not rate the target would
have been observed.
Studies 1 and 2 conformed well to predictions expounded from
the proposed model; however, one might argue that cognitive
evaluation and interpersonal attraction are synonymous constructs
within the attitude framework and that the reinforcing properties of
similarity are influencing all aspects of one’s attitude toward
another, specifically, one’s cognitive evaluation and affective eval-
uation of the target individual. If this notion has merit, it would be
no surprise that cognitive evaluation mediates the similarity effect.
Although the possible collinearity of cognitive evaluation and
interpersonal attraction is certainly worthy of consideration, we
should note that although Studies 1 and 2 provide evidence for the
mediation of the similarity effect by cognitive evaluation, no
evidence for the reciprocal mediation (similarity–evaluation link
by interpersonal attraction) was found. In spite of this evidence
against the overlap of cognitive evaluation and interpersonal at-
traction, we conducted Study 3 to investigate the independence or
collinearity of these two concepts. We expected that the concepts
would be correlated but would respond differently to a unique
interpersonal situation, one of social evaluative threat. Our efforts
to create an evaluative threat and nonthreat condition depended on
the “pratfall effect” (Aronson, 1998; Aronson, Willerman, &
Floyd, 1966) and its theoretical explanations.
In their famous pratfall study, Aronson et al. (1966) found that
attraction toward a competent target was greater when the target
committed an embarrassing blunder than when the target did not
commit the blunder. Helmreich, Aronson, and LeFan (1970) ex-
plained that the change in attraction was principally caused by the
“humanizing” effect of the pratfall. Alternatively, it may be that
the blunder’s impact on interpersonal attraction resulted from a
change in the perceived probability that one would be evaluated
negatively by the target (Stapel & Tesser, 2001; Tesser, 2000).
That is, our attraction to a target (i.e., how much would you like to
work, or be, with this person) is influenced by the way we expect
the target to evaluate us. If we expect a person to regard us
negatively (as we might expect when confronted by a flawless
person), we will be less attracted to this target than to a target who
is competent but who is less likely to insult us. In effect, the
blunder increases interpersonal attraction toward a competent
other because it alleviates the danger of a negative evaluation for
one’s self-esteem. Going further, whereas the potential for nega-
tive evaluation of the self should impact attraction to a target, the
potential threat should not influence one’s cognitive evaluation of
the quality of the other person. Indeed, we propose that cognitive
evaluation is a relatively selfless, cognitive process that proceeds
without consideration of threat to or protection of the self.
We reason that threat is a function of the quality of the person
with whom one interacts and the extent to which the person has the
ability to evaluate the self, either positively or negatively. Thus, in
Study 3, we manipulated threat by varying the quality of the target
person (from poor to exceptional) and the intensity of the task with
which participants expected to work with a partner. Half of the
participants expected their contact with the partner to be minimal,
thus avoiding any danger of negative evaluation. The remaining
half of participants expected their work with the partner to be
The indirect effects of similarity on cognitive evaluation via informa
tion valence and of information valence on attraction via cognitive evalu-
ation were computed with the modification of Sobel’s test (reported above)
and thus, are reported as z statistics. The indirect effect of similarity on
attraction via information valence and overall attraction was computed by
LISREL 8.50, which yields a t statistic. Despite the different statistics,
interpretation of the indirect effects is qualitatively similar.
We also assessed the relative fit of the proposed model versus a
competing model in which cognitive evaluation preceded information
valence. Because one model was not nested within the other, a statistical
test of the relative fit was not possible (P. Curran, personal communication,
September 14, 2003). However, we used the Akaike information criterion
(AIC; Akaike, 1987) to compare the two models. Indeed, the proposed
model fit the data more parsimoniously (AIC 37.55) than did this
competing model (AIC 48.01).
MONTOYA AND HORTON
intense and time consuming, thus creating the possibility of neg-
ative evaluation of the self. Because we expected interpersonal
attraction to be self-interested and cognitive evaluation to be
selfless, we predicted that interpersonal attraction and cognitive
evaluation would be highly correlated when there was no threat to
the self (when contact was minimal and when the partner was no
more accomplished than the self). After all, in the absence of
threat, one’s attraction to a partner should be guided by one’s
cognitive evaluation of the partner. In contrast, in the face of threat
to the self (when confronted by an exceptional other with whom
one must work at length), we expected interpersonal attraction and
cognitive evaluation to be relatively unrelated. In this case, one’s
cognitive evaluation should be a function of the quality of the
individual, whereas interpersonal attraction should decrease as the
potential for a negative evaluation of the self increases.
To test this proposition, participants anticipated an interaction
with another participant who varied in ability from poor (low
competence) to exceptional (high competence) on a given at-
tribute. We also manipulated participants’ expectations about the
intensity of the interaction with their partner by informing partic-
ipants that their interaction with their partner would be either
intense (they would work closely together) or minimal (they would
have minimum interaction with their partner). Participants then
expressed their attraction to and their cognitive evaluation of their
partner. Under conditions of low threat (when the other participant
was no more accomplished than the self and when interaction with
an exceptional partner was minimal), we expected that interper-
sonal attraction and cognitive evaluation would covary: As cogni-
tive evaluation increased, so would interpersonal attraction. Under
conditions of high threat (when an intense interaction with an
exceptional partner was expected), we expected cognitive evalua-
tion to be favorable but attraction to be minimal. Alternatively, we
expected that in the minimal interaction condition, cognitive eval-
uation and interpersonal attraction should be highly correlated: As
quality of the partner increased across conditions, so should at-
traction to the partner. In the intense interaction condition, we
expected attraction and cognitive evaluation to covary for those
conditions in which the other participant was not regarded as
exceptional. However, in the exceptional partner–intense interac-
tion condition, we expected participants to evaluate the partner
more favorably but express less attraction than would others.
