You are on page 1of 18

Journal of Personality and Social Psychology

1979, Vol. 37, No. 3, 395-412

A Psychological Taxonomy of Trait-Descriptive Terms:


The Interpersonal Domain
Jerry S. Wiggins
University of British Columbia, Vancouver, Canada

The eventual aim of the research reported is the development of a comprehen-


sive taxonomy of trait-descriptive terms in the English language. Building on
earlier work of Allport, Norman, and Goldberg, preliminary a priori distinctions
were made among different domains of trait categories. General procedures for
developing structured taxonomies within domains were illustrated with reference
to the interpersonal domain. Theoretical considerations dictated the definition
of the universe of content, the choice of measurement model, and the proce-
dures for classifying terms within the domain. Eight adjectival scales were de-
veloped as markers of the principal vectors of the interpersonal domain. The
substantive, structural, and psychometric characteristics of these scales were
found to be highly satisfactory. Hence, they may prove useful both as assess-
ment devices in their own right and as reference points for the classification of
variables in personality and social psychology.

Personality is that branch of psychology (proclivities, propensities, dispositions, incli-


which is concerned with providing a syste- nations) to act or not to act in certain ways
matic account of the ways in which individu- on certain occasions.
als differ from one another (Wiggins, Ren- Although such tendencies are commonly
ner, Clore, & Rose, 1971). Individuals differ called "traits," the use of that term in con-
from one another in a variety of ways: their temporary psychological discourse carries with
anatomical and physiognomic characteristics; it implications of a particular theoretical com-
their personal appearance, grooming, and mitment, a preferred method of scientific in-
manner of dress; their social backgrounds, vestigation, and a philosophical preference for
roles, and other demographic characteristics; certain kinds of explanation in theory con-
their effect on others or social stimulus value; struction. Hence, it is necessary to make it
and at any given moment in time, their clear at the outset that an interest in human
temporary states, moods, attitudes, and ac- tendencies (traits) does not imply a theo-
tivities. But the principal goal of personality retical precommitment to such issues as
study is to provide a systematic account of whether traits are manifestations of genera-
individual differences in human tendencies tive or causal mechanisms (Allport, 1937);
whether trait attributions reflect specific cog-
nitive processes of observers (Heider, 19S8);
The early stages of this research were supported by whether traits are best construed idiographi-
Grant MH 12972 from the National Institute of cally or nomothetically (Allport, 1937; Bern
Mental Health, U.S. Public Health Service (Lewis
R. Goldberg, principal investigator). More recent & Allen, 1974; Kelly, 1955); or whether con-
work was supported by Canada Council Research sistencies in human tendencies are largely due
Grant S75-0109. An earlier version of this article to environmental or situational constancies
was presented at the meeting of the Society of
Multivariate Experimental Psychology, Gleneden (Mischel, 1968).
Beach, Oregon, November 1975. In my view, consistent patterns of human
Requests for reprints should be sent to Jerry S. conduct constitute the basic data of person-
Wiggins, Department of Psychology, University of
British Columbia, Vancouver, British Columbia, ality study, which require rather than provide
Canada V6T 1WS. explanation (Wiggins, Note 1). In approach-
Copyright 1979 by the American Psychological Association, Inc. 0022-3514/79/3703-0395$00.75

395
396 JERRY S. WIGGINS

ing this task of accounting for individual think that one of the most important cate-
differences in human tendencies, I share with gories of traits is that which may be desig-
others the conviction that the natural lan- nated "interpersonal." Within the realm of
guage provides a convenient starting place things that people do to each other, it is
(Allport, 1937; Cattell, 1957; Goldberg, Note desirable to make a further theoretical dis-
2; Norman, Note 3). The universe of content tinction between interpersonal exchanges
of human tendencies is contained within the based on love and status and interpersonal
covers of an unabridged dictionary of the exchanges based on goods, money, and ser-
English language. I also share with these vices (Foa & Foa, 1974). The former are
writers the conviction that an adequate tax- called interpersonal traits and the latter ma-
onomy of trait-descriptive terms must pre- terial traits.
cede meaningful empirical studies of human Another equally "real" way in which indi-
tendencies. There have been a number of viduals are distinguishable from one another
systematic efforts to develop a personality is in terms of their styles of emotional reac-
taxonomy over the past 40 years, and my own tivity, which we refer to as "temperament."
work has attempted to capitalize on these It is also possible to distinguish individuals
earlier efforts. on the basis of the particular roles and status
Allport and Odbert (1936) examined the they hold within the framework of our social
approximately half million separate entries or institutions. Additionally, there are "charac-
derivatives included in Webster's New Inter- ter" terms that represent appraisals of an
national Dictionary (1924) for terms that individual based on a code of proper behav-
appeared "to distinguish the behavior of one ior. And finally, there are many words that
human being from that of another" (p. 24) refer to individual differences in qualities of
and identified a pool of 17,953 terms having mind as manifested in thought, perception,
this characteristic. Norman (Note 3) scanned and speech.
the entire contents of Webster's Third New Consider the following trait-descriptive ad-
International Dictionary Unabridged (1961) jectives: aggressive, miserly, lively, ceremoni-
for additional terms that had not been in- ous, dishonest, and analytical. I would classify
cluded in the Allport-Odbert list. The universe these terms as representative of the six trait
of content thus denned was approximately categories just mentioned, namely, interper-
27,000 terms. Subsequently, Norman and sonal traits, material traits, temperamental
Goldberg were able to reduce this list by traits, social roles, character, and mental
eliminating obscure, inappropriate, and ar- predicates. Here I differ from Allport, Nor-
chaic terms. The focus of all of these investi- man, and Goldberg, who view all of these
gations has been on a subset of approximately terms as "stable biophysical traits," without
3,600 terms that Allport initially called "sta- attempting to differentiate the different kinds
ble biophysical traits." Allport considered of descriptive jobs such terms perform for us.
these to be "real" traits as opposed to tem- To these earlier authors' important distinc-
porary states, moods, social roles, physical tion between stable traits and temporary
characteristics, and so forth. states, moods, social evaluations, and so forth,
However one conceives of stable biophysical I would add finer distinctions within the cate-
traits, I see them as no more nor less real gory of stable traits.
than any other kinds of human character- The present article describes our first ef-
istics. I agree with Guilford (1959) that "a forts to develop a taxonomy of personality
trait is any distinguishable, relatively en- trait-descriptive terms in the English lan-
during way in which one individual differs guage. The universe of content is taken to be
from others" (p. 6). But I see the major Norman's (Note 3) total lexicon of 18,125
taxonomic task as that of specifying the dif- terms. Within this broader framework, I in-
ferent kinds of ways in which individuals tend to focus on the "prime" categories of
differ from each other. One kind of way in Norman, which involve 4,063 relatively fa-
which individuals differ from each other is in miliar and nonobscure terms. Even restricting
terms of what they do to each other. Thus, I our attention to prime terms, we are still
TAXONOMY OF INTERPERSONAL TRAITS 397

Facet I
DIRECTION

Facet II
OBJECT

Facet III
STATUS[ [ LOVE |STATUS] JSTATUS| | LOVE | LOVE [ [STATUS
RESOURCE

INTERPERSONAL VARIABLES

NO ( Gregarious-Extroverted )
PA ( Ambitious-Dominant) +

BC ( Arrogant-Calculating ) -t-

DE ( Cold-Quarrelsome ) +

FG ( Aloof-Introverted ) -
HI (Lazy-Submissive ) + - - -
JK (Unassuming-Ingenuous ) + + - -
LM ( Warm - Agreeable) + + +

Figure 1. Facet composition of interpersonal variables (after Foa & Foa, 1974). ( + = acceptance;
— = rejection.)

