Fulltext

University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln
Management Department Faculty Publications Management Department
2010
Perceiver Effects as Projective Tests: What Your Perceptions of

Others Say about You
Dustin Wood
Wake Forest University, [email protected]
Peter D. Harms
University of Nebraska - Lincoln, [email protected]
Simine Vazire
Washington University in St Louis
Follow this and additional works at: https://digitalcommons.unl.edu/managementfacpub
Part of the Management Sciences and Quantitative Methods Commons
Wood, Dustin; Harms, Peter D.; and Vazire, Simine, "Perceiver Effects as Projective Tests: What Your
Perceptions of Others Say about You" (2010). Management Department Faculty Publications. 78.
https://digitalcommons.unl.edu/managementfacpub/78
This Article is brought to you for free and open access by the Management Department at
DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Management Department
Faculty Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln.
Published in Journal of Personality and Social Psychology (2010) 99(1). Copyright 2010, American Psychological Association. Used by permission.
DOI: 10.1037/a0019390. “This article may not exactly replicate the final version published in the Journal of Personality and Social Psychology. It is
not the copy of record.”
Perceiver Effects as Projective Tests: What Your Perceptions of Others Say about You
Dustin Wood, Department of Psychology, Wake Forest University, [email protected]
Peter Harms, College of Business Administration, University of Nebraska-Lincoln, [email protected]
Simine Vazire, Department of Psychology, Washington University in St. Louis
In three studies, we document various properties of perceiver effects—or how an individual generally tends to describe other peo-
ple in a population. First, we document that perceiver effects have consistent relationships with dispositional characteristics of
the perceiver, ranging from self-reported personality traits and academic performance to well-being and measures of personal-
ity disorders, to how liked the person is by peers. Second, we document that the covariation in perceiver effects among trait di-
mensions can be adequately captured by a single factor consisting of how positively others are seen across a wide range of traits
(e.g., how nice, interesting, trustworthy, happy, and stable others are generally seen). Third, we estimate the one-year stability
of perceiver effects and show that individual differences in the typical perception of others have a level of stability comparable
to that of personality traits. The results provide compelling evidence that how individuals generally perceive others is a stable
individual difference that reveals much about the perceiver’s own personality.
Keywords: Person perception; Perceiver effect; Perceiver biases; Rating bias