Another potential technique for assessing the role of threat was
to manipulate the final component of Tesser’s (2000) social com-
parative model—the degree to which the trait is central to the
self-concept. Tesser suggests that one is more likely to experience
threat when feedback implicates a dimension that is central (i.e.,
important) to the self-concept. We manipulated centrality by giv-
ing half of the participants feedback regarding intelligence (a
central dimension) and the remaining half feedback regarding
artistic ability (peripheral dimension). Feedback on a central di-
mension, compared with a peripheral dimension, should provide
more threat and, as a result, a greater opportunity to evaluate the
differentiation between attraction and cognitive evaluation. We
expected an interaction such that a decrease in interpersonal at-
traction for an exceptional partner with whom one expected to
interact intensely would occur only when the partner was superior
on a central, rather than peripheral, dimension.
Participants and Experimental Design
Two hundred forty-seven participants (91 men and 156 women) from the
University of North Carolina at Chapel Hill participated in this study in
partial fulfillment of an introductory psychology requirement. Participants
were assigned randomly to one of 16 experimental cells created by a 4
(partner ability: control vs. poor vs. average vs. exceptional) 2 (inter-
action intensity: intense vs. minimal) 2 (dimension centrality: central vs.
peripheral) factorial design.
Participants reported to the experimental session in groups of between 3
and 6 persons. As participants arrived, they were seated in individual
private rooms. Participants were told that the experiment would be in three
parts. First, participants rated themselves on 10 dimensions (e.g., intelli-
gence, social skills, artistic ability, leadership ability) on 11-point rating
Next, participants were told that they would complete a task with another
participant and that this task would begin in a few minutes. Before the next
part of the study could occur, however, the experimenter had to pair the
participants in this experimental session with participants from a different
experimental session that was occurring concurrently “down the hall.”
Participants were told that the other session was identical to their own, with
one exception: Participants in the other session were rating each other,
instead of themselves, on the personality dimensions. The manipulation of
threat followed this information. Participants in the intense interaction
condition were told that their partner would soon come into their individual
room to work closely with them on a collaborative task. In the minimal
interaction condition, participants were told that their partner would remain
in their own room down the hall, and that they would have limited
correspondence with their partner.
Additionally, and ostensibly to facilitate the interaction, we told partic-
ipants that we were going to provide to them some information about their
future partner before they began the collaborative task. Participants re-
ceived feedback about the intellectual ability (central trait) or artistic ability
(peripheral trait) of their future partner. To manipulate ability of their
future partner, participants either received no feedback at all (control
participants) or were informed that their partner had scored at the 50th
(poor), 75th (average), or 100th (exceptional) percentile of the given
After viewing the information (or, in the control condition, immediately
after the manipulation of the degree of expected interaction), participants
completed the Cognitive Evaluation Questionnaire and the Interpersonal
Attraction Questionnaire. Participants were thanked for their time, de-
briefed, and then dismissed.
We began by calculating composite indices for interpersonal
.87) and cognitive evaluation (
.76). We then
subjected interpersonal attraction and cognitive evaluation to a 4
(partner ability: control vs. poor vs. average vs. exceptional) 2
(interaction intensity: intense vs. minimal) 2 (dimension cen-
trality: central vs. peripheral) ANOVA.
With respect to interpersonal attraction, the main effect of
ability of partner reached significance, F(3, 240) 3.88, p .05,
COGNITIVE EVALUATION AND ATTRACTION
.04. Participants were more attracted to the exceptional
partner (100th percentile) than to any other partner. This main
effect was qualified by the expected Interaction Intensity Part-
ner Ability interaction, F(3, 240) 3.07, p .05,
3 displays the means for interpersonal attraction as a function of
interaction intensity and partner ability. Simple effects tests re-
vealed that consistent with expectations, participants who thought
they were to interact with an exceptional partner were less at-
tracted to the partner in the face of an intense (M 5.90, SD
0.96) rather than a minimal (M 6.45, SD 1.23) interaction,
t(57) 2.19, p .05. In contrast, the control group was more
attracted to the target in the intense (M 5.99, SD 1.21) as
compared with the minimal (M 5.47, SD 0.83) interaction
condition, t(71) 2.05, p .05. Attraction did not vary as a
function of interaction intensity in either the poor partner condi-
5.70 vs. M
5.62), t(57) 0.29, p .05,
or the average partner condition (M
6.20 vs. M
6.04), t(48) 0.53, p .05. The main effect of threat did not
reach significance, F(1, 240) 0.16, p .68,
pectedly, the Interaction Intensity Partner Ability Type of
Feedback interaction failed to reach significance, F(3, 240)
2.27, p .13,
.001. Attraction to a target of exceptional
ability varied as a function of interaction intensity but not as a
function of the type of feedback.
With regard to cognitive evaluation, the expected main effect of
partner ability was significant, F(3, 240) 21.29, p .05,
.21. Figure 4 displays the means for cognitive evaluation as a
function of interaction intensity and partner ability. A set of
orthogonal contrasts revealed that (a) the control group rated their
partners less favorably than did the experimental groups, F(1,
240) 48.54, p .05; (b) participants evaluated the average and
exceptional partners (combined) more favorably than they did the
poor partners, F(1, 240) 7.78, p .05; and (c) the exceptional
partner was rated more favorably than the average partner, F(1,
240) 7.40, p .05. As expected, neither the Interaction Inten-
sity Partner Ability interaction, F(3, 240) 0.31, p .81,
.00, nor the main effect of interaction intensity, F(1, 240) 0.004,
.00, reached significance. The Interaction Inten
sity Partner Ability Type of Feedback interaction failed to
reach significance, F(3, 240) 0.40, p .75,
evaluation did not vary as a function of interaction intensity (and
thus was not affected by threat to the self) but was a near-linear
function of partner ability.
Correlation Between Cognitive Evaluation and
To further assess the relationship between cognitive evaluation
and interpersonal attraction as a function of threat, we computed
the correlation between cognitive evaluation and interpersonal
attraction separately for the threat and no-threat conditions. We
hypothesized that there would be no relationship between inter-
personal attraction and cognitive evaluation when participants
were confronted with a potential threat to the self (where a nega-
tive evaluation was possible): when they interacted intensely with
an exceptional partner. Alternatively, we expected to find a cor-
relation between cognitive evaluation and interpersonal attraction
when participants were not confronted by a threat (i.e., when no
negative evaluation was possible): when interacting minimally
with a partner or when interacting intensely with a partner who
As expected, the correlation between cognitive evaluation and
interpersonal attraction was not significant in the threat condition
(the intense interaction–exceptional partner condition), r(41)
.14, p .35, but was significant in the no-threat conditions (all
other conditions) combined, r(209) .38, p .01. Though this
large descriptive difference did not reach standard levels of sig-
nificance (z 1.47, p .15), the correlations suggest that attrac-
tion and cognitive evaluation covaried to a greater degree when the
self was not threatened by the possibility of a negative evaluation.