confronted with a list of staggering size. Our The Domain of Interpersonal Traits
enterprise required a clear-cut game plan or
research strategy to guide us through this sea The line of reasoning to be presented origi-
of words. nated in the interpersonal theory of Sullivan
We began by defining a limited taxonomy (1953) and was later operationalized in a set
of personality measurement procedures by
on an a priori basis. We attempted to specify
Leary (1957) and his colleagues. Foa (1961,
different domains of human characteristics 1965) integrated this theory with Guttman's
based on the different descriptive jobs per- (1954) order analysis, and the general orien-
formed by the words within the domains. tation was extended to domains other than
Next, we attempted preliminary a priori classi- the interpersonal by Rinn (1965). Notable
fications within domains, mainly as a means interpersonal systems based on a circumplex
of keeping track of terms and lightening our model have also been described by Becker and
clerical burdens. The overall strategy for Krug (1964), Benjamin (1974), Lorr and
developing a taxonomy within a single domain McNair (1963), Schaefer (1959), and Stern
may be honorifically described as "iterative," (1970). This general framework was inte-
although "trial and error" may be a more apt grated with interpersonal exchange theory
(Homans, 1961) by Carson (1969) and by
descriptor. Our main concern was the avoid-
Foa and Foa (1974).
ance of premature "fixing" of a taxonomy.
In their most recent theoretical statement,
Hence, all preliminary classifications were Foa and Foa (1974) describe the development
tentative in nature and subject to continual of cognitive categories of social perception as
revision in the light of other developing cate- a progressive differentiation of structure in-
gories. This rather complex interactive strat- volving facets of directionality, object, and
egy will be illustrated with reference to the resource. A somewhat simplified version of
development of a taxonomy within the do- Foa and Foa's representation of this structure
main of interpersonal traits. is illustrated at the top of Figure 1. The earli-
398 JERRY S. WIGGINS

est cognitive schemata are based on the dis- jacent variable by only one element. This is
crimination of the directionality of social also true of the first (NO) and last (LM)
events (give vs. take, accept vs. reject). With variables, and hence the structural relations
the acquisition of the concept of social object among this set of variables can be represented
(self vs. other) four categories of social mean- as a circumplex (Guttman, 1954).
ing are discriminated (e.g., give to self, take The labels that have been attached to the
away from other). Out of an initially undif- interpersonal variables in Figure 1 (gregari-
ferentiated matrix of resource classes, services ous-extraverted, ambitious-dominant, etc.)
are differentiated from love, and the latter is are meant to capture the flavor of terms that
further differentiated into love and status. share the same profile of semantic features
With the distinction between love and status, and may serve more as tags than as definitions.
earlier facets become reorganized, and it is Thus, for example, the rather inelegant label
possible to distinguish eight features of social of "lazy-submissive" is attached to interper-
meaning (e.g., granting status to oneself, sonal transactions involving incompetence,
denying love to another). These eight features passive resistance, submission, or obedience.
may be thought of as part of a semantic code These otherwise diverse attributes share in
strip that provides the basic discriminations common the semantic features of denying
for encoding and decoding interpersonal events status to self, denying love to both self and
(Osgood, 1970). other, and granting status to other. A fuller
Within the above context, interpersonal listing of representative terms in each cate-
events may be denned as dyadic interactions gory may be found in Table 2, later in this
that have relatively clear-cut social (status) article.
and emotional (love) consequences for both Were we to collect personality measure-
participants (self and other). This definition ments on the eight variables listed in Figure
provides a theoretical basis for distinguishing 1, their intercorrelations would, in theory,
interpersonal traits from other categories of show the pattern illustrated in Table 1. The
trait descriptors, such as temperament, moods, correlation of a variable with itself is as-
cognitive traits, and physical characteristics. sumed to be unity, so that the principal di-
Under the semantic features in Figure 1 agonal of this matrix contains ones. The cor-
are listed eight theoretical interpersonal vari- relations along this main diagonal are large
ables. The organization of these variables is and positive, and they decrease across suc-
thought to be determined by an interrelated cessive minor diagonals to the w/2nd variable,
set of societal rules that impart meaning to where they are a minimum. The correlations
social events (Wiggins, Note 1). Thus, ac- then increase up to a large positive value in
tions that have the same profile of semantic the off-diagonal matrix. The circumplexity of
features are categorized as belonging to the this or any other set of variables can be
same response class. The semantic features of evaluated directly from the intercorrelation
each of the eight variables appear as rows in matrix. A rigorous procedure for assessing the
Figure 1. Note that the first variable (NO) circumplexity of a set of variables involves
is coded on all the positive (accept) categories plotting the correlations of each variable
for self and other with respect to both love (ordinate) with the other variables in se-
and status. The first variable is in marked quence (abscissa). This procedure should gen-
contrast to FG for which all values are nega- erate a series of overlapping sine curves that
tive (reject). Variables NO and FG have no can be evaluated for goodness of fit (Stern,
features in common, but since acceptance and 1970).
rejection are logically opposed concepts, it An alternative procedure for evaluating cir-
would be expected that the two variables cumplexity is to extract the first two principal
would be strongly negatively correlated. Simi- components from the matrix of intercorrela-
lar relationships can be seen to exist between tions and to examine the plot of the variables
PA and HI, BC and JK, and DE and LM. Note on the two components. Figure 2 presents the
also that each variable differs from its ad- plot of the eight variables on two principal
TAXONOMY OF INTERPERSONAL TRAITS 399

Table 1
Hypothetical Correlations Among Eight Interpersonal Variables

Variable PA BC DE FG HI JK LM NO

PA 1.00
BC .50 1.00
DE .00 .50 1.00
FG -.50 .00 .50 1.00
HI -1.00 -.50 .00 .50 1.00
JK -.50 -1.00 -.50 .00 .50 1.00
LM .00 -.50 -1.00 -.50 .00 .50 1.00
NO .50 .00 -.50 -1.00 -.50 .00 .50 1.00

Note. PA = ambitious-dominant; BC = arrogant-calculating; DE = cold-quarrelsome; FG = aloof-


introverted; HI = lazy-submissive; JK = unassuming-ingenuous; LM = warm-agreeable; NO = gre-
garious-extraverted.

components extracted from the intercorrela- There are two distinct advantages to the
tions in Table 1. As can be seen from this representation of interpersonal variables by a
figure, the intercorrelations among variables two-dimensional circumplex. The first is that
in Table 1 form a perfect, equally spaced cir- it provides an explicit conceptual definition
cumplex. This pattern follows, both theoreti- of the universe of content of interpersonal be-
cally and empirically, from the pattern of havior. Any behavior that meets the defini-
shared semantic features among the variables tion of a meaningful interpersonal event given
illustrated at the bottom of Figure 1. Variables above must be capable of being represented as
that share three features in common are ad- a vector originating from the center of the
jacent to each other on the circle; variables circle. Thus, the specific system proposed
that have no features in common are opposite here is potentially falsifiable. The second ad-
each other. vantage of the circumplex model is that it
AMBITIOUS
DOMINANT
(PA)
o

ARROGANT - GREGARIOUS -
CALCULATING 0 o EXTRAVERTED
(BC) (NO)

COLD - WARM -
QUARRELSOME o o AGREEABLE
(DE) (LM)

ALOOF - UNASSUMING
INTROVERTED o o INGENUOUS
(FG) (JK)
o
LAZY-
SUBMISSIVE
(HI)

Figure 2. Perfect circumplex of interpersonal variables.