We do not see the world as it is, we see the world as we are. et al., 2004). Similarly, cognitive accounts of disagreeable or
—The Talmud antisocial behavior suggest that these behavior patterns may
originate in part from perceptual tendencies to more readi-
We are what we think. All that we are arises with our ly perceive others as hostile and threatening, which prepares
thoughts. With our thoughts, we make our world. individuals to respond aggressively (Dodge & Crick, 1990;
—The Buddha Raine, 2008). In short, many behavioral patterns common-
ly studied by psychologists are thought to be caused in part
Given that observer reports are usually collected to learn by how individuals perceive others in their environments,
about the target rather than the person providing the rating, which in turn shapes the behavioral options that individuals
tendencies for raters to judge the same target differently are see as desirable, adaptive, or appropriate (Campbell, Mill-
generally regarded as perceiver bias, or scale-use bias, and er, Lubetsky, & O’Connell, 1964). When considered in this
considered nuisance variance to be removed or minimized way, individual differences in how raters tend to rate real
to the extent possible (Hoyt, 2000). However, we can also in- targets on personality scales can be considered to be ecologi-
vert the usual use of observer reports and see what observ- cally valid indicators of individuals’ typical construal of oth-
er reports reveal about the raters. Indeed, numerous theorists ers, which in turn is likely to be a major cause of their behav-
have suggested that raters’ perceptions of others are one of ior. Despite this, there have been few or relatively limited ex-
the most important determinants of their behavior (e.g., Beck, aminations of how perceiver effects relate to other disposi-
Freeman, & Davis, 2004; Kelly, 1963; Laing, 1967; Mischel & tional characteristics.
Shoda, 1995; Reis, 2008). Indeed, perceiver effects, which re- A limitation of previous studies linking personality to per-
fers to general tendencies to perceive or judge others in par- ceiver effects is that these investigations tended to assess
ticular ways (Kenny, 1994), can be thought to be a core com- perceptions of others only through statement-based instru-
ponent of a number of constructs commonly studied by psy- ments such as the Personality Belief Questionnaire (Beck et
chologists. For instance, Machiavellianism is usually mea- al., 2001) or through lab scenarios in which participants rat-
sured in part by asking individuals the extent to which they ed fictional targets (e.g., Lakey & Cassady, 1990). Few stud-
perceive a lack of sincerity, integrity, or selflessness in others’ ies have investigated how a broad range of personality and
actions (Christie & Geis, 1970), and narcissistic behavior is dispositional variables are associated with how individuals
thought to be prompted in part by a belief that other people perceive real others in their social environments. Our goal in
are inferior, uninteresting, and unworthy of attention (Beck the present research is thus to conduct an in-depth investiga-
tion of the nature of perceiver effects. We focus on three ma-
jor issues. First, we explore whether perceiver effects are as-
We would like to thank Andrew Beer and Alecia Santuzzi
sociated with dispositional variables beyond “assumed sim-
for their helpful comments on drafts of this article; Richard
ilarity” effects (Cronbach, 1955). Second, we explore wheth-
Gonzalez for helpful statistical advice relevant to analyses in
er the major dimensions of individual differences in perceiv-
the article; and Daniel Bureau, Shannon Stark, and Kristof
er effects are comparable to the major dimensions of individ-
Supinski for their invaluable help in the data collection for
ual differences found in self-judgments (e.g., the Big Five).
this project.
Third, we explore the stability of perceiver effects over the
Manuscript received October 4, 2009; revision received Jan- period of a year. We outline each of these goals in the follow-
uary 27, 2010; accepted March 3, 2010. ing three sections.
Perceiver Effects as Projective Tests 175
Beyond Assumed Similarity: The Dispositional Correlates of of which may match or even exceed the magnitude of as-
Perceptions of Others sumed similarity correlations.
One of the most common ways that investigators have at- Many normal and abnormal personality traits are thought
tempted to understand the dispositional correlates of per- to be characterized by systematic biases in the perception of
ceiver effects is to correlate participants’ perceiver effect esti- others, and these do not always reflect assumed similarity re-
mates with participants’ own self-judgments on the same di- lationships. For instance, although narcissists may perceive
mension (e.g., Kenny, 1994; Srivastava, Guglielmo, & Beer, others as being uninteresting or worthless, this may not re-
2010). In other words, researchers have examined whether flect how they see themselves (American Psychiatric Associ-
people tend to see others the same way they see themselves. ation, 1994). Similarly, individuals displaying behaviors typ-
These correlations are usually referred to as assumed similarity ical of paranoid personality disorder may believe that others
bias (Cronbach, 1955), although they have been referred to by are malevolent and untrustworthy, even though they may
numerous other names, including projection bias and the self- not see themselves that way (American Psychiatric Associ-
based heuristic (Hoch, 1987; Ready, Clark, Watson, & Wester- ation, 1994). In addition to exploring divergent correlations
house, 2000). The initial goal in measuring assumed similar- between perceiver effects and self-perceptions of the Big
ity has been to examine its influence on accuracy in interper- Five, we thus also cast a wider net to investigate the relation-
sonal perception, rather than as an interesting phenomenon ships between perceiver effects and dispositional variables
in itself. As first noted by Cronbach (1955), an individual may beyond the Big Five. Given the importance of individual dif-
achieve accuracy in judging a target’s personality if (a) the ferences in the perception of others in some theories of per-
target is actually similar to the rater and (b) the rater simply sonality disorders (Beck et al., 2004), we explore the relation-
describes his or her own personality. If these conditions both ships between perceiver effects and measures of personality
exist, the individual can achieve accurate judgments without disorders. We also explore the relationship between perceiv-
attending to any information about the target. er effects and measures of well-being and of the quality of so-
Despite the limited utility of this explanation for under- cial relationships.
standing the accuracy of personality judgments (due to the
limited actual personality similarity of raters and targets in Major Dimensions of Perceiver Effects
most circumstances), the frequent tendency to examine per- Most of the research on how individuals tend to rate others
ceiver effects through the lens of assumed similarity seems has implicitly assumed that the structure of perceiver effects
to have led to a general tendency to report “convergent” cor- does not differ much from the structure of self-judgments.
relations between perceiver effects and self-reported person- That is to say, if the Big Five is regarded as an adequate struc-
ality traits (e.g., do agreeable individuals also see others as ture of the major dimensions of self-reports or single-peer
agreeable?) but not “divergent” correlations (e.g., do agree- ratings, then it is assumed to also be an adequate structure
able individuals see others as more intelligent?). However, for perceiver effects. However, as detailed by Kenny (1994),
there are reasons to suspect that these divergent correlations single-personality judgments—whether of oneself or of an-
exist. Early discussions of the perceiver effect considered the other person—can be viewed as the composite of three ma-
possibility of “complementary projection,” whereby people jor effects: (a) how the rater tends to rate others in the popu-
act in ways that complement the way they tend to see others lation in general (perceiver effects), (b) how the target tends to
(e.g., acting dominant because they believe others are sub- be rated by others in the population in general (target effects),
missive; Campbell et al., 1964). Further, as noted by Beer and and (c) how the rater tends to uniquely judge particular tar-
Watson (2008a), there are indications that individual differ- gets in ways that cannot be accounted for by these other ef-
ences in perceiver effects may have a simpler dimensionali- fects (relationship effects; in self-judgments this is how the rat-
ty than self-perceptions. Consistent with this, there are indi- er uniquely views him- or herself), where population here re-
cations that women tend to rate others more positively than fers to the people who can rate others or be rated by others.
do men across all Big Five dimensions simultaneously (Win- Although the Big Five and similar taxonomies might ade-
quist, Mohr, & Kenny, 1998), which would be unlikely to oc- quately describe the major dimensions of the combination of
cur if perceiver effects for all Big Five dimensions were or- these three effects that emerges in self and other ratings, they
thogonal. may not capture the major dimensions of each effect consid-
If perceiver effects have a simpler structure than the self- ered separately.
ratings or single-peer judgments that were used to uncover Little research has investigated the dimensionality of per-
the Big Five, we should not expect the relationships between ceiver effects. There are reasons to think that the structure of
self-ratings and other-perceptions to be trait-specific, as is perceiver effects can be adequately described by fewer fac-
implied by the assumed similarity hypothesis. For instance, tors than self-reported personality traits. This may happen
there are indications that agreeable individuals describe oth- because perceiver effects can be thought of as individual dif-
ers more positively on a range of traits beyond simply agree- ferences in how people perceive the same “generalized oth-
ableness; agreeableness has been related to lower levels of er.” That is, we can think of the comparison of how differ-
prejudice toward a wide range of groups (Graziano, Bruce, ent individuals rate many targets on average as indicating
Sheese, & Tobin, 2007) and to both the perceived agreeable- how the individuals differ in describing the same “average
ness and extraversion of others (Graziano & Tobin, 2002; oth- person” in the population they are rating. Some research has
er Big Five traits were not measured). We suspect that there suggested that when all raters are judging the same target
are many divergent or nonparallel relationships that may ex- (e.g., a particular political candidate) instead of different tar-
ist between self-ratings and perceiver effects (e.g., self-rated gets (e.g., rating themselves, a friend, or a spouse), the factor
extraversion and perceived agreeableness of others), many structure of trait ratings may become simpler and less differ-
176 Wood, Harms & Vazire in Journal of Personality and Social Psychology (2010) 99(1)
entiated (Caprara, Barbaranelli, Fraley, & Vecchione, 2007). the stability of perceiver effects over a year (Study 3). We
This may happen in part because individual differences in conclude the article with a discussion of perceiver effects as
ratings in such a situation are no longer influenced by differ- individual difference variables.
ences in the target being rated. This reasoning, and the like-
lihood that many individual difference characteristics (such Study 1: Beyond Assumed Similarity
as gender and agreeableness) correlate with perceiver esti- Our first goal was to explore how variation in perceptions
mates of multiple traits simultaneously, lead us to suspect of others is associated with the perceiver’s own personality
that there may be fewer major dimensions in the general per- traits. The relationships between how individuals judge oth-
ception of others than in self-perceptions. Similarly, a recent ers and how they judge themselves have been investigated
investigation by Srivastava and colleagues (2010) found that several times; however, this is most frequently done to doc-
although covariation between perceiver effect estimates cre- ument “assumed similarity” (e.g., does perceiving oneself as
ated by averaging scores a rater provided of different targets extraverted correlate with perceiving others as extraverted?).
could be adequately explained by the five-factor structure These studies have tended to show that people judge oth-
generally found in self-ratings or ratings from knowledge- ers in a similar manner to how they judge themselves, al-
able informants (Goldberg, 1993), the model fitted better though the strength of assumed similarity effects varies by
when these factors were allowed to covary through a com- trait. Within the Big Five framework, the strongest effects are
mon factor of overall positive evaluation (i.e., whether oth- generally observed for agreeableness, and the weakest for
ers were seen as generally having positive vs. negative qual- extraversion (e.g., Beer & Watson, 2008b; Kenny, 1994). Giv-
ities). Accordingly, we attempt to identify the number of fac- en that researchers often limit their analysis to convergent
tors necessary to explain the dimensionality of perceiver ef- correlations between self-perceptions and perceiver effects
fects through the use of factor analysis. (i.e., assumed similarity correlations), we examine the diver-
gent correlations, which explore whether perceptions of oth-
Stability of the Perceptions of Others ers correlate not just with self-judgments of the same trait but
Another topic that has received little attention in the liter- with self-judgments of different traits. Documenting diver-
ature concerns the stability of individual differences in per- gent correlations between how individuals see themselves
ceiver effects over time. Malloy, Sugarman, Montvilo, and and how they see others would clearly indicate that there
Ben-Zeev (1995) reported low levels of one-year stability in is more to the perceiver effects than simply the projection of
perceiver effect estimates among elementary-school-aged one’s own self-image onto others.
children (rs ≈ 0.20), whereas target effects showed higher sta-
bilities over the same intervals. Similarly, Srivastava and col- Method
leagues (2010) found stabilities of perceiver effect estimates Participants. A total of 165 undergraduate students (100 fe-
to be approximately 0.50 over the period of a week. Howev- male) from a large state university in Texas participated in
er, the small amount of research on the topic leaves the sta- exchange for either course credit or a cash reward ($10 and a
bility of perceiver effects an open question. 1 in 10 chance to win $100 more). Participants were asked to
The stability of an individual’s tendency to see people in cer- sign up in groups of five friends who were all previously ac-
tain ways is important for understanding the extent to which quainted. Some groups included dyads that had not met be-
an individual’s general perceptions of others are “traitlike.” fore, but 97% of the dyads were previously acquainted. On
Further, many theorists are interested in perceptual tenden- average, participants had known their group members for
cies as causal variables, noting that individuals’ perceptions over three years (M = 3.12, SD = 4.29). The students ranged
of their environment are a proximal source of their behav- in age from 18 to 22 years (M = 18.8, SD = 1.7). According to
ior, in that people often act on how they perceive their en- self-reported ethnicity, 70 participants were Asian or Pacific
vironments even when their perceptions do not reflect the Islander, 65 were Caucasian, 19 were Hispanic, 9 were Afri-
objective environment (Beck et al., 2004; Kelly, 1963; Reis, can American, one was of mixed ethnicity, and one did not
2008). Consequently, if individual differences in the gener- provide an ethnicity.1
al perceptions of others are relatively stable over time, this Procedure and materials. Participants were recruited by
should increase individuals’ ability to direct and maintain posting flyers in dorms, making announcements in introduc-
their patterns of behavior over time, as the consistency of an tory psychology classes, and handing out candy and flyers at
environment serves to increase the potential for the environ- busy campus intersections. Participants signed up by visiting
ment to direct lasting dispositional change (e.g., Caspi, Her- a website and completing a form, which required five names
bener, & Ozer, 1992; Roberts, Wood, & Caspi, 2008). Conse- of previously acquainted friends who were all undergradu-
quently, empirically estimating the stability of perceiver ef- ate students and at least 18 years of age. No details about the
fects should provide a better idea of the extent to which per- purpose of the study were given in the advertisements or on
ceiver dimensions reflect stable ways people construe their the website. Participants signed up their group for a specific
interpersonal environments. 3-hr session. All sessions were held on Saturday and Sunday
We address these questions in three studies. First, we ex- afternoons. After signing up, participants received an e-mail
plore how perceiver effects of Big Five traits are related to with directions to the laboratory.
self-reported personality traits and other dispositional vari-
ables (Studies 1, 2, and 3). Then, using data sets with a broad-
er range of peer-rated trait adjectives, we conduct analyses 1
Other data from this sample have been used in Vazire (2010) and
in which we explore the major dimensions of covariation Back et al. (in press). The analyses from these other investigations do
among perceiver effects (Studies 2 and 3). Finally, we explore not overlap with the current analyses.
Upon arriving at the laboratory, each “friend group” was tween perceiver effects of different traits will result in a ma-
shown to a room where an experimenter described the study trix in which different traits are linked not just through per-
and administered the consent forms. Participants then com- ceiver effects but also through target effects and relationship
pleted a battery of measures including round-robin ratings effects due to the fact that characteristics of the same targets
of their group members, including themselves. Group mem- being rated represent a significant percentage of the variance
bers were seated together at a table, but folders were put up in the perceiver effect estimates that are created in this man-
so that they could not see each other’s ratings. ner, especially when the number of ratings being averaged is
Participants rated each group member on a 40-item inven- relatively small. To investigate how perceiver effects for dif-
tory. The first 10 items consisted of the Ten Item Personali- ferent traits are related to one another, we thus created a cor-
ty Inventory (Gosling, Rentfrow, & Swann, 2003), which in relation matrix consisting solely of correlations relating judg-
past research has shown high convergent validity with oth- ments a rater made of one target to judgments the same rat-
er widely used Big Five inventories in self- and observer-re- er had made of a different target (i.e., how do a rater’s im-
ports and has had very good test-retest reliability (mean r pressions of target i correlate with the rater’s impressions of
= 0.72 over 6 weeks; Gosling et al., 2003). Participants rated a different target i´?). If raters are randomly assigned to tar-
each group member on the same 15-point Likert-type scale, gets, the only characteristics linking a rater’s judgments of
ranging from 1 (Not at all) to 15 (Extremely). Participants two different targets are tendencies for raters to judge targets
wrote the letter corresponding to the group member they in the sample in particular ways (i.e., perceiver effects). Con-
were rating above the number they chose for that person and sequently, this perceiver-effect matrix can be used to examine
were told not to use the same number twice on a single scale how perceiver judgments are related to one another along
(i.e., participants had to give each group member different both same and different dimensions simultaneously (e.g., if
scores from each other on a given item). The remaining 30 a judge rates one target as extraverted, how likely is the same
items were unrelated to the analyses conducted here. judge to rate a different target as extraverted? or agreeable?).
Participants also completed a number of other measures This matrix was formed by creating all combinations of a rat-
(see Vazire, 2006, 2010, for a full description). Of relevance er’s judgments of two different targets and using the dou-
to the present analyses, participants completed the Beck De- ble-entry method described by Gonzalez and Griffin (2000;
pression Inventory (BDI; Beck, Rial, & Rickels, 1974; an item Griffin & Gonzalez, 1995) to make the correlations invariant
dealing with suicidality was omitted for ethical reasons; α = against the arbitrary assignment of the ratings into columns
0.68), the 16-item version of the Narcissistic Personality In- when estimating the correlations.
ventory (NPI; Ames, Rose, & Anderson, 2006; α = 0.70), and The diagonals of the matrix give the ICCs of trait perceiv-
the Wonderlic IQ test (Wonderlic, 1983). College GPA and er effects, which detail the extent to which ratings that an in-
SAT scores were also obtained from the registrar two years dividual provides of different individuals on the same trait
after the rating data were collected. correlate with one another. We refer to the off-diagonals as
the “cross-class” correlations of perceiver effects, which de-
Analyses tail the extent that ratings a rater has made of one target cor-
Modeling relationships between perceiver characteristics and per- relate with different trait ratings the rater has made of a dif-
ceiver effects. In the results that we report for this and the later ferent target (e.g., how much do an individual’s ratings of
studies, we report the correlations linking a perceiver’s dis- one target’s extraversion correlate with the individual’s rat-
positional characteristics (e.g., a rater’s gender, GPA, self-ratings of a different target’s agreeableness?). To our knowl-
ed extraversion) to how the perceiver sees a single other target edge, this is the first time cross-class correlations have been
on some dimension (e.g., extraversion). These correlations reported. These can be used to understand the dimensionali-
can be interpreted as measuring how personality differences ty of perceiver effects. The cross-class correlations should be
among perceivers on dimension X relate to how the perceiv- approximately zero if the perceiver effects for Big Five traits
ers judge a single target on dimension Y. Consequently, the are independent of one another, and at the other extreme
correlations we report are somewhat smaller than if multiple they should be approximately the same size as the ICCs if
ratings by the individual perceivers were aggregated. To es- the perceiver effects of different Big Five traits are expres-
timate how dispositional characteristics are associated with sions of a single perceiver dimension.
how individuals perceive others in the population in general,
we estimated a second correlation by dividing this correla- Results and Discussion
tion by the square root of the intraclass correlation (ICC) as- Intraclass and cross-class correlations of perceiver effects.
sociated with the perceiver effect (Malloy et al., 1995), which Using the perceiver-effect matrix described previously, we
we report in all studies. Note also that estimating this second were able to estimate the extent to which a rater’s perceptions
correlation is equivalent to a partial correction for unreliabil- of one target were associated with the rater’s perceptions of a
ity (i.e., the unreliable expression of perceiver effects in a per- different target; these are shown in the left half of the top sec-
ceiver’s judgment of a single target); it does not correct for tion of Table 1. The magnitude of the ICCs of perceiver ef-
unreliability on the side of the dispositional variable, such as fects differed substantially across the Big Five. The greatest
self-reported extraversion. ICCs were found for openness (r = 0.18), emotional stability (r
Modeling relations between perceiver dimensions. Perhaps the = 0.16), and agreeableness (r = 0.13). Substantially lower esti-
most intuitive way to measure the relation between perceiv- mates were found for conscientiousness (r = 0.04) and extra-
er effects for different traits is to average trait ratings that a version (r = -0.03). The results suggest that there were moder-
rater has collected across targets and correlate these averages ate individual differences in how people tended to perceive
together. However, this method for estimating relations be- targets on agreeableness, emotional stability, and openness
Table 1
Relationships Between Ratings of Others and Self-Ratings for Big Five Characteristics

r between ratings of a target (rows) and
r between ratings of two different targets self-ratings (columns)
Perceiver effect dimension 1 2 3 4 5 1 2 3 4 5