One may suggest that the threat from the other person may have
differentially influenced the affective and behavioral components of their
attraction assessment. Under conditions of threat, the correlation between
affective items and behavioral items, r(41) .74, p .05, was not
different than that under conditions of no threat, r(209) .69, p .05 (z
0.71, p .48).
Figure 3. Relationship between interaction intensity and ability of partner
on interpersonal attraction (Study 3).
Figure 4. Relationship between interaction intensity and ability of partner
on cognitive evaluation (Study 3).
MONTOYA AND HORTON
We followed the procedure outlined in Studies 1 and 2 to
determine if cognitive evaluation mediated the effects of quality of
partner on interpersonal attraction. As a reminder, we hypothe-
sized that cognitive evaluation should mediate the effect of partner
ability on interpersonal attraction only in the face of a minimal
interaction with the other participant (when there was no threat to
the self). We conducted two mediational analyses, one for the
minimal interaction condition and a second for the intense inter-
Minimal interaction condition. To represent the four-level
partner ability variable, we created three contrast variables: (a)
exceptional versus control, poor, and average; (b) average versus
control and poor; (c) and poor versus control. The first contrast
was the contrast of interest; however, all contrasts were included in
regression equations in order to represent fully the four-level
partner ability factor.
Participants were more attracted to the exceptional partner than
to the other partners,
.29, t(129) 3.49, p .05, and
evaluated the exceptional partner more favorably on cognitive
evaluation than the other partners,
.38, t(129) 4.79, p .05,
effectively fulfilling the first two mediational requirements. With
respect to the third mediational requirement, cognitive evaluation
predicted interpersonal attraction significantly,
4.75, p .05. Finally, when including the three similarity contrasts
and cognitive evaluation as predictors of interpersonal attraction,
cognitive evaluation predicted interpersonal attraction signifi-
.29, t(129) 3.28, p .05, and the critical contrast
.17, t(129) 2.04, p .05. The indirect
effect of partner ability on interpersonal attraction via cognitive
evaluation also reached significance (z 2.67, p .05). Cognitive
evaluation partially mediated the influence of partner ability on
Intense interaction condition. The same contrasts were used in
the analyses for the intense interaction condition. Consistent with
our expectations, participants evaluated the exceptional partner
more favorably than the other partners,
.40, t(119) 4.93,
p .05, but did not express more attraction toward the exceptional
partner relative to the other partners,
.02, t(119) 0.29, p
.77. These results conformed to our hypotheses and effectively
derailed the mediation under investigation.
Study 3 provides support for the relative independence of the
cognitive evaluation and interpersonal attraction constructs: The
two measures were influenced differently by the manipulations of
interaction intensity and partner ability. As expected, when partic-
ipants were not faced with the possibility of a negative evaluation
from the partner (e.g., in the minimal interaction condition and
when faced with an intense interaction with a poor or average
partner), interpersonal attraction increased as the quality of the
partner increased. In contrast, when a negative evaluation was
made possible by an expected intense interaction with a highly
competent partner, interpersonal attraction decreased. Addition-
ally, cognitive evaluation of the partner was unaffected by inter-
action intensity and increased, relatively linearly, with ability of
the partner. The data confirm that interpersonal attraction and
cognitive evaluation are unique constructs; however, this unique-
ness does not mean that they are completely unrelated.
In fact, consistent with Studies 1 and 2, cognitive evaluation
mediated the impact of partner ability on interpersonal attraction
when there was no threat to the self (in the minimal interaction
condition). This result further demonstrates that interpersonal at-
traction is often grounded in one’s cognitive evaluation of the
target. However, in the face of specific self-relevant circumstances
(e.g., in the presence of a potential threat to self-esteem), the two
variables follow unique paths. Such findings are consistent with
past research. For instance, Amabile (1983) found that although
participants regarded individuals who criticized others as more
intelligent, these individuals were also rated as less likable. Fur-
ther, Herbst, Gaertner, and Insko (2003) found that when partici-
pants were confronted by a target whose ability exceeded that of
their ideal self, they evaluated the target favorably but expressed
less attraction to the target than they had to other less able targets.
The current data fit with this existing literature and suggest that
self-relevant motives, activated by a threat to the self, attenuate the
relation between cognitive evaluation and attraction.
This series of studies identifies a novel mechanism that under-
lies the similarity effect. Whereas previous theories have tapped
the reinforcement value of similar attitudes (e.g., Byrne & Clore,
1967) or the evaluative nature of the attitudes or personality traits
attributed to a similar partner (Ajzen, 1974; Kaplan & Anderson,
1973), we posited that similar attitudes, though evaluated posi-
tively themselves, also imply additional positive information about
a target (in the form of additional descriptive attitudes and per-
sonality traits) and that this positive information guides a favorable
cognitive evaluation of the target that, in turn, drives attraction.