400 JERRY S. WIGGINS

alerts the investigator to noticeable "gaps" in of the taxonomy, the principal distinctions
the interpersonal space of a given set of vari- were between interpersonal trait terms, tem-
ables. Strictly empirical procedures of vari- peramental trait terms, and mental predicate
able selection are likely to deemphasize the terms. Tentative categories of "attitudinal
importance of certain variables that are im- terms" and "constancy" were employed at this
plied by the logic of the circumplex system point, as well as a "miscellaneous" category
but that are underrepresented in the English that involved various dimensions that could
language. Although the system of interper- not be clearly classified into other categories.
sonal variables under discussion is limited to In the initial sorting of the 1,710 adjectives,
eight variables, it could in principle be equally approximately 800 terms were identified as
well represented by 16, 32, or 64 variables. "interpersonal" under our working definition
The thinness with which we slice the circum- of interpersonal traits.
plex pie is limited by the reliability with For purposes of preliminary categorization,
which respondents can distinguish between we selected Leary's (1957) system of inter-
closely synonymous words or phrases. personal traits, because it appeared to be the
It should be evident that the preceding most explicit system described in the litera-
theoretical considerations substantially con- ture, an entire book being devoted to the topic.
strain the final form that a taxonomy of in- Two colleagues 1 and I thoroughly familiarized
terpersonal traits may assume. The definition ourselves with the system and its theoretical
of interpersonal events, the specification of framework. Working as a team, we consid-
their underlying facet structure, and the selec- ered each of the approximately 800 terms
tion of a measurement model to represent re- previously classified as interpersonal and at-
lationships among variables all express a tempted to classify each within one of the 16
commitment to a particular, albeit widely interpersonal vectors. Adjectives that could
shared, conceptualization of the domain under not be classified within the 16 dimensions and
investigation. In this sense, the eventual tax- that did not appear to belong in one of the
onomy of trait-descriptive terms will be a other five a priori categories were temporarily
"psychological" taxonomy rather than a set aside for further analysis. There were re-
strictly "semantic" taxonomy based on dic- markably few such adjectives. With consid-
tionary definitions. It is assumed that the erable effort we were able to distribute 567
semantic structures underlying social percep- adjectives across the 16 categories with unani-
tion in this culture cannot be inferred in any mous agreement among three raters.
obvious way from dictionary definitions. In-
vestigators who start with different assump- Selection oj Preliminary Markers
tions and who utilize different measurement
models would undoubtedly devise somewhat Social desirability scale values for all ad-
different taxonomies. Structural relationships jectives were available from a previous study
of the kind at issue here are not "discovered" by Norman (Note 2), and mean self-endorse-
(Loevinger, 1957). They are postulated and ment frequencies were available from a more
then evaluated for goodness of fit. recent study by Goldberg (Note 2 ) . On the
basis of these itemmetric data, items were
selected to serve as preliminary nuclear
Development of Interpersonal Clusters clusters within each of the 16 interpersonal
categories. These items were selected in such
Rational Categorization
a way as to represent the category unambigu-
In a preliminary attempt to make the dis- ously across the range of endorsement fre-
tinction among kinds of trait terms discussed quencies and desirability values. Approxi-
above, I classified all of the trait descriptors
in Goldberg's (Note 2) pool of 1,710 adjec- 1
1 am grateful to James M. Kilkowski and Alex-
tives into one or another of seven a priori ander Galvin for their help and support in this
categories. At that stage in the development enterprise.
TAXONOMY OF INTERPERSONAL TRAITS 401

mately 12 items were selected for each cate- the method of principal components. Figure
gory. 3 displays the pattern of octant variables
In Goldberg's sample of 70 male and 117 loading on the first two factors in the Univer-
female University of Oregon undergraduates, sity of British Columbia sample. The pattern
self-ratings were obtained on a 9-place scale of loadings for 16 interpersonal variables is
for each of the 1,710 trait descriptors. Using considerably less orderly than this, due no
Goldberg's data, our next step was to obtain doubt to the fact that the vector scores are
the intercorrelations among terms selected to based on fewer items in a relatively small
be preliminary markers of each of the cate- sample of subjects. But the pattern of octant
gories of interpersonal behavior. By a cluster- scores clearly displays some of the difficulties
ing procedure, a smaller and more homoge- we encountered in our attempt to map out
neous subset of items was selected within each Leary's (1957) system.
category, with approximately six items per The most striking feature of Figure 3 is
cluster. the lack of variables in the upper right-hand
The next step was to consider all items quadrant. This gap in the Leary (1957) cir-
that had been rationally classified as falling cumplex has been previously noted by several
within the 16 categories as potential candi- authors including Stern (1970), who felt that
dates for addition to or deletion from each the failure of Octants PA and NO to "close"
nuclear interpersonal cluster. For this pur- raises the question of whether the Leary sys-
pose, we scored the preliminary clusters as 16 tem is a circumplex. Lorr and McNair (1965)
scales and examined the correlations of indi- noted this gap also and attempted to close it
vidual items with these scales with an eye with additional substantive variables. An-
toward their circumplex patterning. Thus, we other notable departure from expectations is
examined the correlation of each of the 567 the location of Octant NO, which not only
items with the 16 preliminary clusters ordered fails to appear in the upper quadrant but
by their hypothetical position in the circum- appears below Octant LM. In a sense, then,
plex. Items were sought that had positive cor- we partially succeeded in replicating the
relations with adjacent clusters, zero correla- Leary system with trait-descriptive adjectives,
tions with orthogonal clusters, and negative but in so doing we carried over the faults of
correlations with opposite clusters. For most the system as well.2
items, this circumplex patterning across 16
clusters was far from perfect. Item selection Development oj Bipolar Clusters
was further complicated by the fact that some
of the initial nuclear clusters were clearly out At this point it occurred to us that a 16-
of place on the circumplex, so that we con- variable circumplex can be rather easily con-
stantly had to keep in mind the inadequacies structed from eight genuine bipolar dimen-
of our original clusters. By this bootstrap sions. One of the conceptual difficulties we
procedure, we selected a set of eight adjectives experienced in working with the Leary sys-
for each of the 16 vectors that, we hoped, tem was a decided lack of bipolarity between
would increase the homogeneity of their vector vectors that appeared opposite each other on
and correct the previous shortcomings of that the circle. In particular, the following im-
vector. plicitly bipolar contrasts did not make a
We assembled the 128 adjectives chosen to great deal of sense: success versus masochism,
mark the 16 vectors into a test format and narcissism versus conformity, rebellion versus
administered it to a small group of students
from the University of British Columbia. The 2
Juris I. Berzins (personal communication, March
students were requested to rate the self-ap- 8, 1977) plotted the loadings of the original Leary
plicability of the adjectives on a 9-place Likert (1957) Interpersonal Checklist (ICL) octants on the
scale. Item responses were summed for both first two principal components in samples of 685
high school students and 1,109 college students. The
octant and sixteenth variables. Intercorrela- plots for both samples are indistinguishable from
tions among the variables were factored by Figure 3. (See also Rinn, 1965, p. 458, Figure 6.)
402 JERRY S. WIGGINS

succ;ESS-
POVVER
(PA )
)
NARCISSISM -
EXPLOITATION o
(BC)

PUNISHMENT - COLLABORATION •
HOSTILITY LOVE
0
( DE ) ° (LM)

REBELLION - TENDERNESS -
DISTRUST GENEROSITY
(FG) ( NO)

CONFORMITY
o TRUST
( JK)

MASOCHISM -
WEAKNESS
(HI)

Figure 3. Rotated components of Leary (1957) interpersonal variables.