Study 1: Ratings of friends
1. Extraversion -0.03 -0.04 0.07 0.08* 0.03 0.03
2. Agreeableness 0.04 0.13 -0.01 0.19* -0.01 0.14* 0.00
3. Conscientiousness 0.05 0.04 0.04 0.09* 0.10* 0.09* 0.07 0.03
4. Emotional stability 0.00 0.16 0.05 0.16 0.03 0.13* 0.02 0.14* 0.02
5. Openness to experience 0.00 0.08 0.09 0.07 0.18 -0.01 0.12* 0.06 -0.05 0.16*
Study 2: Ratings of dormitory floormates

1. Extraversion 0.29 0.04 0.18* 0.13* 0.03 0.10*
2. Agreeableness 0.17 0.21 0.05 0.29* 0.20* 0.07* 0.13*
3. Conscientiousness 0.21 0.20 0.24 0.05 0.25* 0.19* 0.07* 0.08*
4. Emotional stability 0.20 0.22 0.22 0.28 0.04 0.22* 0.20* 0.15* 0.04
5. Openness to experience 0.21 0.15 0.18 0.16 0.23 0.10* 0.25* 0.15* 0.04 0.17*
Study 3: Ratings of organization members

1. Extraversion 0.15 0.07 0.08* 0.00 0.04 0.08*
2. Agreeableness 0.20 0.27 0.13* 0.24* 0.12* 0.09* 0.12*
3. Conscientiousness 0.20 0.16 0.15 0.13* 0.17* 0.11* 0.07 0.10*
4. Emotional stability 0.11 0.12 0.05 0.23 0.10* 0.09* 0.04 0.21* 0.08*
5. Openness to experience 0.21 0.27 0.20 0.09 0.39 0.12* 0.27* 0.09* 0.01 0.19*
Note. Study 1: N = 649 total ratings from 165 raters; Study 2: N = 1,004 total ratings from 364 raters; Study 3: N = 802 total ratings from 311 raters.
Correlations are between perceptions of others’ Big Five traits (in the rows and first set of columns) and self-rated Big Five traits (in the second
set of columns). The type of rating used to capture perceptions of others and the number of ratings used to estimate the correlations are given
separately for each study. The same-trait perceiver effects and the assumed similarity correlations are shown in italics. The highest correlation
between how the participant rated targets and a self-rated personality trait is in bold for each row.
* p < 0.05.
but negligible individual differences in how people perceived the correlation between how participants see themselves and
targets on conscientiousness and extraversion.2 how they see someone else on the same trait). The highest
The cross-class correlations were not particularly large in correlation for each row is shown in bold to indicate which
the current study; however, they nonetheless appeared to be self-rated Big Five dimension has the highest association
relatively large in relation to the size of the ICCs. The ICCs with a given perceiver dimension.
for Big Five perceiver effects averaged 0.10. The cross-class Relations between perceiver effects and self-rated personal-
correlations in the matrix averaged 0.06, indicating that they ity traits. As can be seen in Table 1, four of the Big Five per-
were roughly 60% of the size of the same-trait correlations. sonality traits showed significant assumed similarity corre-
Indeed, individuals who rated one individual as agreeable lations. The largest effect was found for agreeableness (r =
were likely to rate a different individual as emotionally sta- 0.19), followed by openness (r = 0.16), emotional stability (r
ble and vice versa (r = 0.16); this cross-class correlation was = 0.14), and conscientiousness (r = 0.09, all ps < 0.05). Only
equal in magnitude to the size of either trait’s ICC, indicating extraversion failed to show an assumed similarity effect (r =
that the perceiver effects for these traits should be highly as- -0.04). Following past research, this indicates that individu-
sociated with one another. als are more likely than chance to describe others as they see
Relation between ratings of others and dispositional charac- themselves.
teristics. We next correlated scores that the participants gave
to single targets on the Big Five with the participants’ own
scores on the Big Five. To estimate the relationship between 2
The variance of perceiver effects is estimated more commonly
dispositional variables and ratings of others in this and the through linear mixed models as the ICC associated with raters. When
remaining studies, we correlated participants’ scores on dis- done this way, the variances were estimated at 0.00 for extraversion,
0.14 for agreeableness, 0.04 for conscientiousness, 0.17 for emotional
positional variables to the ratings participants provided for
stability, and 0.18 for openness, which indicates that the two meth-
targets they rated. Again, the correlations we show relate a ods provide similar answers. It should be noted that negative ICCs
rater’s dispositional variables to how the rater perceives a are not expected to occur when raters rate a large number of people;
single target. Because multiple target ratings are nested with- they should only occur when raters are actively contrasting their rat-
in raters, statistical significance values for these correlations ings from their previous ratings and/or when raters have rated few
were obtained through linear mixed modeling. These corre- targets (Kenny, Mannetti, Pierro, Livi, & Kashy, 2002). Because par-
lations are shown on the right side of the top section of Ta- ticipants were told they could not give the same value twice for dif-
ble 1. Correlations on the diagonal of the correlation matrix ferent targets, this should have generally decreased ICCs and could
have produced the contrast conditions necessary for negative ICCs.
represent the traditional assumed similarity correlations (i.e.,
Of more interest were the correlations on the off-diago- ceiver effects were almost certainly artificially lowered due
nals of the correlation matrix. We found numerous correla- to the fact that the procedure forced raters to give each tar-
tions between self-perceptions and perceptions of others for get a different rating. Because perceiver effects are by defini-
non-matching traits. Self-ratings of agreeableness were asso- tion differences in how individuals generally rate others in
ciated with perceiving others as significantly more conscien- a population, forcing people to vary their ratings more than
tious, emotionally stable, and open (all rs ≥ 0.10). Interesting- they normally should result in an underestimation of the ex-
ly, these relationships between perceptions of others and self- tent to which perceiver effects infuse ratings. Additionally,
rated agreeableness were all nearly equal in magnitude to the groups in this study were made up of people who were gen-
magnitude of the assumed similarity effects for each trait. Ad- erally friends with one another. Because participants formed
ditionally, self-ratings of emotional stability were associated their own groups, a perceiver’s judgments could reflect not
with perceiving others as agreeable (r = 0.14), and self-ratings just perceiver effects but also affiliative preferences (e.g., ten-
of conscientiousness were associated with perceiving others as dencies to rate others in the group as extraverted could be
extraverted (r = 0.08). Finally, self-ratings of extraversion were due to tendencies to select extraverted friends). Consequent-
associated with perceiving others as conscientious (r = 0.09) ly we continued by studying a different context, in which
despite the fact that extraversion did not show an assumed participants were less likely to rate friends and were free to
similarity correlation. The regular observation of off-diagonal give multiple targets the same rating.
effects clearly indicates that perceiver effects reflect more than Despite these limitations of Study 1, the finding that char-
simply projections of one’s own self-image onto other people. acteristics such as agreeableness, depression, and narcissism
Relations between perceiver effects and other dispositional were correlated with perceptions of others along many or all
characteristics. We also examined how perceiver effects were Big Five dimensions simultaneously was consistent with our
associated with a range of other characteristics assessed in expectation that perceiver effects have correlates beyond as-
this sample. These correlations are shown in Table 2. sumed similarity and may have a simpler structure than self-
First, we found that gender was associated with perceiver ratings. In Study 2, we explored these questions more directly.
effects, with women tending to judge others as significant-
ly more extraverted, conscientious, and open to experience Method
than did men (all rs between 0.09 and 0.18). A rater’s level of Participants. A total of 643 students were invited to partic-
depression, based on the rater’s BDI score, was also associat- ipate in a study called the Personality and Social Relation-
ed with a range of perceptions of others, with depressed rat- ship Study, which was described as investigating the rela-
ers tending to judge others as significantly more extravert- tionships students had with their peers within the dormitory
ed (r = 0.08), less agreeable (r = -0.09), and less emotional- system at Wake Forest University. Participants completed the
ly stable (r = -0.12). Narcissism scores from the NPI were as- study online and were given $15 after completing the study,
sociated with negative perceptions across every dimension which involved rating their own personality characteristics
of personality, with higher narcissism being associated with and those of three randomly selected members of their floor
judging others as less extraverted, agreeable, conscientious, as described further later in this paragraph. A total of 365 in-
emotionally stable, and open to experience (all rs ≤ -0.11, ps dividuals completed the measures used in the present inves-
< 0.05). Finally, the measures of intellectual ability showed tigation. Of these participants, 219 were female (60%). A total
some association with perceptions of others. Higher GPA, of 1,004 usable ratings were collected from these participants;
SAT, and IQ scores were all associated with tendencies to ratings were excluded if they contained missing values or if
perceive others as less open to experience (all rs ≤ -0.08, ps < the person did not vary the ratings. The study was conduct-
0.05). In addition, higher GPA was associated with a small ed among residents of freshman dormitories near the end of
tendency to rate others as higher in agreeableness (r = 0.08), their second semester of school, after they had been living on
and higher SAT scores were associated with a small tenden- their floor for approximately seven months. University hous-
cy to rate others as lower in conscientiousness (r = -0.09). ing administrators reported that freshmen had been random-
ly assigned to their dormitories and roommates, with the only
Study 2: Structure and Correlates of Perceiver Effects constraint that they had been assigned to same-sex floors.
The nonparallel relationships between dispositional char-
acteristics and perceiver effects suggested there was more to Table 2
perceiver effects than simple assumed similarity biases. Peo- Correlations Between Participant Characteristics and Per-
ple high in agreeableness were significantly more likely to ceiver Effects (Study 1)
Participant Perceiver effect
rate targets as having positive levels of almost all Big Five
characteristic E A C S O
dimensions, whereas narcissism was associated with more Gender (0=M, 1=F) 0.17* 0.05 0.09* -0.04 0.18*
negative views of others along all Big Five dimensions. Fur- Depression (BDI) 0.08* -0.09* -0.02 -0.12* 0.05
thermore, tendencies to perceive others as extraverted were Narcissism (NPI) -0.14* -0.11* -0.12* -0.11* -0.14*
not associated with the perceiver’s self-rated extraversion GPA -0.01 0.08* -0.04 0.04 -0.09*
but were associated with the perceiver’s gender and level of SAT -0.07 -0.03 -0.09* -0.02 0.11*
depression and narcissism. IQ (Wonderlic) -0.03 0.01 -0.04 0.00 -0.08*
Despite the promising nature of the first study, it had sever- Note. All correlations are based on 602 or more ratings from 152 or
al important limitations. Our analyses of the perceiver-effect more different raters. E = extraversion; A = agreeableness; C = con-
matrix provided some indication that the perceiver effects scientiousness; S = emotional stability; O = openness; BDI = Beck De-
for different traits may be associated with one another. How- pression Inventory; NPI = Narcissistic Personality Inventory; Won-
ever, the magnitude of both intraclass and cross-class per- derlic = Wonderlic IQ test.
* p < 0.05.
Materials and Procedure General liking by others. Participants in the study rated the
Inventory of individual differences in the lexicon. In order extent to which they liked every other individual living on
to obtain a broad survey of participant characteristics, we their dorm floor, using a 7-point scale ranging from 1 (strong-
asked participants to complete an early 57-item version of ly dislike) to 7 (strongly like). The average agreement between
the Inventory of Individual Differences in the Lexicon (IIDL; two different liking judgments of the same target (the lik-
Wood, Nye, & Saucier, 2009). The IIDL is designed to mea- ing ICC) was estimated as r = 0.32. Liking ratings were ag-
sure a broad array of distinguishable individual differenc- gregated across anyone who rated the participant to form an
es that can be identified within the lexicon through the use estimate of the extent to which the individual was generally
of a single item consisting of a pair of synonymous trait ad- liked by others. Given that the average number of raters per
jectives (e.g., bold/assertive, outgoing/sociable, enthusiastic/excit- target was 12.9 (range: 3-22), the expected reliability of the
ed).3 Participants completed the inventory to describe them- average liking estimates using the Spearman-Brown prophe-
selves, using the instructions “How do you see yourself cy formula was estimated at 0.86.
in general?” and given the stem “I see myself as someone
who is . . .” Four items of the IIDL pertaining more to demo- Results
graphic or nonbehavioral characteristics (i.e., items concern- Assumed similarity. We first examined the assumed sim-
ing perceptions of a target’s height, health, wealth, and age) ilarity correlations for Big Five traits. As shown in the right
were excluded. Participants first rated their own personal- half of the middle section of Table 1, there were significant
ity on the IIDL, and then at the end of the survey they rat- assumed similarity effects for all Big Five traits except extra-
ed up to three different residents of their floor on the IIDL. version. However, in this study, the best predictor of a per-
In order to increase the likelihood that participants would ceiver effect for any Big Five dimension was invariably self-
be assigned to rate people they knew, we asked participants reports of agreeableness. This again pointed to the limita-
to rate randomly selected targets who lived within a couple tions of assumed similarity as an explanation of the sources
doors of their own room. of perceiver effects for different traits.
Analyses in a sample that contained both the Big Five In- Intraclass and cross-class correlations of trait perceiver ef-
