Professional Documents
Culture Documents
John Et Al 2008 PDF
John Et Al 2008 PDF
Paradigm Shift
to the Integrative Big Five Trait Taxonomy
History, Measurement, and Conceptual Issues
Oliver P. John
Laura P. Naumann
Christopher J. Soto
Theorist I II III IV V
Bales (1970) DominantInitiative SocialEmotional Orientation Task Orientationa
Block & Block (1980) Undercontrol Overcontrol Ego-Resiliencyb (R)
A. H. Buss & Plomin Activity Sociability Impulsivity (R) Emotionality
(1975)
Cattell (1943) Exvia (vs. Invia) Pathemia (vs. Cortertia) Superego Strength Adjustment (R) (vs. Anxiety) Independence
Clark & Watson (1999) Positive Emotionality Constraint (vs. Disinhibition) c
Negative Emotionality
Comrey (1970) Extraversion, Activity Femininity (vs. Masculinity) Orderliness, Social Emotional Stability (R) Rebelliousness
Conformity
Eysenck (1986) Extraversion Psychoticismc (R) Neuroticism
Gough (1987)
CPI Vectors Externality Norm-Favoring Self-Realizationd (R)
CPI Scales Sociability Femininity Achievement Well-Being (R) Achievement via
viaConformance Independence
Guilford (1975) Social Activity Paranoid Disposition (R) Thinking Introversion Emotional Stability (R)
Hogan (1986) Sociability Likeability Prudence (vs. Impulsivity) Adjustment (R) Intellectance
Jackson (1984) Outgoing, Social Leadership Self-Protective Orientation Work Orientation Dependence (R) AestheticIntellectual
(R)
MMPI; Myers & Extraversion (vs. Feeling (vs. Thinking) Judging (vs. Perceiving) Intuition (vs. Sensing)
McCauley (1985) Introversion)
MBTI; Tellegen (1982) Positive Emotionality Constraint Negative Emotionality Absorption
Agentive Communal
Tellegen et al. (2003) Histrionic Paranoid (R) Compulsive Borderline Schizotypal
Wiggins (1979)
e
Power/Dominance Nurturance (Conscientiousness) (Neuroticism) (Openness)
Note. Based on John (1990) and McCrae and John (1992). (R) indicates that the dimension was reverse-scored in the direction opposite to that of the Big Five label listed above.
MBTI, MyersBriggs Type Indicator; MMPI-2, Minnesota Multiphasic Personality Inventory2.
a
This dimension contrasts a work-directed, emotionally neutral orientation with an erratic, emotionally expressive orientation (Bales, 1970), and thus seems to combine elements of
both Conscientiousness and Neuroticism.
b
Ego-resiliency seems to subsume aspects of both Openness and low Neuroticism, because an ego-resilient individual is considered both intellectually resourceful and effective in
controlling anxiety (Block & Block, 1980). However, Robins, John, and Caspi (1994) found that in adolescents, ego-resiliency is related to all of the Big Five dimensions in the well-
adjusted direction. The ego-control construct was related to Extraversion, Conscientiousness, and Agreeableness, with the undercontrolled pole most similar to Extraversion and
overcontrolled pole most similar to Conscientiousness and Agreeableness.
c
High scores on the GTS Constraint scale are correlated with both Agreeableness and Conscientious (L. A. Clark & Watson, 1999; Markon et al., 2005). Conversely, the EPQ
Psychoticism scale is associated with low scores on both Agreeableness and Conscientiousness (Goldberg & Rosolack, 1994; McCrae & Costa, 1985a).
115
d
The third vector scale on the CPI (Gough, 1987) measures levels of psychological integration and realization and should reflect aspects of both low Neuroticism (e.g., Well-Being)
and high Openness (e.g., Achievement via Independence).
e
Wiggins (1979) originally focused on Dominance and Nurturance, which define the interpersonal circumplex. The dimensions in parentheses indicate that Trapnell and Wiggins
(1990) added adjective scales for Conscientiousness, Neuroticism, and Openness after the emergence of the Big Five (see also Wiggins, 1995).