tenderness, distrust versus generosity, and theoretically orthogonal clusters. Thus, a


punishment versus collaboration. good dominant item should have a high posi-
In Leary's (1957) original system the ten- tive correlation with a dominant cluster, a
derness-generosity octant (NO) was concep- high negative correlation with a submissive
tualized as the bipolar contrast to the rebel- cluster, and essentially zero correlations with
lious-distrustful octant (FG). This likely ac- quarrelsome and agreeable clusters. We used
counts for the noticeable gap in the upper a tentative set of 16 four-item bipolar clusters
right quadrant of the system (Figure 3). as markers to select items having the prop-
Tenderness and generosity are not bipolar erties just described. This enabled us to re-
contrasts to rebellious and distrustful behav- vise and expand our tentative clusters into
ior. Tenderness and generosity are too weak eight-item variables. Factor analysis of these
and loving to be placed this high on the circle. eight-item variables in Goldberg's (Note 2)
As Lorr and McNair (1965) noted earlier, sample of 187 subjects revealed the clearest
the NO octant, which falls between dominance circumplex structure we had yet encountered
and love, reflects a socially exhibitionistic in our own work or in the literature. This re-
style of behaving (gregarious-extraverted). sult was replicated in an additional sample
With a little effort, our revised representa- of 119 subjects from the University of British
tion of the 16 interpersonal variables can be Columbia.
read from Figure 2. Each variable represents
a bipolar contrast to the variable appearing Development of Final Scales
opposite it on the circle. The contrasts are
dominant-submissive, arrogant-unassuming, The items in our eight-item bipolar inter-
calculating-ingenuous, cold-warm, quarrel- personal clusters were developed from the
some-agreeable, aloof-gregarious, introverted- pool of 567 adjectives classified as interper-
extraverted, and ambitious-lazy. sonal in our initial rational sorting of the
In developing our bipolar clusters, we at- total list of 1,710 adjectives. The possibility
tempted to select items that were highly nega- clearly existed that some adjectives from
tively correlated with their opposite cluster among the 1,143 not considered would be
and that had zero correlations with their appropriately placed within the 16 interper-
TAXONOMY OF INTERPERSONAL TRAITS 403

sonal categories. Consequently, a program were unclassifiable in any category, and ad-
was written to examine the relationships be- jectives that were candidates for deletion on
tween all 1,710 adjectives and the 16 bipolar the grounds of ambiguity, obscurity, or lack
interpersonal clusters. The program correlates of personological relevance. In addition, all
each adjective with the 16 bipolar clusters, adjectives in the interpersonal category other
selects the cluster with which the adjective than the 400 ( 1 6 X 2 5 ) marker adjectives
is most highly correlated, and then prints out were classified as doubtful for the purpose of
an ordered list of variables within each of the this analysis. There were 768 such adjectives
16 categories. This listing provided the cor- designated as questionable by this criterion.
relation of each item with the cluster with Each of the 768 adjectives was correlated with
which it was most highly correlated. It also all of the 54 taxonomic category scales. Two
provided the correlation with the bipolar op- of us then examined this pattern of correla-
posite cluster and with the two orthogonal tions for all 768 words and assigned them to
clusters. This procedure produced 16 lists one of the 54 categories on the basis of both
that located all 1,710 adjectives with respect conceptual and empirical (correlational) con-
to the interpersonal cluster with which they siderations. Some items were retained in their
were most highly correlated. original interpersonal category, others moved
Two of us then went through these lists to other categories, and some deleted from
and examined the correlation of each item with further consideration. By these procedures, a
the cluster from the category to which it had revised 54-category taxonomy was developed.
been assigned, as well as the correlation of Within this revised taxonomy, 864 adjectives
the item with its opposite and orthogonal were classified as interpersonal.
clusters.3 Items that clearly belonged in the The selection of items for inclusion in the
interpersonal category to which they had been final interpersonal trait scales involved one
assigned were retained. Items that did not more cycle in our iterative procedures The
appear to belong in the interpersonal category 2S "best" items previously identified in each
to which they were assigned were placed in of the 16 interpersonal categories were scored
one of eight other taxonomic categories that as reference clusters. These reference clusters
then existed (e.g., temperament, character, provided a broader representation of the con-
material traits, etc.). tent of the final interpersonal variables than
At this point, an attempt was made to de- did our earlier clusters. Within each category,
velop more refined subcategories within the each item was correlated with its own 2 5-item
broader taxa. Preliminary groupings of clusters cluster and with its 25-item opposite and
were formed within the domains of tempera- orthogonal clusters. The items were then
ment, character, attitudes, mental predicates, ordered within each category by their cor-
and social roles. This classification added 38 relations with their own clusters.
new clusters to the 16 interpersonal clusters, Working independently, three of us selected
making a total of 54 categories within the the "best" eight items in each of the 16 in-
preliminary taxonomy. These 54 categories terpersonal categories. One rater made selec-
were scored as scales by computing, for each tions strictly on the basis of a summary nu-
subject, the sum of his or her ratings to the merical index based on item correlations with
items in that category. In the case of the 16 same, opposite, and orthogonal clusters, a
interpersonal categories, the 25 best items, as procedure referred to as empirical. Another
determined from item-cluster correlations, rater attended to the same item correlations,
form the reference scale for each category. but attempted to choose the more "meaning-
The next stage of analysis was designed to ful" adjectives from pairs that had roughly
determine more exactly the classification of similar empirical characteristics, a procedure
doubtful adjectives in the preceding taxonomic
sort. This included adjectives that were in- 3
I would like to acknowledge the substantial help
terpersonal but that seemed to be in the of Ana Holzmuller in this and many other phases of
wrong interpersonal category, adjectives that the project.
404 JERRY S. WIGGINS

designated quasi-empirical. The third rater Australian samples, suggested that the com-
selected sets of eight adjectives that were bination set of 128 adjectives had a slight
"psychologically cohesive" in terms of his edge over the item-ordering, empirical, and
conception of the constructs under investiga- quasi-empirical procedures and a substantial
tion, the rational procedure. advantage over the rational procedure, which
The empirical and quasi-empirical item- fared rather badly. This combination set of
selection procedures yielded sets of 16 scales adjectives, which has been employed in all
with highly similar psychometric properties. subsequent investigations, is listed in Table 2.
For a specific sixteenth, it was possible to
identify one or the other of the two scales Generalizability of Circumplex Structure
as being preferable, either on grounds of a
higher alpha coefficient or on the grounds of The present taxonomy of the interpersonal
a better position in the 16-variable circum- domain is based on an explicit structural
plex plot. Consequently, a combination set of model (Guttman, 1954) that follows from a
8 variables was formed from the best of the facet analysis of cognitive categories of so-
empirical and quasiempirical sixteenths. cial perception (Foa & Foa, 1974). On the
An additional procedure for item selection basis of both theoretical and psychometric
was based on a more fine-grained analysis of considerations, a set of eight 16-item scales
the circumplex properties of individual items. were developed as marker variables of the
Each item was correlated with the 16 25-item principal vectors of this system. In addition
clusters that served as markers of the inter- to being potentially useful personality assess-
personal variables. This pattern of correla- ment measures in their own right, these scales
tions was then inspected for the number of enable one to classify any interpersonal trait
departures from perfect circumplex ordering descriptor by establishing its location within
that occurred. Items with the smallest num- the circumplex space. However, the mean-
ber of departures from this pattern were re- ingfulness of such classification depends
tained in a set of scales labeled "item order- heavily on the generalizability of the present
ing." circumplex structure to samples other than
The final selection of a 128-item set of in- those involved in the derivation of the eight
terpersonal variables was based on an index interpersonal scales. The scales were derived
of circumplexity that permits comparison primarily from Goldberg's (Note 2) sample
among competing scale sets (Wiggins & Mar- of American students and cross-validated at
ston, Note 4). Very briefly, two factors were various stages in samples of Canadian stu-
extracted from the intercorrelations among dents and in Norman's sample of Australian
interpersonal variables in a given scale set. A students. Although relatively diverse samples
hypothetical factor matrix was constructed to of students were employed in scale deriva-
represent a perfect circumplex solution for tion, evidence for the generalizability of the
variables of this level of reliability (as esti- circumplex structure must come from other
mated from communalities). The sum of the sources. Four samples of subjects were tested
absolute differences between elements in the for this purpose.
hypothetical factor matrix and elements in
the obtained factor matrix was taken as an
index of circumplexity. Each of the five sets Cross-Validation Samples
of interpersonal scales was then compared on Sample A. Subjects were recruited by ad-
this index. Both octants and sixteenths were vertisements to participate in a "personal-
evaluated in Goldberg's (Note 2) sample of ity study" that involved two separate testing
187 University of Oregon undergraduates and sessions of approximately 2 hours each
in Norman's sample of 123 University of (Marston, Note 5). Subjects were paid $10
Western Australia undergraduates, An over- each for their participation and were provided
all index of circumplexity, based on both 8- individual feedback if requested. The 128 in-
and 16-variable solutions in American and terpersonal adjectives were embedded in a
TAXONOMY OF INTERPERSONAL TRAITS 405