ventory (John & Srivastava, 1999) and IIDL reported by fects. We examined the extent to which a rater’s judgments
Wood and colleagues (2009) suggested that Big Five marker of a target’s traits were associated with the same rater’s judg-
scales could be created by averaging selected IIDL items. An ments of a different target’s traits. As can be seen in the left
extraversion scale was made from items outgoing/extraverted, side of the middle section of Table 1, the cross-class corre-
enthusiastic/excited, loud/noisy, and bashful/shy (reverse-scored lations differed substantially from zero, indicating that per-
[RS]); in self-ratings the internal consistency was α = 0.65. ceiver effects for the Big Five were not independent of one
An agreeableness scale was made from the items kind-heart- another. In fact, the average cross-class correlation of the per-
ed/caring, supportive/encouraging, unsympathetic/unfriend- ceiver effects for Big Five traits was 0.19, versus an average
ly (RS), and angry/hostile (RS); α = 0.73. A conscientiousness ICC of .25, indicating that cross-trait correlations were on av-
scale was made from the items organized/efficient, dependable/ erage 75% of the size of same-trait correlations. This provides
reliable, messy/sloppy (RS), and unreliable/undependable (RS); α clear indication that the perceiver effects for Big Five traits
= 0.72. An emotional stability scale was made from the items are far from orthogonal from one another.4 Structure of per-
insecure/unsure (RS), tense/nervous (RS), sad/unhappy (RS) and ceiver effects within the IIDL. We next examined the struc-
calm/relaxed; α = 0.70. And an openness scale was made from ture of perceiver effects by conducting a factor analysis on
the items creative/artistic, complex/deep, intelligent/smart and the perceiver-effect correlation matrix for all 57 IIDL items.
narrow-minded/close-minded (RS); α = 0.43. Principal axis factoring was used to extract factors, and the
Personality disorder scales. To investigate how perceiver first 10 eigenvalues were 7.6, 2.7, 1.4, 1.2, 1.2, 1.1, 1.1, 1.1, 1.0,
effects might be associated with personality disorders, we and 1.0. Examination of the scree plot showed little reason to
had participants complete selected subscales from the Millon extract more than two factors. This was supported by exam-
Clinical Multiaxial Inventory-III (MCMI-III; Millon, 1997). ining the three-factor solution, which revealed no items on
We selected subscales that we suspected would be most re- the third factor with loadings above 0.25. Consequently, we
lated to individual differences in perceiver effects as assessed describe the features of the one- and two-factor solutions.
here, and we included at least one disorder from each of the
three major clusters of personality disorders (American Psy- 3
The IIDL items are discussed in the text by italicizing the adjec-
chiatric Association, 1994). We chose to include the subscales tives and separating them by a slash mark (e.g., outgoing/sociable) de-
assessing narcissistic, paranoid, antisocial, avoidant, border- spite the fact that participants are presented with the same items in a
slightly different format (i.e., “outgoing, sociable”). This convention
line, and obsessive- compulsive personality disorders. The
was adopted to avoid a confusing proliferation of commas and quo-
scales were measured using standard scoring for the MCMI- tation marks in the text.
III scales, with the exception that items pertaining to drug 4
It is possible that the high intercorrelations among perceiver effects
use or suicidality were eliminated to alleviate participant for Big Five traits may be a property of the IIDL estimates of the Big
concerns about confidentiality. This resulted in the elimina- Five rather than of perceiver effects and thus would be observed in
tion of five of the 79 total items. self-reports. This did not appear to be the case. To examine this, we
Life satisfaction. Life satisfaction was measured using a correlated self-ratings of Big Five traits with self-ratings on the same
shortened three-item version of the Satisfaction with Life scales collected approximately four days later. The size of off-diago-
nal correlations (e.g., self-ratings of extraversion with self-ratings of
Scale (Pavot & Diener, 1993). The items were “In most ways
agreeableness four days later) was only 26% of the magnitude of on-
my life is close to my ideal,” “The conditions of my life are diagonal correlations (e.g., self-ratings of extraversion with self-rat-
excellent,” and “I am satisfied with my life” (α = 0.87). ings of extraversion four days later).
The loadings of IIDL items on the first unrotated factor is strength of association between dispositional variables and
shown in Table 3. The factor contrasted positive and nega- how organization members are rated in general, these corre-
tive perceptions of others, with the highest loading items of lations can be divided by the square root of the perceiver ICC.
the dimension indicating that the dimension concerned in- In the present study, this means that the relationships be-
dividual differences in the extent to which individuals tend- tween dispositional characteristics and how others in the or-
ed to rate others as having a host of positive characteristics, ganization are rated in general are estimated to be 62% larger
ranging from being interesting, impressive, trustworthy, and than the relationships we report between dispositional char-
friendly to being happy and emotionally stable. acteristics and how single targets are rated. For instance, the
When two factors were extracted and rotated using princi- -0.25 correlation we found between self-perceptions of being
pal axis factoring and varimax rotation, the positive and neg- angry/hostile and positive ratings of a single peer would cor-
ative poles of the dimension identified in the one-factor so- respond to an expected -0.25√0.38 = -0.40 correlation between
lution split into separate factors. The highest loading items self-perceptions of being angry/hostile and how positively oth-
of the first factor consisted of the items great/wonderful, ac- ers in the organization are judged in general.
complished/successful, skilled/talented, affectionate/passionate, ex- We first examined how positive perceptions of others was
traordinary/exceptional, kind-hearted/caring, happy/joyful, and related to gender and found that women were considerably
cheerful/good-humored (all factor loadings were between 0.53 more likely to rate a single target positively than were men
and 0.42), which indicates that the factor represents varia- (r = 0.22). However, given that participants invariably rated
tion in perceptions of the dynamism, status, or agentic ten- targets of their own gender in the present study, this could
dencies of others. The highest loading items for the second be due either to differences in how men and women rate oth-
dimension consisted of the adjective pairs unstable/disturbed, ers (i.e., women may tend to rate any target more positive-
ashamed/guilt-prone, unsympathetic/unfriendly, cruel/abusive, ly) or to real differences between men and women (i.e., wom-
bad/immoral, tense/nervous, angry/hostile, and sad/unhappy (all en may tend to be perceived by raters of either gender more
loadings were between 0.47 and 0.38), which indicates that positively than are men). Due to the inability to disentan-
the factor represents variation in how antisocial and neurot- gle these explanations without cross-gender ratings, we con-
ic others are typically perceived. trolled for gender in all remaining analyses, so that relation-
Additional analyses suggested that the separate positive ships between positivity and other dispositional characteris-
and negative perceptions of others found in the two-factor tics may be more easily interpreted independent of the con-
solution may not differ from each other importantly. Analy- founds of the gender of the raters and of the targets rated.
sis of these two dimensions suggested that when factor scor- Personality traits. We report the relationship between the
ing was not forced on these dimensions, they showed sub- positivity of perceptions of others and self-ratings of person-
stantial negative correlations with one another. In particular, ality separately for each Big Five domain.
the cross-trait perceiver effect (how much rating a target as Within the domain of extraversion, the positivity of per-
antisocial is associated with rating a different target as inter- ceptions of others was consistently associated with self-rat-
esting) was over half the size of the same-trait perceiver ef- ings indicating positive affectivity (e.g., happy/joyful, enthusi-
fect for these dimensions, suggesting that the latent correla- astic/excited). Other aspects of extraversion more indicative
tion between these two dimensions was greater than -0.50. of sociability or agentic behavior were generally unassociat-
Additionally, the correlates of the two scales were largely ed with positive judgments of others.
mirrors of one another, with the dispositional characteristics There were consistent strong associations between aspects of
that were correlated with positive perceptions being corre- agreeableness and conscientiousness and the positivity of judg-
lated in the opposite direction and at roughly the same mag- ments of others. Tendencies to judge others positively were as-
nitude with the second dimension (the column-vector corre- sociated with every dimension related to agreeableness, with
lation of the two dimensions across the characteristics shown the strongest relationships observed for characteristics asso-
in Table 3 was -0.92). ciated with adversarial behavior (e.g., unsympathetic/unfriend-
Given the redundancy of the pattern of relationships with ly, angry/hostile). Similarly, most aspects of self-rated conscien-
dispositional variables for the two factors, we considered tiousness were associated with the positivity of perceptions of
only the properties concerning the first unrotated factor in others, with the strongest effects for dependability.
the analyses that follow. To do this, we created a perceiver The positivity of perceptions of others was also associat-
score by averaging the 10 positive items with the strongest ed with some aspects of emotional stability and openness.
loadings and the 10 negative items with the strongest load- Individuals who described others more positively tended
ings after reverse-scoring to create a 20-item measure; we re- to describe themselves as less unstable/disturbed and crabby/
fer to this dimension as the perceived positivity of others. We es- grouchy, whereas aspects concerning anxiety or confidence
timated the perceiver ICC for the positivity dimension to be showed no significant associations. Within the domain of
0.38, indicating that there were fairly large individual differ- openness to experience, the positivity of perceptions of oth-
ences in how positively raters judged targets in the group in ers was associated with evaluating oneself as more intelli-
general and that these ratings have a fairly large impact on gent/smart and less narrow-minded/close-minded, but was not
single-observer reports. significantly associated with other aspects of openness.
Relationships between perceived positivity of others and Many self-ratings that were relatively unassociated with
dispositional characteristics. We next examined how the per- the Big Five were also associated with positive ratings of oth-
ceived positivity of others was associated with other dispo- ers: Individuals who described themselves as well-liked/like-
sitional variables. As in Study 1, the correlations that are re- able, great/wonderful, and extraordinary/exceptional, and as less
ported show how dispositional variables are associated with bad/immoral, annoying/aggravating, and dumb/stupid tended to
perceptions of a single target of the organization; to show the rate others more positively.
Table 3
Correlations between Positivity of Perceptions of Others and Dispositional Variables in a Dormitory Sample (Study 2)
Big Five r between self-rating Perceiver intraclass Loading on
Measure and item associations and Positivity factor correlation Positivity factor
Inventory of Individual Differences in the Lexicon (IIDL)
Outgoing/extraverted E 0.06 0.11 0.34
Enthusiastic/excited E 0.17 0.22 0.41
Bold/assertive E -0.04 0.16 0.25
Happy/joyful E, A, S 0.23 0.26 0.46
Loud/noisy E -0.02 -0.03 0.04
Energetic/active E 0.15 0.21 0.37
Funny/amusing E 0.21 0.22 0.41
Brave/fearless E, O -0.03 0.17 0.26
Bashful/shy -E, -S 0.00 0.06 -0.21
Kind-hearted/caring A 0.22 0.23 0.44
Thankful/grateful A 0.22 0.22 0.39
Affectionate/passionate A, E 0.17 0.33 0.46
Courteous/polite A, C 0.23 0.17 0.38
Truthful/honest A 0.23 0.22 0.42
Cheerful/good-humored A, E 0.23 0.25 0.47
Supportive/encouraging A, E 0.24 0.22 0.40
Short-tempered/impatient -S, -A -0.13 0.14 -0.25
Unsympathetic/unfriendly -A, -E -0.27 0.24 -0.44
Angry/hostile -A, -S -0.25 0.21 -0.36
Conceited/egotistical -A -0.20 0.10 -0.20
Cruel/abusive -A -0.22 0.29 -0.37
Insulting/offensive -A -0.19 0.14 -0.29
Controlling/dominant -A -0.12 0.10 -0.08
Organized/efficient C 0.17 0.08 0.12
Dependable/reliable C 0.20 0.16 0.37
Level-headed/sensible C 0.18 0.13 0.31
Accomplished/successful C, E 0.14 0.28 0.50
Competent/capable C 0.14 0.22 0.41
Messy/sloppy -C -0.17 0.07 -0.02
Unreliable/undependable -C -0.23 0.20 -0.32
Childish/immature -C -0.08 0.16 -0.21
Awkward/clumsy -C, -S -0.05 0.19 -0.28
Calm/relaxed S 0.11 0.16 0.26
Confident/self-assured S, E 0.06 0.20 0.37
Satisfied/secure S, C 0.14 0.14 0.31
Tense/nervous -S -0.06 0.24 -0.35
Ashamed/guilt-prone -S -0.10 0.25 -0.33
Unstable/disturbed -S -0.26 0.30 -0.42
Insecure/unsure -S, -E -0.10 0.15 -0.25
Sad/unhappy -S, -E, -A -0.17 0.23 -0.32
Crabby/grouchy -S, -A -0.20 0.18 -0.33
Lonely/lonesome -S -0.18 0.20 -0.35
Creative/artistic O 0.04 0.24 0.28
Intelligent/smart O 0.15 0.26 0.42
Complex/deep O 0.01 0.24 0.21
Skilled/talented O 0.08 0.28 0.47
Traditional/conventional -O 0.09 0.10 0.16
Narrow-minded/close-minded -O, -A -0.14 0.16 -0.26
Prominent/well-known E 0.01 0.21 0.26
Well-liked/likeable E, A 0.21 0.16 0.39
Great/wonderful E 0.10 0.32 0.55
Weird/strange -C -0.06 0.27 -0.24
Attractive/good-looking E 0.01 — —
Annoying/aggravating — -0.20 0.16 -0.28
Bad/immoral — -0.26 0.25 -0.39
Dumb/stupid — -0.16 0.22 -0.39
Extraordinary/exceptional — 0.10 0.27 0.46
Stylish/fashionable — 0.05 0.26 0.36
Additional variables
Height — -0.11 — —
Religiosity — 0.06 — —
Political orientation — 0.01 — —
Table 3 (continued)
Big Five r between self-rating Perceiver intraclass Loading on
Measure and item associations and Positivity factor correlation Positivity factor
GPA — 0.11 — —
Average liking from floormates — 0.16 — —
Life satisfaction — 0.11 — —
Millon Clinical Multiaxial Inventory-III