116 II. THEORETICAL PERSPECTIVES
ality psychology, a taxonomy would permit Outline and Goals of This Chapter
researchers to study specified domains of re-
lated personality characteristics, rather than The first version of this chapter (John, 1990)
examining separately the thousands of par- offered a comprehensive and detailed review
ticular attributes that make human beings of most of the available research. This is no
individual and unique. Moreover, a generally longer possible as we are writing this chap-
accepted taxonomy would facilitate the ac- ter in 2007. What has happened? Figure 4.1
cumulation and communication of empirical uses publication trends over the past 25 years
findings by offering a standard vocabulary, to illustrate how fundamentally the field has
or nomenclature. changed. Specifically, we show the number of
After decades of research, the field has publications related to the Big Five personal-
now achieved an initial consensus on a gen- ity traits for each 5-year interval, beginning
eral taxonomy of personality traits, the Big in the early 1980s, obtained from keyword
Five personality dimensions. These dimen- searches of the PsycINFO database. To pro-
sions do not represent a particular theoretical vide a comparison, we also show the publica-
perspective but were derived from analyses tion trend for the influential models developed
of the natural-language terms people use to earlier by Cattell and by Eysenck. Although
describe themselves and others. Rather than both were then close to retirement age, their
replacing all previous systems, the Big Five influence had continued during the 1980s. In
taxonomy serves an integrative function be- fact, both Cattell (1990) and Eysenck (1990)
cause it can represent the various and diverse had written chapters on personality traits for
systems of personality description in a com- the first edition of this handbook.
mon framework, as shown by the columns What did we expect to find? Our intu-
organizing Table 4.1. itions suggested that publications on the Big
1750
1500
Number of publications
1250
1000
Cattell/Eysenck
750 Big Five/FFM
500
250
0
1980 1985 1990 1995 2000 2005
1984 1989 1994 1999 2004 2009
Publication year
FIGURE 4.1. Number of publications related to either the Big Five personality traits or to the influential
models developed earlier by Cattell and by Eysenck (in 5-year intervals), identified in keyword searches
of the PsycINFO database. Note: The numbers identified in the figure as Cattell/Eysenck refer to the sum
of all articles that used one of the measures developed by either Eysenck or Cattell as a keyword, such as
EPI, EPQ, and 16PF; those identified as Big Five/FFM are the sum of all articles that used as one
of their keywords Big Five, Five Factor Model, 5 Factor Model, and + personality (to rule out
misidentifications of articles using these keywords in other literatures, such as the big five game animals
in Africa). To estimate the projected publication trends for 20052009 (which were not yet available
when this chapter was completed), we computed the sum of articles for 2005 and 2006 and multiplied
that 2-year period by 2.5.
4. Big Five Trait Taxonomy 117
Five had increased substantially since the (1936), various psychologists have turned to
mid-1980s, with Cattells and Eysencks in- the natural language as a source of attributes
fluence decreasing. But we were surprised by for a scientific taxonomy. This work, begin-
the data. First, the ascent of the Big Five hap- ning with the extraction of all personality-
pened much more gradually than we had ex- relevant terms from the dictionary, has been
pected, and Cattells and Eysencks influence guided by the lexical approach (see John et
held steady much longer. As Figure 4.1 shows, al., 1988; Saucier & Goldberg, 1996b). The
it took until the late 1990s for the number lexical hypothesis posits that most of the so-
of Big Five publications to finally overtake cially relevant and salient personality char-
the two older models. Second, whereas ref- acteristics have become encoded in the natu-
erences to Cattell and Eysenck models have ral language (e.g., Allport, 1937). Thus, the
finally begun to decline in absolute numbers, personality vocabulary contained in the dic-
their decline has been small compared to the tionaries of a natural language provides an
amazing increase in research publications on extensive, yet finite, set of attributes that the
the Big Five. By 2006, the last year for which people speaking that language have found
we had figures available, the number of Big important and useful in their daily interac-
Five publications exceeded 300 per year, tions (Goldberg, 1981).
compared with less than 50 for the two older
models.