Table 2
Final Set of Interpersonal Adjective Scales

P. Ambitious A. Dominant B. Arrogant C. Calculating


(Success) (Power) (Narcissism) (Exploitation)
Persevering Dominant Bigheaded Sly
Persistent Assertive Boisterous Tricky
Industrious Forceful Conceited Wily
Self-disciplined Domineering Boastful Cunning
Organized Firm Overforward Overcunning
Deliberative Self-confident Swellheaded Crafty
Stable Self-assured Cocky Calculating
Steady Un-self-conscious Flaunty Exploitative
D. Cold E. Quarrelsome F. Aloof G. Introverted
(Hate) (Hostility) (Disaffiliation) (Withdrawal)
Warmthless Impolite Antisocial Silent
Unsympathetic Uncordial Unneighborly Shy
Ironhearted Discourteous Impersonal Introverted
Uncharitable Ungracious Unsociable Bashful
Coldhearted Disrespectful Distant Inward
Hardhearted Uncooperative Dissocial Unrevealing
Cruel Ill-mannered Unsmiling Unsparkling
Ruthless Uncivil Uncheery Undemonstrative
H. Lazy I. Submissive J. Unassuming K. Ingenuous
(Failure) (Weakness) (Modesty) (Trust)
Unproductive Self-doubting Nonegotistical Uncunning
Lazy Self-effacing Undemanding Uncalculating
Unthorough Timid Unvain Uncrafty
Unindustrious Meek Unwild Unwily
Inconsistent Unbold Unargumentative Unsly
Disorganized Unaggressive Boastless Guileless
Unbusinesslike Forceless Pretenseless Undevious
Impractical Unauthoritative Conceitless Undeceptive
L. Warm M. Agreeable N. Gregarious O. Extraverted
(Love) (Collaboration) (Affiliation) (Outgoingness)
Tenderhearted Courteous Friendly Outgoing
Gentlehearted Charitable Genial Extraverted
Tender Well-mannered Neighborly Vivacious
Kind Respectful Companionable Jovial
Emotional Cordial Approachable Enthusiastic
Sympathetic Cooperative Congenial Cheerful
Softhearted Accommodating Good-natured Perky
Appreciative Forgiving Pleasant Unshy

list of 60S adjectives whose order was ran- the Greater Vancouver area (Merritt, Note
domized separately for each subject. Four 6). The 128 interpersonal adjectives were ad-
personality inventories were administered in ministered in a single booklet along with a
varying orders along with the total list of value survey and a standardized personality
adjectives. The 152 subjects (51 men and inventory. Testing was accomplished indi-
101 women) included both undergraduate and vidually, in small groups, or on a take-home
graduate students from a variety of academic basis. One hundred subjects (29 men and 71
majors at the University of British Columbia. women) completed the interpersonal adjec-
Sample B. Representative samples of vol- tive form.
unteer and professional workers were recruited Sample C. Students in the second term of
from three different social service agencies in an introductory psychology class at the Uni-
406 JERRY S. WIGGINS

SAMPLE A SAMPLE B

PA PA
o o
o NO
BC o o NO
BC o
o LM
DE o

DE o
FG o o JK
o JK
FG o o o
HI
HI

PA PA
o o

BC o BC o ONO
o NO

o LM DE o o LM
DE o

FG o
O JK FG o JK
o O
HI HI

SAMPLE C SAMPLE D
Figure 4. Structure of final interpersonal variables in four samples. (PA = ambitious-dominant;
BC = arrogant-calculating; DE — cold-quarrelsome; FG = aloof-introverted; HI = lazy-submis-
sive ; JK = unassuming-ingenuous; LM — warm-agreeable.)

versity of British Columbia were administered cipal components were then extracted from
the 128 adjectives, embedded in a larger list each of the intercorrelation matrices. In the
of adjectives, during a regular class period. respective analyses, these two components ac-
The students were given a guest lecture on counted for 76.1% (Sample A), 67.9% (Sam-
personality testing in exchange for their par- ple B, 72.0% (Sample C), and 74.8% (Sample
ticipation. Data were collected for 132 sub- D) of the total variance. For comparability
jects (57 men and 75 women). of inspection, the second component of each
Sample D. Summer students enrolled in solution was hand rotated to pass through
an introductory psychology class at the Uni- the ambitious-dominant vector (PA). Figure
versity of British Columbia were administered 4 depicts the loadings of the eight interper-
the 128 adjectives, embedded in a larger list, sonal scales on two hand-rotated principal
during a regular class period. The students components in each of four samples.
later received their own standardized scores Perfect, evenly spaced circumplexes (Fig-
on the interpersonal variables and a lecture ure 2) are not expected in real data, because
on interpersonal behavior. Data were collected of measurement error, and hence the struc-
for 139 subjects (58 men and 81 women). tures in Figure 4 are more properly described
as quasi-circumplexes (Guttman, 1954). Al-
Circumplex Analyses though the amount of "quasiness" one is
Intercorrelations were obtained among the prepared to tolerate in empirical data is to
eight interpersonal adjective scales in each of some extent a matter of taste, it seems fair
the four samples just described. Two prin- to say that the quasi-circumplex structures in
TAXONOMY OF INTERPERSONAL TRAITS 407

Table 3
Psychometric Characteristics of Interpersonal Adjective Scales
Internal consistency
(coefficient alpha)
Social desirability Range of
Men Women Total rating values
(n = 236) (n = 374) (N = 610) (N = 100) Total from the

label M SD M SD M SD M SD (N = 610) samples

PA 5.87 .95 5.72 1.01 5.79 .99 6.77 1.10 .855 .831-.880
BC 4.30 1.02 3.60 1.07 3.87 1.05 3.66 .98 .870 .84S-.885
DE 2. 95 .89 2.48 .80 2.66 .84 2.31 .81 .877 .862-.8S9
FG 4.28 1.19 3.87 1.17 4.03 1.17 3.54 .71 .891 .887-.S94
HI 3.95 .94 4.13 1.01 4.06 .99 3.37 .61 .809 .769-838
JK 4.64 .93 5.14 1.00 4.95 .97 5.73 .95 .801 .743-.840
LM 6.64 .80 7.08 .76 6.91 .77 7.53 .73 .865 .841-.879
NO 6.17 .96 6.54 .96 6.40 .96 7.52 .66 .897 .887-.917

Note. Scale labels stand for ambitious-dominant (PA); arrogant-calculating (BC); cold-quarrelsome (DE); aloof-introverted
(FG); lazy-submissive (HI); unassuming-ingenuous (JK); warm-agreeable (LM); and gregarious-extraverted (NO).
ft
Scale values range from 1 (extremely inaccurate description) to 9 (extremely accurate description).