Studies 1 and 2 provide evidence for each component of this model
within the similarity paradigm: (a) Cognitive evaluation mediated
the effect of similarity on interpersonal attraction, and (b) infor-
mation valence mediated the impact of similarity on cognitive
evaluation. It is important to note that Study 3 provides critical
evidence that interpersonal attraction and cognitive evaluation are
An assessment of a target via a cognitive evaluation provides a
parsimonious explanation for some of the inconsistencies in the
similarity literature. For instance, people are not attracted to targets
who share with them characteristics that they regard negatively
(e.g., Ajzen, 1974; Novak & Lerner, 1968). By our reasoning,
similarity on negative characteristics would imply negative infor-
mation about a target, which would make one’s cognitive evalu-
ation of the individual poor, thus not creating attraction. Addition-
ally, attitude similarity elicits more attraction than does personality
similarity (Byrne, London, & Griffitt, 1968; Clore & Baldridge,
1968). It is important to note that attitudes tend to imply more
information about an individual’s character than do personality
traits (Horton & Montoya, 2004). Thus, given that both similar
attitudes and personality traits convey positive information about a
target, similar attitudes convey more positive information than
similar personality traits. This additional information translates
into a more positive evaluation of the target and, in turn, more
attraction to the target as a result of attitude, compared with
COGNITIVE EVALUATION AND ATTRACTION
Why Should Cognitive Evaluation Be Important to
Though we have described cognitive evaluation as an explana-
tion for the similarity effect, cognitive evaluation is also the
cornerstone of the proposed model of attraction that we believe
accounts well for a number of factors that impact interpersonal
attraction and close relationships. This model is similar to an
information integration model. However, its uniqueness lies in the
identification of cognitive evaluation as the proximal determinant
of attraction and as the intermediate step between attitude–
personality trait evaluation and attraction. The model’s application
to other attraction and relationship variables also depends on this
For example, evolutionary psychological models have identified
specific qualities or traits that lead to increased levels of interper-
sonal attraction (Buss & Barnes, 1986; Buss et al., 1990), and these
models have identified sex differences in mate preferences that
correspond to evolutionary needs. Women tend to value earning
capacity, high status, and ambition in mates more than do men
(Buss & Barnes, 1986; Howard, Blumstein, & Schwartz, 1987). In
contrast, men, relative to women, value physical attractiveness and
youth in mates. According to the theories of natural and sexual
selection (Darwin, 1859; Symons, 1979), these trait preferences
reflect the fundamental drive to increase genetic fitness and evo-
lutionary success, a drive that is ideally served by different char-
acteristics in mates dependent on sex. Mating with someone who
does not possess the preferred characteristics would, theoretically,
result in fewer, less fit offspring compared with mating with one
who possesses such characteristics. In the language of the pro-
posed model of attraction, the evolutionarily preferred character-
istics define for each sex what is a “high-quality” mate (a mate
who would be evaluated positively): Those who possess the char-
acteristics are high quality. Those who do not are “low quality.”
Men and women should be attracted to high-quality individuals in
order to produce the most, and most fit, offspring.
Additionally, the social psychological literature is littered with
theoretical perspectives that emphasize the role of target quality
for predicting attraction. Self-expansion theory suggests that mate
selection is guided by efforts to expand the self. High-quality
partners may provide self-expansion opportunities that low-quality
partners do not (Aron, Paris, & Aron, 1995). Vicarious enhance-
ment perspectives suggest that individuals are drawn to partners
because of the social prestige or self-esteem boost they receive
from identification with the partner. In this case, quality of the
partner may be defined as superior attractiveness or financial
security and may also be defined socially, as status or prestige.
Role of Cognition and Affect in the Attraction Process
The current findings have implications not only for the role of
cognition in the interpersonal attraction process but also for the
interplay between cognition and affect in attitude formation. Be-
cause this model stresses the role of a cognitive evaluation as the
proximal determinant of attraction, we suggest that attraction is
guided by a cognitive assimilation of the information that is
available about a target. This emphasis on the cognitive bases of
attraction is not only consistent with attitude models (Ajzen &
Fishbein, 1977) and information integration perspectives on attrac-
tion (Anderson, 1971) but is also bolstered by a growing body of
research that regards cognitive processes as primary for a number
of evaluative responses. For instance, cognitive evaluation has
been tapped as critical for understanding how prejudice develops
from stereotypes (Devine, 1989; Hoyle, Pinkley, & Insko, 1989),
for how consumer satisfaction flows from product information
(Oliver, 1993), for assessing physical attractiveness from a cogni-
tive appraisal of appearance (Kenrick, Montello, Gutierres, &
Trost, 1993), for predicting emotional attachment to a group from
an evaluation of the group (Molm, Takahashi, & Peterson, 2000),
and for predicting anti-Black affect from a cognitive assessment of
preadult experiences (Sears & Henry, 2003; for a theoretical re-
view, see Ajzen & Fishbein, 1977; Fishbein & Ajzen, 1972).
We should note here, as we have above, that the proposed model
contrasts with models that emphasize the affective underpinnings
of the similarity effect (Byrne, 1971). In fact, we cannot rule out
such models with the current data. After all, in Study 3 we found
that attraction, but not cognitive evaluation, was influenced by a
threat to self-esteem (an affective threat). This finding suggests
that, indeed, attraction involves an affective component that is
influenced by affective considerations. In the face of a threat to the
self, self-protective motives guide decreased attraction to an ex-
ceptional individual who could evaluate the self negatively (also
see Fromkin, 1972; Sedikides, 1993; Sigall & Landy, 1973).
However, it is particularly noteworthy that cognitive evaluation
was the proximal determinant of attraction in all three studies
reported here, mediating the similarity effect in both Studies 1 and
2 and the influence of partner ability in Study 3. Additionally,
attraction did not mediate the impact of similarity on cognitive
evaluation in Study 1, a result that would have been predicted by
affective models of attraction. In short, even though affect has
been demonstrated to be a determinant of attraction, the data
suggest that one’s cognitive evaluation of the quality of the partner
is a proximal determinant of attraction.
Finally, we should note that the findings reported here are
consistent methodologically and empirically with much of the
research conducted within the similarity paradigm. In much of this
research, attraction has been operationalized as the composite of
two items from the Interpersonal Judgment Scale (IJS; Byrne,
1971), one affective item (i.e., “How much do think you will like
this person?”) and one behavioral item (i.e., “How much would
you want to work with this person?”). The four items that precede
the attraction items on the IJS, normally regarded as filler items
(Byrne, 1971), assess the participant’s evaluation of the target’s
intelligence, morality, knowledge of current events, and adjust-
ment. We would argue that these four items make up an assess-
ment of the overall quality of the target (akin to our cognitive
evaluation scale). Given the Study 1 finding that similarity im-
pacted attraction only when evaluation was made salient, it is not
surprising to us that Montoya et al. (2004) found a descriptively
more powerful similarity effect in studies that used the IJS with,
rather than without, those four filler items.