Allport and Odberts Psycholexical Study:
In the 9 years since the previous version
Traits,States, Activities, and Evaluations
of this chapter (John & Srivastava, 1999)
was completed, almost 2,000 new publica- Following Baumgartens (1933) work in Ger-
tions on the Big Five have appeared. As a man, Allport and Odbert (1936) conducted a
result, we can now cover only a small frac- seminal lexical study of the personality terms
tion of all the relevant work in this chapter. in an unabridged English dictionary. They in-
Our main goal remains to provide a general cluded all terms that could be used to distin-
overview and introduction to the field that guish the behavior of one human being from
focuses on the main issues and can serve as that of another (Allport & Odbert, 1936,
a useful reference resource. We therefore p.24) and identified almost 18,000 terms
refer the reader to more specialized sources a semantic nightmare (Allport, 1937,
or reviews as needed. We begin our chapter pp.353354) that would keep psychologists
with the history of the Big Five, including the at work for a life time (Allport & Odbert,
discovery of the five dimensions, research 1936, p. vi). Indeed, this task has preoccu-
replicating and extending the model, its con- pied personality psychologists for more than
vergence with research in the questionnaire 60 years (for details, see John et al., 1988;
tradition, and the development of several in- John, 1990).
struments to measure the Big Five. Then we What kinds of person descriptors are in-
compare three of the most frequently used cluded in the dictionary? Allport and Odbert
instruments and discuss some new data re- identified four major categories: (1) personal-
garding their reliability and validity. Finally, ity traits (e.g., sociable, aggressive, and fear-
we address a number of conceptual issues, in- ful), defined as generalized and personal-
cluding how the Big Five taxonomy is struc- ized determining tendenciesconsistent and
tured hierarchically, how the five dimensions stable modes of an individuals adjustment
develop, whether they predict important life to his environment (p. 26); (2) temporary
outcomes, and whether they are descriptive states, moods, and activities, such as afraid,
or explanatory concepts. rejoicing, and elated; (3) highly evaluative
judgments of personal conduct and reputa-
tion, such as excellent, worthy, average, and
The Lexical Approach and Discovery irritatingalthough these terms presuppose
of the Big Five some traits within the individual, they do not
indicate the specific attributes that gave rise
One starting place for a shared taxonomy to the individuals evaluation by others or by
is the natural language of personality de- society in general; and (4) physical character-
scription. Beginning with Klages (1932), istics, capacities and talents, and other terms
Baumgarten (1933), and Allport and Odbert of doubtful relevance to personality. Norman
118 II. THEORETICAL PERSPECTIVES
(1967) elaborated these classifications into e valuation categories have recently received
seven content categories: Individuals can be more attention (Almagor, Tellegen, & Waller,
described by their enduring traits (e.g., irras- 1995; Benet-Martnez & Waller, 1997).
cible), by the internal states they typically ex- Using both semantic and empirical clus-
perience (furious), by the physical states they tering procedures as well as his own reviews
endure (trembling), by the activities they en- of the literature available at the time (for re-
gage in (screaming), by the effects they have views, see John, 1990; John et al., 1988), Cat-
on others (frightening), by the roles they play tell reduced the 4,500 trait terms to a mere
(murderer), and by social evaluations of their 35 variables, eliminating more than 99% of
conduct (unacceptable, bad). Moreover, in- the initial terms. This drastic reduction was
dividuals differ in their anatomical and mor- dictated primarily by the data-analytic limi-
phological characteristics (short) and in the tations of his time, which made factor analy-
personal and societal evaluations attached to ses of large variable sets prohibitively cost-
these appearance characteristics (cute). ly and complex. Using this small set of 35
Both Allport and Odbert (1936) and variables, Cattell conducted several oblique
Norman (1967) classified the terms culled factor analyses (i.e., allowing for correlated
from the dictionary into mutually exclusive factors) and concluded that he had identified
categories. However, their categories clearly 12 factors, which became part of his 16 Per-
overlap and have fuzzy boundaries. Chaplin, sonality Factors (16PF) questionnaire (Cat-
John, and Goldberg (1988) proposed a pro- tell, Eber, & Tatsuoka, 1970).