Figure 4 are among the clearest reported in Sex differences in self-report on the inter-
the personality literature to date. The cir- personal circumplex provide a graphic repre-
cumplex ordering obtained in scale derivation sentation of sex role stereotypes in North
samples is generalizable across relatively di- American society (Wiggins & Holzmuller,
verse samples of subjects tested in differing 1978). These differences are sample dependent
contexts. Hence, the eight scales may prove and may possibly decrease in future years.
useful both as measuring instruments and as Samples of university students are not uni-
reference points for the classification of items formly "stereotyped." They contain differing
and scales in the interpersonal domain. proportions of complexly "sex-reversed" sub-
jects, such as men who score high on lazy-
submissive and on aloof-introverted and
Psychometric Characteristics
women who score high on ambitious-dominant
Sex Differences and on gregarious-extraverted (Wiggins &
Holzmuller, Note 7).
Table 3 presents normative and psycho-
metric data for each of the eight interpersonal Social Desirability
adjective scales. The first six columns con-
tain means and standard deviations in a com- Social desirability ratings of the individual
bined sample of 610 North American univer- adjectives were obtained from Norman's
sity students. Inspection of the separate means monograph (Note 3). One hundred Univer-
for men and women reveals clear-cut, and to sity of Michigan undergraduates (SO men and
some extent predictable, sex differences in SO women) rated the adjectives on a 9-place
self-report. In comparison with women, men social desirability rating scale. The means
present themselves as more ambitious-domi- and standard deviations for the 16 adjectives
nant, arrogant-calculating, cold-quarrelsome, in each scale appear in columns 7 and 8 of
and aloof-introverted. Women present them- Table 3. Comparing the total-sample mean
selves as more lazy-submissive, unassuming- self-report scale scores with corresponding
ingenuous, warm-agreeable, and gregarious- mean social desirability ratings, it is clear that
extraverted. All of these mean differences are the well-known relationship between endorse-
statistically reliable. Surprisingly, the smallest ment and desirability is found within the in-
differences occur on ambitious-dominant (p terpersonal domain (Edwards, 19S7a). More-
< .03) and lazy-submissive ( / > < . 0 1 ) ; all over, the similarity between mean endorse-
other differences are highly reliable (p < ment and mean desirability with respect to
.0001). the patterning or ordering of interpersonal
408 JERRY S. WIGGINS

variables is open to alternative interpretations cipal itemmetric criterion was that of struc-
of the circumplex structure in terms of stylistic tural fidelity (Loevinger, 1957). By a variety
response variables.4 of procedures, we attempted to ensure that
A variety of procedures have been sug- each item selected was properly located with
gested for coping with the endorsement-de- reference to a circumplex that we had adopted
sirability confound in self-report personality as the most appropriate measurement model
data. On the assumption that in principle, re- for representing the relationships among cate-
sponses within all interpersonal vectors should gories of interpersonal perception. Items were
be equiprobable, LaForge and Suczek (1955) also selected that had relatively high correla-
tried to write desirable phrases for undesir- tions with appropriate preliminary scales.
able dimensions ("Can be strict if necessary") Final scales thus constructed are likely to
and undesirable phrases for desirable dimen- be internally consistent, although this must
sions ("Spoils people with kindness"). Jack- be demonstrated rather than assumed.
son's (1970) differential reliability index re- The next-to-last column of Table 3 presents
moves desirability variance and enhances con- coefficient alpha (Cronbach, 1951) estimates
tent saturation at the stage of item selection. of internal consistency for each of the inter-
Norman (Note 9) has described analysis of personal scales in a combined sample of uni-
variance procedures for removing desirability versity students. The final column of the table
variance from the Subject X Item response indicates the range of alpha coefficients ob-
matrix. Finally, several individual difference tained in the four samples that constituted the
measures of social desirability response style combined student group. Although some of
have been constructed to permit the assess- the interpersonal scales are more reliable
ment of desirability responding for each sub- than others, all of them meet a reasonably
ject (e.g., Crowne & Marlowe, 1960; Edwards, stringent requirement of acceptable internal
1957b; Wiggins, 1959). consistency (a > .80). The range of alpha
The extent to which an investigator may coefficients in the different samples tends to
feel the need to "correct for" the ubiquitous be small, and this, together with the repre-
endorsement-desirability confound will vary sentativeness with which the universe of con-
with the purpose of the investigation and tent was sampled, makes it reasonable to as-
with the investigator's theoretical stance on sume that the mean alpha values reported
the meaning of the evaluative dimension of are good estimates of the characteristic degree
affective meaning. A set of interpersonal vari- of interitem structure for the different inter-
ables that did not differ in desirability (or personal variables (Loevinger, 1957). In this
did not reveal sex differences) would be a respect, the extraversion-introversion dimen-
feeble representation of real-life categories sion (NO-FG) is the most internally cohesive,
of social perception. On the other hand, basic and the variables of unassuming-ingenuous
investigations of the process of responding to (JK) and lazy-submissive (HI) are the least.
personality inventory items may require vari-
ance decomposition procedures such as those Discussion
of Norman (Note 9). In any event, interpreta-
The long-range goal of the research re-
tions of individual or group scores on the in-
ported in this article was the development of
terpersonal scales should be made with refer-
a psychological taxonomy that would encom-
ence to normative data of the kind provided
pass the approximately 4,000 relatively fa-
in Table 3. Whether a person or a group
miliar trait-descriptive terms identified by
scores high on warm-agreeable or low on
cold-quarrelsome can be judged only with 4
For example, Douglas N. Jackson (personal com-
reference to the scores of others. munication, November 27, 1975) is convinced that
the present interpersonal circumplex structure, as
Internal Consistency well as others reported in the literature, can be
accounted for in terms of his threshold theory of
In the selection of item sets to mark the desirability responding (Rogers, 1971; Jackson, Note
vectors of the interpersonal domain, the prin- 8).
TAXONOMY OF INTERPERSONAL TRAITS 409

Norman (Note 3). The research strategy rather than as precise logical distinctions. The
adopted for this task relied heavily on cer- circumplex model appears to be particularly
tain a priori distinctions between different well suited for representing elements (in the
domains of human characteristics. In the pres- present case, adjectives) whose class mem-
ent work, the domain of interpersonal traits bership is continuous rather than discrete.
was defined in such a way as to distinguish Although often working in apparent isola-
it from other domains such as temperament, tion from one another, numerous investigators
character, moods, and cognitive traits. These, have proposed highly similar models of inter-
or similar, distinctions have been made by personal behavior for the study of parents
most personality theorists concerned with (Chance, 1959; Roe & Siegelman, 1963;
taxonomic issues (e.g., Allport, 1937; Cattell, Schaefer, 1959), children (Baumrind & Black,
1950; Murray, 1938). However within the 1967; Becker & Krug, 1964; Schaefer & Bay-
interpersonal domain, the adoption of a ley, 1963), parents and children (Benjamin,
two-dimensional real-space circumplex model 1974; Foa, Triandis, & Katz, 1966), normal
(Benjamin, 1974) to capture the interrela- and abnormal adults (Leary, 1957; Lorr &
tionships among trait terms represents a per- McNair, 1963), psychotics (Lorr, Klett, &
sonal methodological preference that is per- McNair, 1963), and college students (Stern,
haps not as widely shared. On the basis of 1970). Within many of these diverse pro-
the work reported in this article, it seems grams of research, the circumplex model has
fair to conclude that the domain of interper- provided a nomological network that has
sonal trait descriptors can be classified, quite enhanced the meaning and significance of
precisely, with reference to a circumplex model the separate interpersonal variables employed
—a conclusion that was not obvious when we (Schaefer, 1961). Unfortunately, the potential
began. Since there are, no doubt, many other of the circumplex as an integrative conceptual
ways in which the interpersonal domain could model has not been as widely recognized by
be classified, under different definitions and those whose research paradigms have focused
different models, the specific advantages of the on the intensive study of single dimensions in
present framework must be discussed in more personality and social psychology.
detail. In a recent book that presents summaries
Two-dimensional circular orderings of inter- of research on the major dimensions of per-
personal variables have been reported in the sonality, the editors state:
literature for more than 40 years (Schaefer,
There obviously has been no overarching plan or
1961), and the general conceptual model un- theory, implicit or explicit, guiding the selection of
derlying this structure can be traced as far topics for trait researchers. Indeed, the editors were
back as Galen (Roback, 1928). Remarkable forced to organize the book by means of the un-
"convergences" in conception and structure sophisticated tactic of simply placing the chapters in
have been noted for studies varying widely alphabetical order. (London & Exner, 1978, p. xiv)
in variables, populations, and measurement But surely the major research topics in con-
procedures (Foa, 1961; Schaefer, 1961; Wig- temporary personality and social psychology
gins, 1968). These convergences do not stem such as achievement and power (PA), Machi-
from the similarity of generative mechanisms avellianism (BC), aggression (DE), intro-
postulated by different theorists (needs, in- version (FG), obedience (HI), interpersonal
terests, dynamisms, etc.) to account for sur- trust (JK), altruism and helping behavior
face patterns. Instead, the convergences re- (LM), affiliation and extraversion (NO)—
flect a set of semantic categories that impart can be related to each other by schemes more
a common meaning to social events observed sophisticated than the alphabet. This is not
by a variety of procedures. The categories to claim an established isomorphism between
distinguished by the ordinary language of per- the research topics just enumerated and the
sonality are not sharply demarcated, and present interpersonal taxonomy; rather, it is
they are perhaps best thought of as "fuzzy to illustrate the kinds of questions whose
sets" (Zadeh, Fu, Tanaka, & Shimura, 1975) answers might bring conceptual clarity to
410 JERRY S. WIGGINS