Narcissism — -0.12 — —
Paranoid — -0.20 — —
Antisocial — -0.20 — —
Avoidant — -0.16 — —
Borderline — -0.18 — —
Obsessive-compulsive — 0.17 — —
Note. All analyses are based on at least 913 observer ratings from at least 353 raters; all correlations are conducted controlling for gender. The
“Big Five associations” column shows all Big Five inventory scales that have correlations of at least |r| = 0.30 with the item from a previous in-
vestigation, where E = extraversion, A = agreeableness, C = conscientiousness, S = emotional stability, O = openness to experience (with scales
listed in descending order of highest to lowest correlations). Negative signs (e.g., “-S”) indicate that the item is negatively associated with the
Big Five dimension. Values in bold indicate (a) same-trait perceiver effects greater than 0.20 and factor loadings greater than 0.30, as well as (b)
all statistically significant correlations (p < 0.05) between self-rated IIDL items and the positivity factor. Dashes indicate missing values.
Personality disorders. We next examined how the MC- and emotionally stable. The fact that we were unable to iden-
MI-III measures of personality disorders were associated tify even two relatively orthogonal dimensions of perceiv-
with the positivity of perceptions of others. These correla- er effects suggests that people have relatively undifferenti-
tions are shown in Table 3. Most of the personality disorders ated rater biases, as manifested in how they generally tend
showed the same general pattern of relationships, in which to rate others.
higher scores on personality disorders were associated with How positively individuals rated others also showed siz-
judging targets less positively. This pattern was found for able relationships with a number of dispositional variables.
the MCMI-III measures of narcissistic, paranoid, antisocial, Describing others as interesting, happy, and possessing few-
avoidant, and borderline personality disorders. The associa- er antisocial tendencies was associated with a host of charac-
tions were similar in magnitude across these disorder scales, teristics indicative of good emotional functioning (e.g., hap-
with the personality disorder measure showing associations piness, life satisfaction, and emotional stability) and positive
of approximately |r| = 0.15 with how positively others were relations with others (e.g., friendliness and courtesy vs. hos-
judged. Only the MCMI-III measure of obsessive-compul- tility and undependability), as well as more desirable self-
sive tendencies showed a different pattern, which was the re- evaluations (e.g., self-perceptions of being likable and great).
verse of the general pattern: Higher levels of obsessive-com- Given that the tendency to perceive others positively could
pulsive tendencies were associated with more positive judg- be simply a scale-use bias, it is important to note that posi-
ments of others (rs = 0.17). tive perceptions of others correlated with a number of mea-
Additional variables. We explored the relationship between sures that should be less susceptible to the scale-use biases
perceiver effects and height, religiosity, political orientation, that might infuse abstract trait ratings. As in Study 1, indi-
GPA, life satisfaction, and how much the person was liked viduals who saw others more positively tended to report a
by dormitory floormates. We found that taller raters tended higher GPA. Further, individuals were more likely to rate
to rate targets less positively (rs = -0.11). As in Study 1, self- others positively if they were shorter and female and if oth-
reported GPA was associated with more positive ratings of ers on the floor rated them as likable. These effects strongly
others (r = 0.11). Reports of life satisfaction were also asso- suggest that correlations between the perceived positivity of
ciated with seeing others more positively (r = 0.11). Finally, others and other dispositional variables are not entirely driv-
as might be expected, individuals who rated targets on their en by scale-use differences.
floor more positively tended to be rated as more likable by
their floormates (rs = 0.16). Study 3: Stability of Perceiver Effects
In this final study, we focused on three questions. Using a
Discussion different data set, in which perceiver effects were estimat-
The factor analysis results of Study 2 indicate that the struc- ed through ratings by members of the participants’ frater-
ture of perceiver effects is decidedly simpler than the struc- nity or sorority, we conducted new factor analyses to exam-
ture of self-ratings or ratings of single peers. In particular, ine whether the factor structure of perceiver effects identi-
the results suggest that a single factor concerning how pos- fied in Study 2 could be identified in a new data set with dif-
itively others are perceived is sufficient to capture most of ferent items. Second, we conducted analyses linking perceiv-
the covariation in how individuals tend to see others across er-effect dimensions to a broader array of dispositional vari-
a broad range of traits. This dimension was found when ex- ables, including self- and peer-ratings of personality traits,
amining a suite of traits that are relatively distinct at the lev- well-being measures, and attitudes about power. Given that
el of self-reports or single-peer reports. The single-perceiver- perceiver effects are operationalized as perceptions by mem-
effect dimension reflects the extent to which other people are bers of a defined group and reflect a person’s average ratings
seen as interesting, impressive, trustworthy, friendly, happy, of a person in that context, we also explored how perceiver
effects were associated with the participants’ overall experi- reports that were made of the same individual to see wheth-
ence with that group. Third, we explored the stability of per- er an individual’s perceiver effects were associated with how
ceiver effects over a year to examine whether the stability of they were seen by others. Each participant was rated by an
perceiver effects might be comparable to other disposition- average of 2.1 raters, and the overall reliability of observer
al variables. reports of personality based on interrater agreement and this
number of raters using the Spearman-Brown prophecy for-
Method mula was estimated at 0.55 for extraversion, 0.41 for agree-
Participants and procedure. A total of 366 participants ableness, 0.43 for conscientiousness, and 0.41 for emotion-
were recruited from seven different fraternities and sorori- al stability, and 0.25 for openness. For the relationships we
ties at the University of Illinois at Urbana-Champaign. A to- report between perceiver effects and how the individuals
tal of 203 women were recruited from three sororities, and were rated by peers on Big Five characteristics, the aggregate
163 men were recruited from four fraternities. The partici- peer ratings have been corrected for unreliability, and con-
pation from each organization ranged from 34 members in sequently the correlations should be interpreted as the rela-
one fraternity to 75 in two sororities. Participants completed tionship between the positivity of a participant’s ratings of a
the survey at the organization’s house, usually in the span of single target and the individual’s reputation in the organiza-
about 2 hr. Only fraternities and sororities that were housed tion in general.
on campus were approached, given our desire to examine
settings where members were already acquainted. Partic- Other Individual Difference Measures
ipants were given $10 for completing the first administra- Hope for and fear of power. Participants completed six-
tion of the survey and were given an additional $20 if they item versions of the Hope for Power and Fear of Power scales
completed the second administration of the study a year (Harms, Roberts, & Wood, 2007). The item with the highest
later. The organization was also given money for its assis- item-total correlation with the Hope for Power scale was “I
tance with the study; the amount varied as a function of the am driven to become as powerful as possible” (α = 0.80). The
size of the organization and the percentage of participating item with the highest item-total correlation with the Fear of
members. Organizations and their members were told that Power scale was “The thought of being put in a position of
the answers they provided would be confidential (i.e., nei- authority scares me” (α = 0.80).
ther the participants nor their organizations would be iden- Social dominance orientation. Participants completed an
tified when information concerning the study was shared eight-item version of the Social Dominance Orientation Scale,
with others). which assesses belief in and comfort with group inequalities
Personality trait ratings of self and others. Participants com- (Pratto, Sidanius, Stallworth, & Malle, 1994, p. 757). The item
pleted an inventory containing 75 adjectives, which included with the highest item-total correlations was “It is important
a set of 53 adjectives designed to measure the Big Five (Wal- that we treat other countries as equals” (RS; α = 0.82).
ton & Roberts, 2004) and a number of terms selected to mea- Misuse of power. Participants also completed the eight-
sure subclinical psychopathy, narcissism, and Machiavellian- item Misuse of Power scale (Lee-Chai, Chen, & Chartrand,
ism (the “Dark Triad;” Harms, Wood, Brummel, & Roberts, 2001). This scale assesses how acceptable participants feel it
2010). Participants completed self-ratings with the prompt is to misuse their power and status (α = 0.66).
“How do you see yourself in general?” and with response op- Well-being. To assess well-being, we administered the five-
tions ranging from 1 (Strongly disagree) to 5 (Strongly agree). item Satisfaction with Life Scale (Pavot & Diener, 1993; α =
Items were presented to participants alphabetically. 0.84) and the 20-item Center for Epidemiological Studies-De-
After completing all other survey materials, participants pression scale (Radloff, 1977; α = 0.88).
were asked to rate the personalities of three other members General fit with peers. Participants rated these two items in-
of their organization on the same 75-item trait inventory. The dicating their fit with others who were their age: “I feel that
members they were assigned to rate were randomly selected, I fit in with people my age” and “I feel that I fit in with my
with the only constraint being that people were more likely peers” (α = 0.81).
to rate individuals of their own year or adjacent years (e.g., Satisfaction with organization. Satisfaction with the orga-
seniors were unlikely to rate freshmen, and vice versa). Be- nization was measured with the items “On the whole, being
cause assignments of targets to raters were near-random, in this [fraternity/sorority] is dissatisfying to me” (RS) and
sometimes raters were asked to rate individuals they did not “Overall, I am satisfied with being in this [fraternity/soror-
know or to complete observer reports of themselves. If either ity]” (α = 0.89).
of these situations occurred, individuals were asked to re- Perceived capacity for power. A five-item version of Ander-
turn their form to the experimenter and were then given an- son, John, and Keltner’s (2009) Personal Sense of Power scale
other, randomly selected target to rate. Finally, some partic- was included in the present sample. The scale assesses the
ipants completed observer reports inappropriately by copy- extent to which members believe they can be heard by and
ing their answers from a previous form. We identified these influence other members of their organization (α = 0.74).
individuals by performing a profile correlation across their Organization identity goals. A six-item scale was included
answers. If the profile correlation between two ratings by the that assessed a person’s desire to identify and be recognized
same rater approached unity (i.e., r > 0.99), we excluded all as a member of his or her organization. Items with the high-
of the person’s observer ratings. A total of 908 ratings were est item- total correlations were “I want to be recognized as
included in the analysis from 323 different raters. a member of this [fraternity/sorority] by outsiders” and “I
In addition to using these observer reports for examination want to be a member of this [fraternity/sorority]” (α = 0.87).
of properties of perceiver effects, we also averaged observer Peer-rated liking, knowledge, and influence. Participants
rated the extent to which they knew and liked every other lected among individuals nested in groups (Gonzalez & Grif-
member in their organization and were also asked to rate the fin, 2002; Raudenbush & Bryk, 2002), we combined these vari-
level of influence each member had over organization deci- ances and covariances to estimate the latent stability of per-
sions. After removing self-ratings, we aggregated these rat- ceiver effects over time. By dividing the covariance of Year 1
ings to estimate how much each person was seen as liked, and Year 2 random effects (12) by the square root of the vari-
known, and had status as rated by the person’s peers. By esti- ances for those years (11 and 22), we estimated the level of
mating ICCs from a linear mixed model, we found the inter- rank-order stability that should be expected between reliably
rater agreement to be 0.15 for liking ratings, 0.22 for ratings estimated perceiver effects over a year. In other words, this
of how much members were known by the rater, and 0.35 for stability coefficient allowed us to estimate how much per-
influence ratings. Because participants were rated by an av- ceiver-effect estimates collected in different years would cor-
erage of 28 other people, the reliability of these estimates for relate if we had individuals rate a large number of targets
the average participant for all dimensions was expected to one year and a large number of different targets a year later.
exceed 0.83. As applied to the overall positivity of ratings, this allowed us
Modeling the stability of perceiver effects. The stability of to estimate the extent to which we expected raters who gen-
perceiver effects was estimated by creating a separate file for erally rated targets particularly positively or negatively one
each trait that contained three columns consisting of the rat- year to continue to rate targets particularly positively or neg-
er’s identification number, the year the rating was collected, atively a year later.
and the score that the rater gave the target on that trait. In Note that the formulas given in Figure 1 estimate the corre-
turn, a linear mixed model was constructed in which main ef- lation in which the individuals rate several targets but nev-
fects of the 2 years were specified as fixed predictors and ran- er rate any particular target twice. We thus eliminated peer
dom effects for these variables were specified to allow people ratings in which a participant rated the same target in both
to have different variances across raters for ratings provided years of the survey by randomly eliminating one of their
in Year 1 (τ11) and Year 2 (τ22); an unstructured covariance ma- Year 1 or Year 2 ratings of a target they rated twice. Failing to
trix was specified such that these random effects were freely do so would be expected to artificially influence estimates of
allowed to covary (τ12); the model is shown in Figure 1. the longitudinal stability of perceiver effects by making sta-
By using a simple multilevel statistic generally employed to bilities in perceivers’ ratings a function of not just perceiver
estimate the group-level correlations between variables col- effects but also relationship and target effects.
Variances τ11 and τ22 and covariance τ21 are obtained from this linear mixed model:
Ratingijk = b1k(Year1) + b2k(Year2) + rijk
i = target, j = year of assessment, k = rater
b1j = γ1 + u1k, b2j = γ2 + u2k, rijk ~ N(0, σ2r)
Figure 1. Modeling the stability of perceiver effects.