totype conception where each category is de- Cattell also claimed that his factors
fined in terms of its clear cases rather than its showed excellent correspondence across
boundaries. Chaplin and colleagues applied methods, such as self-reports, ratings by
this prototype conception to traits, states, others, and objective tests; however, these
and activities. Prototypical states were seen claims have not gone unquestioned (e.g.,
as temporary, brief, and externally caused. Becker, 1960; Nowakowska, 1973). More-
Prototypical traits were seen as stable, long- over, reanalyses of Cattells own correlation
lasting, internally caused, and needed to matrices by others have not confirmed the
be observed more frequently and across a number and nature of the factors he pro-
wider range of situations than states before posed (e.g., Tupes & Christal, 1961, 1992).
they were attributed to an individual. These Digman and Takemoto-Chock (1981) con-
findings replicated the earlier classifications cluded that Cattells original model, based
and confirmed that the lay conceptions of on the unfortunate clerical errors noted
traits and states are widely shared and un- here, cannot have been correct (p. 168),
derstood. although the second-order factors of the
16PF show some correspondence between
Cattells system and the Big Five dimensions
Identifying the Major Dimensions of Personality
discovered later.
Description: Cattells Early Efforts
Allport and Odberts (1936) classifications
provided some initial structure for the per- The Big Five Factors
sonality lexicon. However, to be of practical in Personality Trait Ratings
value, a taxonomy must provide a system-
Initial Discovery of the Big Five
atic framework for distinguishing, ordering,
in Cattells Variable List
and naming individual differences in peo-
ples behavior and experience (John, 1989). Cattells pioneering work and the avail-
Aiming for such a taxonomy, Cattell (1943) ability of a relatively short list of variables
used the Allport and Odbert list as a starting stimulated other researchers to examine the
point. Because the size of that list was too dimensional structure of trait ratings. Sev-
overwhelming for research purposes, Cattell eral investigators were involved in the initial
(1943, 1945a, 1945b) began with the subset discovery of the Big Five dimensions. First,
of 4,500 trait terms. Indeed, most taxonomic Fiske (1949) constructed much simplified de-
research has focused on the trait category, scriptions from 22 of Cattells variables; the
although the other categories are no less factor structures derived from self-ratings,
important; the emotional-state and social- ratings by peers, and ratings by psychologi-
4. Big Five Trait Taxonomy 119
cal staff members were highly similar and re- haustive list of personality descriptive terms,
sembled what would later become known as which he sorted into 75 semantic categories.
the Big Five. To clarify these factors, Tupes Goldberg (1990; see also 1981, 1982) used
and Christal (1961) reanalyzed correlation this list to clarify the composition of the Big
matrices from eight samples and found five Five factors and to test their generalizability
relatively strong and recurrent factors and across methodological variations and data
nothing more of any consequence (1961, sources. Goldberg (1990) constructed an
p. 14). This five-factor structure has been inventory of 1,710 trait adjectives and had
replicated by Norman (1963), Borgatta participants rate their own personality. He
(1964), and Digman and Takemoto-Chock then scored Normans semantic categories as
(1981) in lists derived from Cattells 35 vari- scales and factor-analyzed their intercorrela-
ables. Following Norman (1963), the fac- tions in the self-rating data. The first five fac-
tors were initially labeled (I) Extraversion or tors represented the expected Big Five, repli-
Surgency (talkative, assertive, energetic); (II) cated across a variety of different methods of
Agreeableness (good-natured, cooperative, factor extraction and rotation, and remained
trustful); (III) Conscientiousness (orderly, virtually invariant even when more than five
responsible, dependable); (IV) Emotional factors were rotated.