both taxonomic and experimental research. by Plutchik and Platman (1977) suggests a
For example, once it is recognized that the possible reason why this might be the case.
dimensions measured by Bern's (1974) mas- Twelve trait-descriptive adjectives were se-
culinity and femininity scales are the or- lected to represent an interpersonal circum-
thogonal interpersonal dimensions of ambi- plex (Schaefer, 1961). Psychiatrists were
tious-dominant (PA) and warm-agreeable then asked to rate seven diagnostic categories
(LM), then the issue of the "independence" (e.g., paranoid, schizoid, hysterical) with re-
of masculinity and femininity, so conceived, is spect to these adjectives. The mean adjective
seen in a new light (Wiggins & Holzmuller, profiles of the seven diagnostic categories were
1978). intercorrelated, and two factors were ex-
The alphabetical list of human "needs" tracted. The circular ordering of the diag-
provided by Henry Murray (1938) has proved nostic categories (Plutchik & Platman, 1977,
to be an exceptionally fertile source of per- p. 421) bears a striking resemblance to the
sonality dimensions. Within the framework of circular ordering of corresponding MMPI
Murray's personology, several of these dimen- scales reported elsewhere (Schaefer, 1961, p.
sions have been studied in considerable depth 135). The constellations of traits implied
(e.g., Atkinson, 1958; McClelland, 1961; by diagnostic labels can be represented by
Winter, 1973), although little attention has a two-dimensional interpersonal circumplex
been given to the interrelationships among (Schaefer & Plutchik, 1966). To the extent
the needs in the overall system, perhaps be- that the MMPI clinical scales reflect conven-
cause an alphabetical taxonomy provides no tional diagnostic labeling, they too would be
guidance in this matter. Needs selected from expected to exhibit a circular ordering.
Murray's list have formed the basis for two The present taxonomy of interpersonal
multiscale inventories (Edwards, 1959; Jack- traits may prove useful to investigators who
son, 1967) and for scoring keys on an adjec- employ single adjectives as stimulus materials
tive checklist (Gough & Heilbrun, 1965). in studies of interpersonal perception, impres-
There appears to be a consistent factor struc- sion formation, and trait attribution. The
ture associated with the dozen needs these representativeness of the stimulus materials is
instruments happen to share in common, but always an issue in such studies, and the pres-
attempts to interpret this structure have not ent taxonomy provides a systematic basis for
been particularly enlightening from a substan- sampling the entire domain of interpersonal
tive point of view (Huba & Hamilton, 1976). traits as an alternative to, for example, the
In contrast, the strangely neglected work of less differentiated strategy of sampling "fa-
Stern (1958, 1970) presents convincing evi- vorable" traits. Elaborate sampling procedures
dence that a circumplex model provides a for the study of one or more specific dimen-
meaningful representation of the full range of sions of interpersonal behavior are made pos-
Murray's need variables. sible by the availability of a list of 864 terms
Two-dimensional circumplex models may that have been classified within 16 categories.
have utility outside the realm of ordinary The position of an adjective within a given
language-trait description associated with per- category is indexed by the correlation of that
sonality inventories and adjective checklists. adjective with its own, its opposite, and its
The circular ordering of intercorrelations
orthogonal clusters. In addition, the earlier
among clinical scales from the MMPI has
been noted (Schaefer, 1961), and it has itemmetric work of Norman (Note 3) pro-
been argued that these scales can be inter- vides information on these same adjectives
preted within a two-dimensional circular with respect to such characteristics as social
model of personality (Kassebaum, Couch, desirability, difficulty level, self-endorsement,
& Slater, 1959). Given the manner in which and attributions to liked, indifferent, and dis-
the MMPI clinical scales were constructed, liked others. Although such itemmetric data
this argument appears rather farfetched on are useful primarily for calibration of stimu-
initial consideration. However, a recent study lus materials, they may themselves be profit-
TAXONOMY OF INTERPERSONAL TRAITS 411