Results ly intercorrelated and that relations with dispositional vari-

Relation between self-ratings and perceiver effects for Big ables mirrored each other. The column-vector correlations of
Five traits. As in the two earlier studies, we once again began the two scales for the variables used in Table 4 was r = -0.94, in-
by exploring the relationship between self-ratings and per- dicating that traits that correlated with one dimension would
ceiver effects on the Big Five traits. These are shown in the correlate roughly with the same magnitude and in the oppo-
bottom right side of Table 1. Similar to results from the previ- site direction with the second dimension. Given the failure of
ous two studies, assumed similarity effects were found for all the two dimensions to show discriminant patterns of associa-
Big Five traits except extraversion (rs ≥ 0.11), but there were tions with disposition, we focused on the one-factor solution.
also numerous associations on the off-diagonals of the corre- As in Study 2, the first unrotated factor served to combine the
lation matrix. Self-ratings of agreeableness were significant- two factors into an overall positivity dimension representing
ly associated with judging others more positively along every a contrast of positive versus negative perceptions. The highest
Big Five dimension (rs between 0.09 and 0.21; ps < 0.05) and in loading negative items were jealous, insensitive, abusive, cold, in-
most cases showed higher correlations with perceiver effects efficient, harsh, envious, anxious, greedy, and deceitful (all load-
than did the correlation for the assumed similarity effect for ings between 0.45 and 0.39), and the highest loading positive
the same trait. Interestingly, as in Study 1, extraversion was items were kind, energetic, talkative, sympathetic, attractive, imag-
associated with perceiving others more positively along ev- inative, honest, trustful, creative, and warm (all loadings between
ery Big Five dimension except extraversion (all rs ≥ 0.10; ps ≤ -0.38 and -0.32). We created scores by averaging these 20 items
0.05), and openness self-ratings were associated with perceiv- after reverse-coding the negative items. As in Study 2, scores
ing others more positively across all Big Five dimensions. on this dimension thus represented how much people saw
Covariation among Big Five perceiver effects. We again ex- others as trustworthy, nice, interesting, and possessing other
amined the extent to which a rater’s judgments of a target’s positive characteristics. We again estimated the perceiver ICC
traits were associated with the same rater’s judgments of the for this dimension. This correlation was estimated at 0.35, very
same and different characteristics in a different target. As can close to the size of the similar dimension in Study 2.
be seen in the bottom left side of Table 1, and as was found
Table 4
in Study 2, the cross-class correlations between trait perceiv-
Correlations Between Positivity of Perceptions of Others and
er effects differed substantially from zero and were nearly the
Dispositional Variables in Greek Organizations (Study 3)
same size as the perceiver ICCs. In fact, the average cross-
Measure Rated positivity of target
class correlation of the perceiver effects for Big Five traits was General dispositional measures
0.16, whereas the average ICC was 0.20, indicating that cross- Self-rated personality
class correlations were more than 80% of the size of ICCs. As Extraversion 0.11
in Study 2, this indicates that the perceiver effects for different Agreeableness 0.33
Big Five traits were highly associated with one another. Conscientiousness 0.14
Structure of perceiver effects. As in Study 2, we conducted a Emotional stability 0.18
factor analysis on the square symmetric correlation matrix that Openness 0.16
Peer-rated personality
can be constructed by correlating all pairs of ratings that a rater
Extraversion -0.08
provided of different targets (i.e., correlating the ratings an indi- Agreeableness 0.13
vidual made of target i with the ratings the individual made of a Conscientiousness 0.22
different target i´). Given the fact that the 13 adjectives beginning Emotional stability 0.15
with un- (e.g., “unintelligent,” “uncreative”) appeared to elicit Openness 0.13
acquiescence or satisficing biases (e.g., some participants provid- Additional measures
ed the same value for all 13 items), we chose to exclude these Age -0.04
terms from the analysis, resulting in a total of 62 items. ACT Assessment score 0.10
How religious 0.07
We then conducted a factor analysis on the perceiver-effect
How liberal/conservative -0.03
matrix using principal axis factor extraction with varimax ro- Fear of Power scale -0.11
tations. The first 10 eigenvalues from the matrix were 5.9, 2.1, Need for Power scale -0.20
1.5, 1.4, 1.3, 1.3, 1.2, 1.2, 1.2, and 1.1. These results again strong- Social Dominance Orientation Scale -0.21
ly pointed to a one-factor or two-factor solution; these were in- Misuse of Power scale -0.22
vestigated in more detail. First, in the two-factor solution the General fit with peers 0.20
highest loading items on the rotated first factor were jealous, CES-Depression scale -0.18
manipulative, harsh, fretful, insensitive, deceitful, anxious, envious, Satisfaction with Life Scale 0.21
Organization-contextualized measures
and abusive (all loadings between 0.41 and 0.38), and on the sec-
Organization identity goals 0.32
ond rotated factor the highest loading items were relaxed, intel- Organization satisfaction 0.31
lectual, energetic, creative, attractive, practical, warm, and imagina- Personal Sense of Power scale 0.24
tive (all loadings between 0.35 and 0.29). Despite the differenc- Average rated influence 0.04
es in the item pools, the two factors looked fairly similar to the Average rated knowledge 0.04
dimensions identified in Study 2, reflecting perceptions of oth- Average rated liking 0.14
ers as possessing negative/antisocial and positive/impressive Note. All correlations are based on at least 802 ratings from at least
characteristics, respectively. However, as in Study 2, analyses 302 raters, except for correlations using peer reports of participant’s
conducted using scores from the two-factor solution demon- personality, which are based on 794 ratings from 279 raters. All sta-
strated that when scale scores of the two dimensions were cre- tistically significant correlations (p < 0.05) are shown in bold. The or-
ated, scores on the two perceiver-effect dimensions were high- ganization of the rater is controlled for all correlations. ACT = Amer-
ican College Testing; CES = Center for Epidemiological Studies.
Correlates of the positivity of perceptions of others. We Discussion