Limitations and Future Directions
The repeated demonstrations in this article of the influential role
of cognition in the attraction process support the notion that the
proposed model of attraction accounts well for laboratory-based
investigations of the similarity effect. One is left to wonder,
MONTOYA AND HORTON
however, whether the proposed model generalizes from the labo-
ratory setting to actual relationships and, additionally, to other
factors that predict attraction. In the absence of empirical data
using actual relationships, we can only speculate about the role
cognition plays in attraction outside of the laboratory. However,
our model’s reliance on a cognitive evaluation would seem to
provide an effective foundation to generalize to a wide array of
laboratory and real-world situations as well as to a variety of
factors that impact attraction.
One of the primary concerns about the generalizability of our
findings to actual relationships lies in our use of the phantom-other
paradigm (in Studies 1 and 2). After all, this paradigm has pro-
duced a similarity effect that has proven difficult to replicate in
actual relationships (see Montoya et al., 2004; Sunnafrank, 1992).
Participants in the phantom-other paradigm assess attraction and
evaluation as a function of one cue to quality: similarity. Alterna-
tively, in actual relationships there are numerous sources of infor-
mation that could affect attraction, including, but not limited to,
physical attractiveness (Sloman & Sloman, 1988; Walster, Aron-
son, Abrahams, & Rottmann, 1966), behavior toward other indi-
viduals (Leahy, 1979), and the response of the target to the self
(Kerr & Kaufman-Gilliland, 1994). We would argue, in fact, that
these other information sources are cues to quality and help de-
termine one’s cognitive evaluation and, resultantly, impact attrac-
tion to the target. In short, the proposed model of attraction, in
which a cognitive evaluation is the key precipitate to attraction,
should generalize beyond the similarity effect to the variety of
forces that act on attraction in actual relationships.
Bearing further on the generalizability of the proposed model is
the question of whether the cognitive evaluation that guides at-
traction is automatic and immediate. We propose that a degree of
cognitive processing occurs immediately on exposure to the stim-
ulus and that this automatic evaluative processing can guide at-
traction. This proposition is supported by the ever-mounting evi-
dence of an influential automatic component to cognitive
evaluations (see Banaji, 2001; Fazio, Sanbonmatsu, Powell, &
Kardes, 1986; Greenwald, McGhee, & Schwartz, 1998). As an
example, Nosek, Banaji, and Greenwald (2003) found that implicit
gender stereotypes regarding math (an automatic cognitive asso-
ciation between males and math) predicted males’ and females’
explicit attitudes (including their attraction) about the subject.
We must also acknowledge here that we are not the first to
propose a model in which cognition precedes an affective re-
sponse. As stated previously, Devine (1989) has espoused such a
model for out-group attitudes, and Lazarus (1982, 1984) described
a similar model in which an individual’s emotional response to a
given situation is initially determined by a cognitive appraisal of
the situation. Such cognitive analyses should extract importance
and meaning from stimuli attributed to a target (e.g., similar or
dissimilar attitudes, favorable or unfavorable personality traits)
and then should influence our evaluation of the individual as a
whole. Given this reasoning, it seems logical that making such
stimuli salient—emphasizing the characteristics or attitudes of the
individual—should polarize the cognitive and, in turn, the affec-
tive responses to a target. The results of Study 1 support this
contention: The similarity effect was more powerful when the
participant evaluated the quality of the target before expressing his
or her attraction to the other target. It seems critical to note that
cognitive processes would not only occur in the laboratory setting
but in actual relationship dynamics. Future research may do well to
further elucidate the relationship between affect and cognition and
the temporal order between the components.
Finally, it is important to note that our cognitive evaluation
measure includes a pair of items related to achievement orienta-
tion. Achievement is a fundamental human motive that is linked to
interpersonal judgment (Austrin, 1965; Gruenfeld & Arbuthnot,
1969; Harackiewicz, Abrahams, & Wageman, 1987; Mowaiye,
1993), and as such, achievement is a valuable inferential proxy for
positive beliefs about an individual. However, the decision to
include achievement-oriented items may limit the use of our in-
strument to predominantly Western, individualist cultures, which
value achievement orientation (see Kanungo & Bhatnagar, 1978;
Niles, 1998). Cross-cultural investigation of the proposed model
may require a revised cognitive evaluation measure that is not
weighted by achievement items. Having said that, we contend that
though the cues to quality to which one attends and the valence of
evaluation that results from specific cues may vary as a function of
culture, the primary link between the cognitive evaluation, what-
ever its source, and interpersonal attraction should be universal.
The current project provides evidence that one’s cognitive eval-
uation of the quality of an individual (a) is distinct from interper-
sonal attraction, (b) mediates the similarity effect, and (c) is
determined by the valence of information one infers from attitudes
attributed to a partner. These findings offer a cognitive interpre-
tation of the similarity effect and speak to a general model of
attraction: Attraction is based on an evaluation of the quality of an
individual. The quality of the individual, in turn, may be evidenced
by a variety of cues: similar attitudes (as demonstrated in Studies
1 and 2), possession of positive qualities (as demonstrated in Study
3), occupational success (Byrne, Clore, & Worchel, 1966), and
physical attractiveness (Sloman & Sloman, 1988; Walster et al.,
1966). Finally, Study 3 suggests that affective and self-relevant
considerations (i.e., threats to self-esteem) combine with cogni-
tively derived assessments (i.e., of the target as superior to the self)
to impact attraction.
Ajzen, I. (1974). Effects of information on interpersonal attraction: Simi-
larity versus affective value. Journal of Personality and Social Psychol-
ogy, 29, 374–380.
Ajzen, I., & Fishbein, M. (1977). Attitude–behavior relations: A theoret-
ical analysis and review of empirical research. Psychological Bulletin,
Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317–332.
Alicke, M. D. (1985). Global self-evaluation as determined by the desir-
ability and controllability of trait adjectives. Journal of Personality and
Social Psychology, 49, 1621–1630.
Amabile, T. M. (1983). Brilliant but cruel: Perceptions of negative evalu-
ators. Journal of Experimental Social Psychology, 19, 146–156.