Stability (calm, not neurotic, not easily up- To ensure independence from any a
set); and (V) Culture (intellectual, polished, priori classification, Goldberg (1990) con-
independent-minded). ducted two additional studies using abbre-
These factors (see Table 4.2 for more re- viated sets of more common terms. In one
cent labels, definitions, and examples) even- study, Goldberg obtained self- and peer rat-
tually became known as the Big Fivea ings on 475 very common trait adjectives,
name Goldberg (1981) chose not to reflect which he had grouped into 131 sets of tight
their intrinsic greatness but to emphasize synonym clusters. The five-factor self- and
that each of these factors is extremely broad. peer-report structures were very similar to
Thus, the Big Five structure does not imply each other and to the structure obtained in
that personality differences can be reduced the more comprehensive list of 1,710 terms.
to only five traits. Rather, these five dimen- Most important were the null results from
sions represent personality at a very broad the search for replicable additional factors.
level of abstraction; each dimension summa- Saucier and Goldberg (1996a) selected 435
rizes a large number of distinct, more specific highly familiar trait adjectives; a factor anal-
personality characteristics. ysis of these adjectives closely replicated the
Big Five. Another thorough search for fac-
tors beyond the Big Five showed that these
Testing the Big Five in a Comprehensive Set
five were the only consistently replicable fac-
ofEnglish Trait Terms
tors (Saucier, 1997).
After a period of dormancy during the 1970s
and early 1980s, research on personality
Assessing the Big Five with Trait Descriptive
structure increased dramatically during the
Adjectives: Simple and Circumplex Approaches
mid-1980s. Factor structures resembling the
Big Five were identified in numerous sets of Goldberg (1990, 1992) distilled his exten-
variables (e.g., Botwin & Buss, 1989; Con- sive taxonomic findings into several adjec-
ley, 1985; De Raad, Mulder, Kloosterman, tive lists. A 50-item instrument using the
& Hofstee, 1988; Digman & Inouye, 1986; so-called transparent format (Goldberg,
Goldberg, 1981, 1990; John, 1990; McCrae 1992) is not used much for research but is
& Costa, 1985a, 1987; Peabody & Goldberg, excellent for instructional purposes (Pervin,
1989; Saucier & Goldberg, 1996a). Because Cervone, & John, 2005): 10 bipolar adjec-
a number of these studies were influenced by tive scales (e.g., quiettalkative) are grouped
Cattells selection of variables (Block, 1995), together under the factor name, thus making
it was important to test the generality of the the constructs being measured transparent to
Big Five in more comprehensive variable the subject. The list used more commonly in
sets. To update the Allport and Odbert list research is the set of 100 unipolar trait de-
and rectify the problems with Cattells reduc- scriptive adjectives (TDA). Goldberg (1992)
tion steps, Norman (1967) compiled an ex- conducted a series of factor-analytic studies
TABLE 4.2. The OCEANa of Personality: Definition and Explication of the Big Five Domains
120
Conceptual Implies an energetic Contrasts a prosocial and Describes socially prescribed Contrasts emotional stability Describes the breadth, depth,
definition approach toward the communal orientation impulse control that and even-temperedness with originality, and complexity
social and material world toward others with facilitates task- and goal- negative emotionality, such of an individuals mental and
and includes traits such antagonism and includes directed behavior, such as feeling anxious, nervous, experiential life.
as sociability, activity, traits such as altruism, as thinking before acting, sad, and tense.
assertiveness, and positive tender-mindedness, trust, delaying gratification,
emotionality. and modesty. following norms and rules,
and planning, organizing,
and prioritizing tasks.