next investigated the relationships between positive percep- Using a different sample and pool of lexical terms, we again
tions of others and dispositional variables. To compensate found that a single positivity factor could adequately account
for the fact that participants rated targets only within their for the covariation in perceiver effects on a broad range of
organization, we controlled for the organization of the rat- traits. This single dimension looked much like the positivity
er in all analyses. Because all organizations were single-sex dimension identified in Study 2; that is, it reflected the ten-
and raters rated members of only their organization, control- dency to perceive people as trustworthy, nice, interesting, and
ling for the organization of rater simultaneously controlled emotionally stable. Further, the correlation between the posi-
for the gender of the rater and target. Again, the correlations tivity of an individual’s judgments of two different targets was
that are reported show how dispositional variables are asso- fairly large, indicating that there are sizable individual differ-
ciated with positive perceptions of a single other member of ences in how positively other targets are judged. As in the first
the organization; to show the strength of association between two studies, we found that variation in perceiver effects cor-
dispositional variables and how others in the organization related in sensible ways with dispositional variables, ranging
are perceived in general, these correlations can be divided by from academic performance and being liked by others to self-
the square root of the perceiver ICC. and peer-reported personality traits, organizational experi-
As can be seen in Table 4, there were various relationships ences, and well-being measures. Finally, our analyses suggest
between how individuals saw others and their dispositional that the individual differences in tendencies to rate others pos-
characteristics. First, individuals who perceived others more itively showed an impressive level of stability over a year.
positively tended to report higher levels of extraversion, con-
scientiousness, emotional stability, openness, and especial- General Discussion
ly agreeableness. In addition, positive perceptions of others The results of the studies presented here indicate that how
correlated with how the raters were judged by others. Indi- we perceive others in our social environments reveals much
viduals who perceived others more positively tended to be about our personality. First, the studies clearly demonstrate
judged by others as significantly more agreeable, conscien- that perceiver effects represent more than simply the projec-
tious, and emotionally stable. tion of an individual’s self-image onto other people. For in-
Individuals who perceived others positively tended to re- stance, in none of the three studies did self-rated extraver-
port a greater sense of fitting in with their same-age peers, sion show a statistically significant relationship with perceiv-
higher life satisfaction, lower depression, and higher ACT ing others as extraverted, despite the fact that self-rated ex-
scores. Having positive perceptions of others was also relat- traversion was regularly associated with perceiving others as
ed to having a lowered desire to have power over others, less possessing a range of other positive characteristics. Similar-
fear of being in a position of power, a lower sense that it is ly, self-rated agreeableness was associated with seeing oth-
acceptable to misuse power, and a lower social dominance ers as more agreeable but also as possessing a host of other
orientation (i.e., belief in the superiority of some groups over positive characteristics.
others). Our investigation also served to follow up on recent find-
Finally, having positive perceptions of others was also as- ings that suggest that perceiver effects may have fewer or-
sociated with having considerably better organizational ex- thogonal factors than do self-ratings (Beer & Watson, 2008a;
periences. People who identified with their organization and Srivastava et al., 2010). Instead of identifying five factors,
reported that they were satisfied and had status in the orga- our analyses suggested that a single-factor model could ef-
nization were considerably more likely to describe members fectively capture most of the covariation in how individuals
of their organization as possessing positive traits (all |rs| be- perceived different targets along different traits. The magni-
tween 0.24 and 0.32). As in Study 2, rating others positively tude of individual differences in the positivity of perceptions
was also associated with being rated as more likable by oth- of others is fairly large, and individual differences in perceiv-
ers in the organization (r = 0.14). er positivity impacted judgments of single targets to a con-
Longitudinal stability of perceived positivity of others. Fi- siderable degree.
nally, we estimated the stability of perceived positivity of Further, we found that individual differences in the positiv-
others using the linear mixed model design described in the ity of perceptions of others show a level of stability compa-
Method section and outlined in Figure 1. This analysis was rable to that of other personality dimensions: approximately
based on 892 ratings from 158 participants who provided rat- 0.69 over the period of a year. It is worth noting that the sta-
ings of other organization members in both years. From this bility of perceiver effects reported here was higher than the
linear mixed model, the variance of observer ratings due to stabilities of 0.20 to 0.50 reported in past studies (Malloy et
rater differences was estimated to be 0.0867 in Year 1 (τ11), al., 1995; Srivastava et al., 2010). The most likely reason for
and 0.0849 in Year 2 (τ22). In contrast, the covariance between this discrepancy is undoubtedly that latent stabilities were re-
Year 1 and Year 2 observer ratings due to rater differences ported here, which correspond to expected stability of per-
(τ12) was estimated to be 0.0594. Using the equations shown ceiver effects if we had averaged a large number of ratings of
in Figure 1, we estimated the stability of perceiver effects others. In contrast, previous investigations of perceiver-effect
over the course of a year to be a correlation of 0.69. This val- stability examined the rank-order stability of perceiver-effect
ue indicates that the average positivity of a person’s percep- estimates created by averaging a small number of rater judg-
tions of a large number of people during one year (averaged ments (between three and eight), which should serve to make
across many ratings to form a reliable estimate) would be ex- the perceiver-effect estimates quite unreliable and which
pected to correlate quite highly with the average positivity of should lead to lower stability estimates even if both measure-
the person’s perceptions a year later. ment occasions occurred nearly contemporaneously. The im-
plication that judgments of others completed a year apart cor- ers through standard response sets, such as social desirabil-
related only slightly less than did judgments completed con- ity. For instance, although academic achievement was de-
temporaneously is somewhat surprising, given the potential termined through registrar records in Study 1, it was mea-
for perceiver effects to be driven by mood effects or other ses- sured through self-reported GPA or scores on standardized
sion-specific effects. Our finding concerning the high stabili- tests in the remaining studies. Although these self-reports
ty of perceiver effects further suggests that how positively we are highly predictive of actual scores, they have nonetheless
tend to perceive others in our social environment is an impor- been linked to self-enhancement biases relative to official re-
tant traitlike individual difference in its own right. cords (Gramzow & Willard, 2006). The fact that perceiver ef-
The next question we explored was how perceptions of oth- fects correlated with how much people are liked by others
ers correlated with dispositional characteristics. This is im- and how their personality is perceived by others (Studies 2
portant because, in many frameworks, our perceptions of and 3) makes clear that perceiver-effect relationships are not
others are thought to be an important cause of our behavior due solely to measurement artifacts of self-reports; however,
(e.g., Dodge & Crick, 1990; Kelly, 1963; Reis, 2008). In each future investigations should extend efforts to link perceiver
study, we found regular relationships between personality effects to non-self-report measures. Researchers should also
traits and perceptions of others. Individuals who perceived explicitly examine the link between positive perceptions of
others positively reported higher agreeableness (particularly others and standard measures of social desirability or self-
aspects associated with friendliness and low hostility); high- enhancement biases.
er life satisfaction; and lower endorsement of measures of In our investigation, we found that perceiver effects for a
personality disorders, depression, and antisocial attitudes. wide range of traits could be explained adequately by a sin-
Individuals who perceived others more positively also tend- gle factor. In fact, despite the fact that multiple factors could
ed to be more liked by others from the group they rated, and be extracted, our explorations found that all such factors
they described their experiences in the group more positive- tended to display similar patterns of associations with dispo-
ly. The causal direction of these associations is unclear and sitional variables. In contrast, a recent investigation conduct-
should be explored in future research. However, the results ed by Srivastava and colleagues (2010) presented more op-
clearly indicate that how positively we perceive others in a timistic evidence for distinguishable dimensions of perceiv-
group shows important relationships with our general emo- er effects beyond an overall evaluation factor. Although the
tions, well-being, goals, values, and attitudes, in addition to structure of perceiver effects was not the primary focus of
our group experiences. the current investigation, the search for perceiver-effect di-
The associations between perceiver effects and personali- mensions beyond overall positivity is an interesting question
ty disorders were particularly interesting. Although theo- for future research. Given that theorists generally postulate
rists often discuss how each personality disorder may have a more complex set of cognitive biases associated with the
its own specific set of cognitive biases (Beck et al., 2004), we perception of others than a single positivity dimension (e.g.,
found that several personality disorders were associated Beck et al., 2004; Kelly, 1963), future investigators should
with a single dimension concerning how generally positive- continue examining whether a larger number of distinct per-
ly others were perceived. If perceiving others less positively ceiver dimensions can be identified that show clearly distinct
is associated with several personality disorders, this has two patterns of dispositional correlates.
important implications for the understanding of personali- Our findings concerning the structure of perceiver effects
ty disorders. First, this may indicate that perceiving others and the associations between perceiver effects and oth-
as being unfriendly, untrustworthy, unhappy, and uninter- er individual differences were quite similar across studies,
esting may be a relatively coordinated set of negative cogni- despite the fact that the targets of perception ranged from
tions that acts as a common cause to several personality dis- friends and fellow organization members to people random-
orders. If so, this may help to explain the phenomenon of co- ly assigned to one’s dormitory. However, it is fair to say that
morbidity, in which individuals are much more likely to be all of these contexts can be thought of as representing how
diagnosable with several personality disorders simultane- the individual sees ingroup members to varying degrees.
ously than would be expected by chance (e.g., Widiger et al., Given that previous studies have shown that perceiver ef-
1991). Second, if negative perceptions of others underlie sev- fects for ingroup and outgroup members can be fairly dis-
eral personality disorders simultaneously, then finding tech- tinct from one another (e.g., Boldry & Kashy, 1999), future
niques to get people to see others more positively could pro- studies might explore the dispositional correlates of perceiv-
mote the cessation of behavior patterns associated with sever effects collected from perceptions of others in a broad-
eral different personality disorders simultaneously. er range of theoretically meaningful contexts, such as out-
group members, strangers, members of the opposite sex, and
Limitations and Future Directions superiors versus subordinates. Such studies could help re-
The current investigation leaves a number of important veal some of the distinct patterns of perceiver-effect associ-