Anderson, N. H. (1971). Integration theory and attitude change. Psycho-
logical Review, 78, 171–206.
Aron, A., Paris, M., & Aron, E. N. (1995). Falling in love: Prospective
studies of self-concept change. Journal of Personality and Social Psy-
chology, 69, 1102–1112.
Aronson, E. (1998). The social animal. Basic Books: New York.
Aronson, E., Willerman, B., & Floyd, J. (1966). The effect of a pratfall on
COGNITIVE EVALUATION AND ATTRACTION
increasing interpersonal attractiveness. Psychonomic Science, 4, 227–
Austrin, H. (1965). Cross validation of an attitude scale for the identifica-
tion of high and low academic achievers. Journal of Educational Re-
search, 58, 426–428.
Banaji, M. R. (2001). Implicit attitudes can be measured. In H. L. Roediger
& J. S. Nairne (Eds.), The nature of remembering: Essays in honor of
Robert G. Crowder (pp. 117–150). Washington, DC: American Psycho-
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
Bond, M. H., Byrne, D., & Diamond, M. J. (1968). Effect of occupational
prestige and attitude similarity on attraction as a function of assumed
similarity of attitudes. Psychological Reports, 23, 1167–1172.
Breckler, S. J. (1984). Empirical validation of affect, cognition, and be-
havior as distinct components of attitude. Journal of Personality and
Social Psychology, 47, 1191–1205.
Breckler, S. J., & Wiggins, E. C. (1989). Affect and evaluation in the
structure of attitudes. Journal of Experimental Social Psychology, 25,
Buss, D. M., Abbott, M., Angleitner, A., Asherian, A., Biaggio, A.,
Blanco-Villasenor, A., et al. (1990). International preferences in select-
ing mates. Journal of Cross-Cultural Psychology, 21, 5–47.
Buss, D. M., & Barnes, M. F. (1986). Preferences in human mate selection.
Journal of Personality and Social Psychology, 70, 559–570.
Byrne, D. (1971). The attraction paradigm. New York: Academic Press.
Byrne, D. (1992). The transition from controlled laboratory experimenta-
tion to less controlled settings: Surprise! Additional variables are oper-
ative. Communication Monographs, 59, 190–198.
Byrne, D. (1997). An overview (and underview) of research and theory
within the attraction paradigm. Journal of Social and Personal Rela-
tionships, 14, 417–431.
Byrne, D., & Blaylock, B. (1963). Similarity and assumed similarity of
attitudes between husbands and wives. Journal of Abnormal and Social
Psychology, 67, 636–640.
Byrne, D., & Clore, G. L. (1967). Effectance arousal and attraction.
Journal of Personality and Social Psychology, 6, 1–18.
Byrne, D., & Clore, G. L. (1970). A reinforcement model of evaluative
responses. Personality: An International Journal, 1, 103–128.
Byrne, D., Clore, G. L., & Worchel, P. (1966). Effect of economic
similarity–dissimilarity on interpersonal attraction. Journal of Person-
ality and Social Psychology, 4, 220–224.
Byrne, D., & Griffitt, W. (1966). Similarity versus liking: A clarification.
Psychonomic Science, 6, 295–296.
Byrne, D., London, O., & Griffitt, W. (1968). The effect of topic impor-
tance and attitude similarity–dissimilarity on attraction in the intra-
stranger design. Psychonomic Science, 11, 303–304.
Byrne, D., Rasche, L., & Kelley, K. (1974). When “I like you” indicates
disagreement: An experimental differentiation of information and affect.
Journal of Research in Personality, 8, 207–217.
Byrne, D., & Wong, T. J. (1962). Racial prejudice, interpersonal attraction,
and assumed dissimilarity of attitudes. Journal of Abnormal and Social
Psychology, 65, 246–253.
Carli, L. L., Ganley, R., & Pierce-Otay, A. (1991). Similarity and satis-
faction in roommate relationships. Personality and Social Psychology
Bulletin, 17, 419–426.
Clore, G. L., & Baldridge, B. (1968). Interpersonal attraction: The role of
agreement and topic interest. Journal of Personality and Social Psychol-
ogy, 9, 340–346.
Clore, G. L., & Gormly, J. B. (1974). Knowing, feeling, and liking: A
psychophysiological study of attraction. Journal of Research in Person-
ality, 8, 218–230.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
Darwin, C. (1859). The origin of species by means of natural selection.
London: John Murray.
Davidson, A. R., & Jaccard, J. J. (1979). Variables that moderate the
attitude–behavior relation: Results of a longitudinal survey. Journal of
Personality and Social Psychology, 37, 1364–1376.
DeNike, D. L., & Leibovitz, M. P. (1969). Accurate anticipation of rein-
forcement in verbal conditioning. Journal of Personality, 37, 158–170.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and
controlled components. Journal of Personality and Social Psychology,
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Orlando,
FL: Harcourt Brace.
Eagly, A. H., & Chaiken, S. (1998). Attitude structure and function. In
D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social
psychology (Vol. 2, pp. 269–322). New York: McGraw-Hill.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986).
On the automatic activation of attitudes. Journal of Personality and
Social Psychology, 50, 229–238.
Fishbein, M., & Ajzen, I. (1972). Attitudes and opinions. Annual Review of
Psychology, 23, 487–544.
Foa, U. G., & Foa, E. B. (1974). Societal structures of the mind. Oxford,
England: Charles Thomas.
Fromkin, H. L. (1972). Feelings of interpersonal undistinctiveness: An
unpleasant affective state. Journal of Experimental Research in Person-
ality, 6, 178–185.
Gaynor, C. A. (1971). A developmental investigation of the law of attrac-
tion. Unpublished doctoral dissertation, University of Oklahoma, Nor-
Gillis, J. C., & Avis, W. E. (1980). The male taller norm in mate selection.
Personality and Social Psychology Bulletin, 6, 396–401.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring
individual differences in implicit cognition: The Implicit Association
Test. Journal of Personality and Social Psychology, 74, 1464–1480.
Gruenfeld, L., & Arbuthnot, J. (1969). Field independence as a conceptual
framework for prediction of variability in ratings of others. Perceptual
and Motor Skills, 28, 31–44.