ably studied for the light they shed on basic Bern, D. J., & Allen, A. On predicting some of the
issues of social perception (Goldberg, 1978). people some of the time: The search for cross-situa-
tional consistencies in behavior. Psychological Re-
view, 1974, 81, 506-520.
Bern, S. L. The measurement of psychological an-
Reference Notes drogyny. Journal of Consulting and Clinical Psy-
chology, 1974,42, 115-162.
1. Wiggins, J. S. In defense of traits. Invited address Benjamin, L. S. Structural analysis of social be-
to the Ninth Annual Symposium on Recent De- havior. Psychological Review, 1974, 81, 392-425.
velopments in the Use of the MMPI, Los Angeles, Carson, R. C. Interaction concepts of personality.
February 28, 1974. Chicago: Aldine, 1969.
2. Goldberg, L. R. Language and personality: De- Cattell, R. B. Personality: A systematic, theoretical,
veloping a taxonomy of trait-descriptive terms. and factual study. New York: McGraw-Hill, 1950.
Invited address to the Division of Evaluation and Cattell, R. B. Personality and motivation structure
Measurement at the 86th Annual Convention of and measurement. Yonkers-on-Hudson, N. Y.:
the American Psychological Association, San Fran- World Book, 1957.
cisco, August 27, 1977. Chance, E. Families in treatment; from the view-
3. Norman, W. T. 2,800 personality trait descriptors: point of the patient, the clinician and the re-
Normative operating characteristics for a uni- searcher. New York: Basic Books, 1959.
versity population. Unpublished manuscript, Uni- Cronbach, L. J. Coefficient alpha and the internal
versity of Michigan, 1967. structure of tests. Psychometrika, 1951, 16, 297-334.
4. Wiggins, J. S. & Marston, B. J. An index for Crowne, D. P., & Marlowe, D. A new scale of social
evaluating circumplexity in two-dimensional real- desirability independent of psychopathology. Jour-
space models. Unpublished manuscript, University nal of Consulting Psychology, 1960, 24, 349-354.
of British Columbia, 1976. Edwards, A. L. Social desirability and probability
of endorsement of items in the Interpersonal Check
5. Marston, B. J. Unpublished materials, University List. Journal of Abnormal and Social Psychology,
of British Columbia, 1976. 1957, 55, 394-396. (a)
6. Merritt, N. G. Personality and value character- Edwards, A. L. The social desirability variable in
istics of volunteers and professionals at three dif- personality assessment and research. New York:
ferent social service agencies. Unpublished master's Holt, Rinehart & Winston, 1957. (b)
thesis, University of British Columbia, 1977. Edwards, A. L. Edwards Personal Preference Sched-
7. Wiggins, J. S., & Holzmuller, A. Further evidence ule. New York: Psychological Corporation, 1959.
on androgyny and interpersonal flexibility. Un- Foa, U. G. Convergences in the analysis of the
published manuscript, University of British Co- structure of interpersonal behavior. Psychological
lumbia, 1978. Review, 1961, 68, 341-353.
8. Jackson, D. N. A threshold model for stylistic re- Foa, U. G. New developments in facet design and
sponding. Paper presented at the 76th Annual analysis. Psychological Review, 1965, 72, 262-274.
Convention of the American Psychological Associa- Foa, U. G., & Foa, E. B. Societal structures of the
tion, San Francisco, September 1968. mind. Springfield, 111.: Charles C Thomas, 1974.
9. Norman, W. T. On decomposing personality rat- Foa, U. G., Triandis, H. C., & Katz, E. W. Cross-
ings. Paper presented at the annual meeting of cultural invariance in the differentiation and or-
the Society of Multivariate Experimental Psy- ganization of family roles. Journal of Personality
chology, London, Canada, November 7,1974. and Social Psychology, 1966, 4, 316-327.
Goldberg, L. R. Differential attribution of trait-
descriptive terms to oneself as compared to well-
References liked, neutral, and disliked others: A psychometric
analysis. Journal of Personality and Social Psy-
chology, 1978, 36, 1012-1028.
Allport, G. W. Personality: A psychological inter-
pretation. New York: Holt, 1937. Gough, H. G., & Heilbrun, A. B. The Adjective
Allport, G. W., & Odbert, H. S. Trait names: A Check List manual. Palo Alto, Calif.: Consulting
Psychologists Press, 1965.
psycho-lexical study. Psychological Monographs,
1936, 47(1, Whole No. 211). Guilford, J. P. Personality. New York: McGraw-Hill,
Atkinson, J. W. (Ed.). Motives in fantasy, action 1959.
and society. Princeton, N.J.: Van Nostrand, 1958. Guttman, L. A new approach to factor analysis: The
Baumrind, D., & Black, A. E. Socialization practices radex. In P. R. Lazarsfeld (Ed.), Mathematical
associated with dimensions of competence in pre- thinking in the social sciences. Glencoe, 111.: Free
school boys and girls. Child Development, 1967, Press, 1954.
38, 291-328. Heider, F. The psychology of interpersonal relations.
Becker, W. C., & Krug, R. S. A circumplex model for New York: Wiley, 1958.
social behavior in children. Child Development, Homans, G. C. Social behavior: Its elementary form.
1964, 35, 371-396. New York: Harcourt, Brace, 1961.
412 JERRY S. WIGGINS

Huba, G. J., & Hamilton, D. L. On the generality of search. Multivariate Behavioral Research Mono-
trait relationships: Some analyses based on Fiske's graphs, 1971, 6, No. 2.
paper. Psychological Bulletin, 1976, 83, 868-876. Schaefer, E. S. A circumplex model for maternal
Jackson, D. N. Personality Research Form manual. behavior. Journal of Abnormal and Social Psy-
Goshen, N.Y.: Research Psychologists Press, 1967. chology, 1959, 59, 226-235.
Jackson, D. N. A sequential system for personality Schaefer, E. S. Converging conceptual models for
scale development. In C. D. Spielberger (Ed.), maternal behavior and child behavior. In J. C.
Current topics in clinical and community psychol- Glidewell (Ed.), Parental attitudes and child be-
ogy (Vol. 2). New York: Academic Press, 1970. havior. Springfield, 111.: Charles C Thomas, 1961.
Kassebaum, G. G., Couch, A. S., & Slater, P. E. Schaefer, E. S., & Bayley, N. Maternal behavior, child
The factorial dimensions of the MMPI. Journal behavior, and their intercorrelations from infancy
of Consulting Psychology, 1959, 23, 226-236. through adolescence. Monographs of the Society
Kelly, G. A. The psychology of personal constructs for Research in Child Development, 1963, 28(3,
(Vols. 1 & 2). New York: Norton, 19SS. Whole No. 87).
LaForge, R., & Suczek, R. The interpersonal dimen- Schaefer, E. S., & Plutchik, R. Interrelationships of
sion of personality: III. An interpersonal check emotions, traits, and diagnostic constructs. Psy-
list. Journal of Personality, 195S, 24, 94-112. chological Reports, 1966, 18, 399-410.
Leary, T. Interpersonal diagnosis of personality. Stern, G. G. Preliminary manual: Activities Index-
New York: Ronald Press, 1957. College Characteristics Index. Syracuse, N.Y.:
Loevinger, J. Objective tests as instruments of psy- Syracuse University Psychological Research Cen-
chological theory. Psychological Reports, 1957, 3, ter, 1958.
635-694 (Monograph No. 9). Stern, G. G. People in context: Measuring person-
London, H., & Exner, J. E., Jr. (Eds.). Dimensions environment congruence in education and indus-
of personality. New York: Wiley, 1978. try. New York: Wiley, 1970.
Lorr, M., Klett, C. J., & McNair, D. M. Syndromes Sullivan, H. S. The interpersonal theory of psychi-
of psychosis. New York: Macmillan, 1963. atry. New York: Norton, 1953.
Lorr, M., & McNair, D. M. An interpersonal behavior Webster's new international dictionary (2nd ed.,
circle. Journal of Abnormal and Social Psychology, unabridged). Springfield, Mass.: Merriam, 1924.
1963, 67, 68-75. Webster's third new international dictionary of the
English language unabridged. Springfield, Mass.:
Lorr, M., & McNair, D. M. Expansion of the inter-
personal behavior circle. Journal of Personality Merriam, 1961.
Wiggins, J. S. Interrelationships among MMPI mea-
and Social Psychology, 1965, 2, 823-830.
sures of dissimulation under standard and social
McClelland, D. C. The achieving society. Princeton,
desirability instructions. Journal of Consulting
N.J.: Van Nostrand, 1961.
Psychology, 1959, 23, 419-427.
Mischel, W. Personality and assessment. New York:
Wiggins, J. S. Personality structure. Annual Review
Wiley, 1968. of Psychology, 1968, 19, 293-350.
Murray, H. A. Explorations in personality. New
Wiggins, J. S., & Holzmuller, A. Psychological an-
York: Oxford University Press, 1938.
drogyny and interpersonal behavior. Journal of
Osgood, C. E. Speculation on the structure of inter-
Consulting and Clinical Psychology, 1978, 46, 40-
personal intentions. Behavioral Science, 1970, 15,
52.
237-254. Wiggins, J. S., Renner, K. E., Clore, G. L., & Rose,
Plutchik, R., & Platman, S. R. Personality connota- R. J. The psychology of personality. Reading,
tions of psychiatric diagnoses. Journal of Nervous
and Mental Disease, 1977, 165, 418-422. Mass.: Addison-Wesley, 1971.
Rinn, J. L. Structure of phenomenal domains. Psy- Winter, D. G. The power motive. New York: Free
Press, 1973.
chological Review, 1965, 72, 445-466. Zadeh, L. A., Fu, K.-S., Tanaka, K., & Shimura, M.
Roback, A. A. The psychology of character. New
(Eds.). Fuzzy sets and their applications to cog-
York: Harcourt, Brace, 1928. nitive and decision processes. New York: Aca-
Roe, A., & Siegelman, M. A parent-child question-
demic Press, 1975.
naire. Child Development, 1963, 34, 355-369.
Rogers, T. B. The process of responding to person-
ality items: Some issues, a theory and some re- Received April 14, 1978 •

You might also like