questions for future research. Although perceiving others ations with different behavioral tendencies that have been
positively was associated with a range of dispositional char- proposed by others.
acteristics, an understanding of the relations between per- Given that the present investigation was initiated first and
ceiver effects and other individual differences will be im- foremost to show the relations between perceptions of oth-
proved by obtaining a broader range of individual difference ers and dispositional characteristics, a limitation of the cur-
measures. Importantly, the majority of the individual differ- rent research is that the design of the studies did not allow
ence variables that were linked to perceptions of others were for adequate tests of the causal direction of the relationships
self-reported and thus could be linked to perceptions of oth- identified. Our results do not indicate, for instance, whether
perceiving others more positively causes people to become Journal of Personality and Social Psychology 77: 1,200-1,212.
friendlier and less aggressive toward others, more satisfied Campbell, D.T., N. Miller, J. Lubetsky & E.J. O’Connell (1964). Vari-
with their lives, and more well-liked— or whether having eties of projection in trait attribution. Psychological Monographs 78
(15, Whole No. 592).
these characteristics causes people to see others more posi-
Caprara, G.V., C. Barbaranelli, R.C. Fraley & M. Vecchione (2007).
tively. We suspect that the causal effect most likely goes in The simplicity of politicians’ personalities across political context:
both directions, and testing the causal links will require the An anomalous replication. International Journal of Psychology 42:
collection of more intensive longitudinal data. We are op- 393-405.
timistic that future research will demonstrate the power of Caspi, A., E.S. Herbener & D.J. Ozer (1992). Shared experiences and
perceiver effects to direct the development of these disposi- the similarity of personalities: a longitudinal study of married cou-
tions for at least two reasons. First, the current findings dem- ples. Journal of Personality and Social Psychology 62: 281-291.
onstrate the existence of moderate associations between pos- Christie, R., & Geis, F. L. (1970). Studies in Machiavellianism. New
York, N.Y.: Academic Press.
itive perceptions of others and dispositional variables rang-
Cronbach, L.J. (1955). Processes affecting scores on “understanding
ing from self-reported personality traits to well-being, men- of others” and “assumed similarity.” Psychological Bulletin 52: 177-
tal health, and social functioning measures. Second, in our 193.
longitudinal analyses, individual differences in how posi- Dodge, K.A., & N.R. Crick (1990). Social information processing bas-
tively others were rated showed a high level of stability. Sta- es of aggressive behavior in children. Personality and Social Psychol-
bility of environments has been identified as a characteristic ogy Bulletin 53: 1,146-1,158.
that should lead to a greater potential for environments to Goldberg, L. (1993). The structure of phenotypic personality traits.
cause dispositional change (Roberts et al., 2008). American Psychologist 48: 26-34.
Gonzalez, R., & D. Griffin (2000). On the statistics of interdepen-
The current research represents a major step in understand-
dence: treating dyadic data with respect. In: W. Ickes & S. Duck
ing the nature of perceiver effects relative to past approaches, (Eds.), The Social Psychology of Personal Relationships (pp. 181-213).
which have largely focused on assumed similarity explana- New York, N.Y.: Wiley.
tions for understanding perceiver effects. The results of the Gonzalez, R., & D. Griffin (2002). Modeling the personality of dyads
current studies provide compelling evidence that individu- and groups. Journal of Personality 70: 901-924.
als’ perceptions of others can reveal much more than merely Gosling, S., P. Rentfrow & W. Swann (2003). A very brief measure
how the individuals see themselves on the same characteris- of the Big-Five personality domains. Journal of Research in Person-
tics. Specifically, their perceptions of others reveal informa- ality 37: 504-528.
Gramzow, R., & G. Willard (2006). Exaggerating current and past
tion about their other personality characteristics, their well-
performance: motivated self-enhancement versus reconstructive
being, their mental health, their social attitudes, and how memory. Personality and Social Psychology Bulletin 32: 1,114-1,125.
they are judged by others. There are many questions that re- Graziano, W.G., J. Bruce, B.E. Sheese & R.M. Tobin (2007). Attrac-
main to be answered to demonstrate that one’s perceptions tion, personality, and prejudice: liking none of the people most of
of others truly direct developmental outcomes in the manner the time. Journal of Personality and Social Psychology 93: 565-582.
suggested by some theorists. However, the findings present- Graziano, W.G., & R.M. Tobin (2002). Agreeableness: dimension of
ed here suggest that the emphasis that various models have personality or social desirability artifact? Journal of Personality 70:
placed on perceptions of others as determinants of behavior 695-727.
Griffin, D., & R. Gonzalez (1995). Correlational analysis of dyad-lev-
may be well placed and is a fertile place for future research.
el data in the exchangeable case. Psychological Bulletin 118: 430-439.
Harms, P.D., B.W. Roberts & D. Wood (2007). Who shall lead? An in-
References
tegrative personality approach to the study of the antecedents of
American Psychiatric Association. (1994). Diagnostic and Statistical
status in informal social organizations. Journal of Research in Person-
Manual of Mental Disorders (4th ed.). Washington, D.C.: American
ality 41: 689-699.
Psychiatric Association.
Harms, P.D., D. Wood, B. Brummel & B. Roberts (2010). The Mini-
Ames, D.R., P. Rose & C.P. Anderson (2006). The NPI-16 as a short
markers of Evil. Manuscript in preparation. University of Nebras-
measure of narcissism. Journal of Research in Personality 40: 440-450.
ka—Lincoln.
Anderson, C., O.P. John & D. Keltner (2009). The Subjective Sense of
Hoch, S.J. (1987). Perceived consensus and predictive accuracy: the
Power: Structure and Antecedents. Unpublished manuscript, Univer-
pros and cons of projection. Journal of Personality and Social Psychol-
sity of California, Berkeley.
ogy 53: 221-234.
Back, M.D., J.M. Stopfer, S. Vazire, S. Gaddis, S.C. Schmukle, B. Eg-
Hoyt, W. (2000). Rater bias in psychological research: when is it a
loff & S.D. Gosling (in press). Facebook profiles reflect actual per-
problem and what can we do about it? Psychological Methods 5: 64-
sonality not self-idealization. Psychological Science.
86.
Beck, A., A. Butler, G. Brown, K. Dahlsgaard, C. Newman & J. Beck
John, O.P., & S. Srivastava (1999). The Big Five trait taxonomy: His-
(2001). Dysfunctional beliefs discriminate personality disorders.
tory, measurement, and theoretical perspectives. In: L.A. Pervin &
Behaviour Research and Therapy 39: 1,213-1,225.
O.P. John (Eds.), Handbook of Personality: Theory and Research (2nd
Beck, A. T., A. Freeman & D.D. Davis (2004). Cognitive Therapy of Per-
ed., pp. 102-138). New York, N.Y.: Guilford Press.
sonality Disorders (2nd ed.). New York, N.Y.: Guilford Press.
Kelly, G.A. (1963). A Theory of Personality: The Psychology of Personal
Beck, A., W. Rial & K. Rickels (1974). Short form of Depression In-
Constructs. New York, N.Y.: Norton.
ventory: Cross-validation. Psychological Reports 34: 1,184-1,186.
Kenny, D.A. (1994). Interpersonal Perception: A Social Relations Analy-
Beer, A., & D. Watson (2008a). Asymmetry in judgments of person-
sis. New York, N.Y.: Guilford Press.
ality: others are less differentiated than the self. Journal of Personal-
Kenny, D.A., L. Mannetti, A. Pierro, S. Livi & D.A. Kashy (2002). The
ity 76: 535-559.
statistical analysis of data from small groups. Journal of Personality
Beer, A., & D. Watson (2008b). Personality judgment at zero acquain-
and Social Psychology 83: 126-137.
tance: Agreement, assumed similarity, and implicit simplicity. Jour-
Laing, R.D. (1967). The Politics of Experience. New York, N.Y.: Pan-
nal of Personality Assessment 90: 250-260.
theon.
Boldry, J., & D. Kashy (1999). Intergroup perception in naturally oc-
Lakey, B., & P.B. Cassady (1990). Cognitive processes in perceived so-
curring groups of differential status: a social relations perspective.
cial support. Journal of Personality and Social Psychology 59: 337-343.
Lee-Chai, A.Y., S. Chen & T.L. Chartrand (2001). From Moses to Reis, H. (2008). Reinvigorating the concept of situation in social psy-
Marcos: individual differences in the use and abuse of power. In: chology. Personality and Social Psychology Review 12: 311-329.
A.Y. Lee-Chai & J.A. Bargh (Eds.), The Use and Abuse of Power (pp. Roberts, B.W., D. Wood & A. Caspi (2008). Personality development.
57-74). Philadelphia, Pa.: Psychology Press. In: O.P. John & R.W. Robins (Eds.), Handbook of Personality: Theo-
Malloy, T.E., D.B. Sugarman, R.K. Montvilo & T. Ben-Zeev (1995). ry and Research (3rd ed., pp. 375-398). New York, N.Y.: Guilford
Children’s interpersonal perceptions: a social relations analysis of Press.
perceiver and target effects. Journal of Personality and Social Psychol- Srivastava, S., S. Guglielmo & J.S. Beer (2010). Perceiving others’ per-
ogy 68: 418-426. sonalities: examining the dimensionality, assumed similarity to the
Millon, T. (1997). Millon Clinical Multiaxial Inventory-III (MCMI-III): self, and stability of perceiver effects. Journal of Personality and So-
Manual (2nd ed.). Minneapolis, Minn.: National Computer Sys- cial Psychology 98: 520-534.
tems. Vazire, S. (2006). The Person from the Inside and Outside. Unpublished
Mischel, W., & Y. Shoda (1995). A cognitive-affective system theo- doctoral dissertation, University of Texas at Austin.
ry of personality: reconceptualizing situations, dispositions, dy- Vazire, S. (2010). Who knows what about a person? The Self-Other
namics, and invariance in personality structure. Psychological Re- Knowledge Asymmetry (SOKA) model. Journal of Personality and
view 102: 246-268. Social Psychology 98: 281-300.
Pavot, W., & E. Diener (1993). Review of the Satisfaction with Life Walton, K. E., & B.W. Roberts (2004). On the relationship between
Scale. Psychological Assessment 5: 164-172. substance use and personality traits: Abstainers are not maladjust-
Pratto, F., J. Sidanius, L.M. Stallworth & B.F. Malle (1994). Social ed. Journal of Research in Personality 38: 515-535.
dominance orientation: a personality variable predicting social Widiger, T., A. Frances, M. Harris, L. Jacobsberg, M. Fyer & D. Man-
and political attitudes. Journal of Personality and Social Psychology ning (1991). Comorbidity among Axis II disorders. In: J. M. Old-
67: 741-763. man (Ed.), Personality Disorders: New Perspectives on Diagnostic Va-
Radloff, L.S. (1977). The CES-D scale: a self-report depression scale lidity (pp. 165-194). Washington, D.C.: American Psychiatric Press.
for research in the general population. Applied Psychological Mea- Winquist, L., C. Mohr & D. Kenny (1998). The female positivity ef-
surement 1: 385-401. fect in the perception of others. Journal of Research in Personality 32:
Raine, A. (2008). From genes to brain to antisocial behavior. Current 370-388.
Directions in Psychological Science 17: 323-328. Wonderlic, E. F. (1983). Wonderlic Personnel Test Manual. Northfield,
Raudenbush, S.W., & A.S. Bryk (2002). Hierarchical Linear Models: Ap- Ill.: Wonderlic & Associates.
plications and Data Analysis Methods. Thousand Oaks, Calif.: Sage. Wood, D., C. Nye & G. Saucier (2009). Derivation and Measurement of
Ready, R.E., L.A. Clark, D. Watson & K. Westerhouse (2000). Self- a Representative Set of Person-Descriptive Adjective Markers from the
and peer-related personality: agreement, trait ratability, and the English Lexicon. Unpublished manuscript, Wake Forest University,
“self-based heuristic.” Journal of Research in Personality 34: 208-224. Winston-Salem, N.C.

Fulltext

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fulltext

Uploaded by

Copyright:

Available Formats

University of Nebraska - Lincoln

DigitalCommons@University of Nebraska - Lincoln

Management Department Faculty Publications Management Department

Perceiver Effects as Projective Tests: What Your Perceptions of

Follow this and additional works at: https://digitalcommons.unl.edu/managementfacpub

Part of the Management Sciences and Quantitative Methods Commons

Keywords: Person perception; Perceiver effect; Perceiver biases; Rating bias

Perceiver effect dimension 1 2 3 4 5 1 2 3 4 5

Study 2: Ratings of dormitory floormates

Study 3: Ratings of organization members

Millon Clinical Multiaxial Inventory-III

Ratingijk = b1k(Year1) + b2k(Year2) + rijk

i = target, j = year of assessment, k = rater

b1j = γ1 + u1k, b2j = γ2 + u2k, rijk ~ N(0, σ2r)

Figure 1. Modeling the stability of perceiver effects.

Results ly intercorrelated and that relations with dispositional vari-

Correlates of the positivity of perceptions of others. We Discussion

You might also like