Harackiewicz, J. M., Abrahams, S., & Wageman, R. (1987). Performance
evaluation and intrinsic motivation: The effects of evaluative focus,
rewards, and achievement orientation. Journal of Personality and Social
Psychology, 53, 1015–1023.
Helmreich, R., Aronson, E., & LeFan, J. (1970). To err is humanizing
sometimes: Effects of self-esteem, competence, and a pratfall on inter-
personal attraction. Journal of Personality and Social Psychology, 16,
Herbst, K. C., Gaertner, L., & Insko, C. A. (2003). My head says yes but
my heart says no: Cognitive and affective attraction as a function of
similarity to ideal self. Journal of Personality and Social Psychology,
Hill, C. T., Rubin, Z., & Peplau, L. A. (1976). Breakups before marriage:
The end of 103 affairs. Journal of Social Issues, 32, 147–168.
Horton, R. S., & Montoya, R. M. (2004). The relationship between implied
information and interpersonal attraction. Manuscript submitted for pub-
Howard, J. A., Blumstein, P., & Schwartz, P. (1987). Social or evolution-
ary theories? Some observations on preferences in human mate selec-
tion. Journal of Personality and Social Psychology, 53, 194–200.
Hoyle, R. H., Pinkley, R. L., & Insko, C. A. (1989). Perceptions of social
behavior: Evidence of differing expectations for interpersonal and inter-
group interaction. Personality and Social Psychology Bulletin, 15, 365–
Insko, C. A., & Cialdini, R. B. (1969). A test of three interpretations of
attitudinal verbal reinforcement. Journal of Personality and Social Psy-
chology, 12, 333–341.
MONTOYA AND HORTON
Jo¨reskog, K. G., & So¨rbom, D. (2001). LISREL 8.50 [Computer software].
Chicago: Scientific Software International.
Judd, C. M., & Kenny, D. A. (1981). Process analysis: Estimating medi-
ation in treatment evaluations. Evaluation Review, 5, 602–619.
Kanungo, R. N., & Bhatnagar, J. K. (1978). Achievement orientation and
occupational values: A comparative study of young French and English
Canadians. Canadian Journal of Behavioural Science, 10, 202–213.
Kaplan, M. F., & Anderson, N. H. (1973). Information integration theory
and reinforcement theory as approaches to interpersonal attraction. Jour-
nal of Personality and Social Psychology, 28, 301–312.
Katz, D., & Stotland, E. (1959). A preliminary statement to a theory of
attitude structure and change. In S. Koch (Ed.), Psychology: A study of
a science (Vol. 3, pp. 423–475). New York: McGraw-Hill.
Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social
psychology. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The
handbook of social psychology (pp. 233–265). Boston: McGraw-Hill.
Kenrick, D. T., Montello, D. R., Gutierres, S. E., & Trost, M. R. (1993).
Effects of physical attractiveness on affect and perceptual judgments:
When social comparison overrides social reinforcement. Personality and
Social Psychology Bulletin, 19, 195–199.
Kerr, N. L., & Kaufman-Gilliland, C. M. (1994). Communication, com-
mitment, and cooperation in social dilemmas. Journal of Personality and
Social Psychology, 66, 513–529.
Lazarus, R. S. (1982). Thoughts on the relations between emotion and
cognition. American Psychologist, 37, 1019–1024.
Lazarus, R. S. (1984). On the primacy of cognition. American Psycholo-
gist, 39, 124–129.
Leahy, R. L. (1979). Development of conceptions of prosocial behavior:
Information affecting rewards given for altruism and kindness. Devel-
opmental Psychology, 15, 34–37.
Lott, A. J., & Lott, B. E. (1969). Liked and disliked persons as reinforcing
stimuli. Journal of Personality and Social Psychology, 11, 129–137.
Lydon, J. E., Jamieson, D. W., & Zanna, M. P. (1988). Interpersonal
similarity and the social and intellectual dimensions of first impressions.
Social Cognition, 6, 269–286.
McLaughlin, B. (1970). Similarity, recall, and appraisal of others. Journal
of Personality, 38, 106–116.
Miller, M. G., & Tesser, A. (1986). Effect of affective and cognitive focus
on the attitude–behavior relation. Journal of Personality and Social
Psychology, 51, 270–276.
Molm, L. D., Takahashi, N., & Peterson, G. (2000). Risk and trust in social
exchange: An experimental test of a classic proposition. American
Journal of Sociology, 105, 1396–1427.
Montoya, R. M., Horton, R. S., & Kirchner, J. (2004). Does similarity in
relationships really cause attraction? A quantitative research synthesis
of the similarity effect. Manuscript submitted for publication.
Mowaiye, O. (1993). Secondary schoolgirls’ perception of women’s roles
and their achievement orientation. School Psychology International, 14,
Murstein, B. I., & Beck, G. D. (1972). Person perception, marriage adjust-
ment, and social desirability. Journal of Consulting and Clinical Psy-
chology, 39, 396–403.
Niles, S. (1998). Achievement goals and means: A cultural comparison.
Journal of Cross-Cultural Psychology, 29, 656–667.
Norman, R. (1975). Affective–cognitive consistency, attitudes, confor-
mity, and behavior. Journal of Personality and Social Psychology, 32,
Nosek, B. A., Banaji, M. R., & Greenwald, A. G. (2002). Math male,
me female, therefore math me. Journal of Personality and Social
Psychology, 83, 44–59.
Novak, D. W., & Lerner, M. J. (1968). Rejection as a consequence of
perceived similarity. Journal of Personality and Social Psychology, 9,
Oliver, R. L. (1993). Cognitive, affective, and attribute bases of the
satisfaction response. Journal of Consumer Research, 20, 418–435.
Orne, M. T. (1962). On the social psychology of the psychological exper-
iment: With particular reference to demand characteristics and their
implications. American Psychologist, 17, 776–783.
Orne, M. T., & Scheibe, K. E. (1964). The contribution of nondeprivation
factors in the production of sensory deprivation effects: The psychology
of the “panic button.” Journal of Abnormal and Social Psychology, 68,
Peterson, J., & Miller, L. (1980). Physical attractiveness and marriage
adjustment in older American couples. Journal of Psychology, 105,
Pilkington, N. W., & Lydon, J. E. (1997). The relative effect of attitude
similarity and attitude dissimilarity on interpersonal attraction: Investi-
gating the moderating roles of prejudice and group membership. Per-
sonality and Social Psychology Bulletin, 23, 107–122.
Rogers, M. T. (1999). Marital satisfaction as it relates to similarity versus
complementarity in personality dimensions (couples). (Doctoral disser-
tation, Miami Institute of Psychology of the Caribbean Center for
Advanced Studies, 1999). Dissertation Abstracts International: Section
B: The Sciences and Engineering, 60(4-B), 1917.
Rosenberg, M. J., Hovland, C. I., McGuire, W. J., Abelson, R. P., &
Brehm, J. W. (1960). Attitude organization and change: An analysis of
consistency among attitude components: Vol. 3. Yale studies in attitude
and communication. Oxford, England: Yale University Press.
Russell, R. J., & Wells, P. A. (1991). Personality similarity and quality of
marriage. Personality and Individual Differences, 12, 407–412.
Sachs, D. H., & Byrne, D. (1970). Differential conditioning of evaluative
responses to neutral stimuli through association with attitude statements.
Journal of Experimental Research in Personality, 4, 181–185.
Schwartz, S. H. (1978). Temporal instability as a moderator of the attitude–
behavior relationship. Journal of Personality and Social Psychology, 36,
Sears, D. O., & Henry, P. J. (2003). The origins of symbolic racism.
Journal of Personality and Social Psychology, 85, 259–275.
Sedikides, C. (1993). Assessment, enhancement, and verification determi-
nants of the self-evaluation process. Journal of Personality and Social
Psychology, 65, 317–338.
Sigall, H., & Landy, D. (1973). Radiating beauty: Effects of having a
physically attractive partner on person perception. Journal of Personal-
ity and Social Psychology, 28, 218–224.
Singh, R. (1973). Affective implications of the weighting coefficient in
attraction research. (Doctoral dissertation, Purdue University, 1973).
Dissertation Abstracts International, 34, 3545–3546.
Sloman, S., & Sloman, L. (1988). Mate selection in the service of human
evolution. Journal of Social Biological Structures, 11, 457–468.
Spuhler, J. N. (1968). Assortative mating with respect to physical charac-
teristics. Eugenics Quarterly, 17, 128–140.
Stallings, R. B. (1970). Personality similarity and evaluative meaning as
conditioners of attraction. Journal of Personality and Social Psychology,
Stapel, D. A., & Tesser, A. (2001). Self-activation increases social com-
parison. Journal of Personality and Social Psychology, 81, 742–750.
Steele, M. E., & McGlynn, R. P. (1979). Effects of affective value and
similarity on attraction. Journal of Social Psychology, 108, 111–119.
Stevens, G., Owens, D., & Schaefer, E. C. (1990). Education and attrac-
tiveness in marriage choices. Social Psychology Quarterly, 55, 62–70.
Sunnafrank, M. (1992). On debunking the attitude similarity myth. Com-
munication Monographs, 59, 164–179.
Susanne, C., & Lepage, Y. (1988). Assortative mating for anthropometric
characters. In C. G. N. Mascie-Taylor & A. J. Boyce (Eds.), Human
mating patterns (pp. 83–99). Cambridge, England: Cambridge Univer-
COGNITIVE EVALUATION AND ATTRACTION
Symons, D. (1979). The evolution of human sexuality. New York: Oxford
Tan, D. T., & Singh, R. (1995). Attitudes and attraction: A developmental
study of the similarity–attraction and dissimilarity–repulsion hypothe-
ses. Personality and Social Psychology Bulletin, 21, 975–986.
Tesser, A. (2000). On the confluence of self-esteem maintenance mecha-
nisms. Personality and Social Psychology Review, 4, 290–299.
Uleman, J. S. (1971). Awareness and motivation in generalized verbal
conditioning. Journal of Experimental Research in Personality, 5, 257–267.
Walster, E., Aronson, V., Abrahams, D., & Rottmann, L. (1966). Impor-
tance of physical attractiveness in dating behavior. Journal of Person-
ality and Social Psychology, 4, 508–516.
Wetzel, C. G., & Insko, C. A. (1982). The similarity–attraction relation-
ship: Is there an ideal one? Journal of Experimental Social Psychology,
Woodmansee, J. L., & Cook, S. W. (1967). Dimensions of verbal racial
attitudes: Their identification and measurement. Journal of Personality
and Social Psychology, 7, 240–250.
Questionnaires Used in the Studies
Interpersonal Attraction Questionnaire
1. I would like to meet my future interaction partner.
2. To what extent do you want to work on the upcoming task with your
3. I would probably dislike talking with my future interaction partner at
a party. (reverse scored)
4. I would enjoy discussing controversial topics with my future interac-
5. My future interaction partner would probably not make a good friend
to me. (reverse scored)
6. How much do you think you will like your partner?
7. I would like to get to know this person better.
8. I think I would enjoy my future interaction partner’s company.
9. To what extent are you looking forward to meeting your partner?
Cognitive Evaluation Questionnaire
1. My future interaction partner is probably well-respected.
2. My future interaction partner is probably good at everything that s/he
3. In general, how good a person do you think your partner is?
4. My future interaction partner will probably be successful in life.
5. My future interaction partner probably achieves all of his/her goals.
6. My future interaction partner could help me accomplish my goals.
7. I think that my future interaction partner would make a good leader.
Received March 25, 2003
Revision received October 31, 2003
Accepted November 5, 2003 䡲
Wanted: Old APA Journals!
APA is continuing its efforts to digitize older journal issues for the PsycARTICLES database.
Thanks to many generous donors, we have made great strides, but we still need many issues,
particularly those published in the 1950s and earlier.
If you have a collection of older journals and are interested in making a donation, please e-mail
firstname.lastname@example.org or visit http://www.apa.org/journals/donations.html for an up-to-date list of the
issues we are seeking.
MONTOYA AND HORTON