Behavioral Approach strangers at a Emphasize the good qualities Arrive early or on time for Accept the good and the Take the time to learn
examples party and introduce myself; of other people when I talk appointments; Study hard bad in my life without something simply for the
Take the lead in organizing about them; Lend things to in order to get the highest complaining or bragging (R); joy of learning; Watch
a project; Keep quiet when I people I know (e.g., class grade in class; Double-check Get upset when somebody is documentaries or educational
disagree with others (R) notes, books, milk); Console a term paper for typing and angry with me; Take it easy TV; Come up with novel set-
a friend who is upset spelling errors; Let dirty and relax (R) ups for my living space; Look
dishes stack up for more for stimulating activities that
than one day (R) break up my routine
Examples High pole: Social status High pole: Better High pole: Higher academic High pole: Poorer coping High pole: Years of
of external in groups and leadership performance in work groups grade-point averages; better and reactions to illness; education completed; better
criteria positions; selection as jury Low pole: Risk for job performance; adherence experience of burnout and performance on creativity
predicted foreperson; positive emotion cardiovascular disease, to their treatment regimens; job changes tests; success in artistic jobs;
expression; number of juvenile delinquency, longer lives Low pole: Feeling committed create distinctive-looking
friends and sex partners interpersonal problems Low pole: Smoking, to work organizations; work and home environments
Low pole: Poorer substance abuse, and greater relationship Low pole: Conservative
relationships with parents; poor diet and exercise satisfaction attitudes and political party
rejection by peers habits; attention-deficit/ preferences
hyperactivity disorder
(ADHD)
Note. Conceptual definitions are based on John and Srivastava (1999). Behavioral examples are based on significant correlations between Big Five Inventory scales and self-reported
act frequencies in an undergraduate sample (N = 375; John & Naumann, 2007). (R) denotes that the act was a reverse-keyed item (i.e., correlated negatively with the Big Five domain).
For more examples of predictive validity criteria and relevant references, see the text. aThe first letter of the Big Five dimensions form the anagram OCEAN.
4. Big Five Trait Taxonomy 121
to develop and refine the TDA, selecting only the tasks most central to human survival are
adjectives that uniquely defined each factor. universal, then the most important individ-
These scales have high internal consistency, ual differences, and the terms people use to
and their factor structure is easily replicated. label these individual differences, should be
Another adjectival measure of the Big universal as well (D. M. Buss, 1996; Hogan,
Five was developed by Wiggins (1995; Trap- 1983). Conversely, if cross-cultural research
nell & Wiggins, 1990), who used trait ad- reveals a culturally specific dimension, varia-
jectives to elaborate the two major dimen- tion on that dimension may be uniquely im-
sions of interpersonal behavior: dominance portant within that cultures particular social
(or agency) and nurturance (or communion). context. Although central from the vantage
As shown in Table 4.1, the first dimension point of the lexical approach, cross-language
resembles Extraversion in the Big Five, and research is difficult and expensive to con-
the second resembles Agreeableness. Wiggins duct, and until the 1990s it was quite rare.
thus extended his circumplex scales by add- In the initial taxonomic studies, English was
ing simple adjective measures for the other the language of choice, primarily because
three Big Five factors (Trapnell & Wiggins, the researchers were American (see John et
1990). al., 1984; John, Angleitner, & Ostendorf,
The circumplex approach has also been 1988).
extended to a perennial problem in lexical re-
search on personality factors: namely, to de-
Initial Studies in Dutch and German
scribe more clearly those characteristics that
fall in the fuzzy regions between the factors. The first two non-English taxonomy proj-
Using 10 two-dimensional circumplexes, ects involved Dutch and German, both Ger-
Hofstee, De Raad, and Goldberg (1992) de- manic languages closely related to English.
vised a novel empirical approach, called the The Dutch projects, carried out by Hofstee,
Abbreviated Big Five Circumplex (AB5C), to De Raad, and their colleagues at the Uni-
represent the two-dimensional space formed versity of Groningen (De Raad et al., 1988;
by each pair of factors and define eight fac- Hofstee et al., 1997; see De Raad, Perugini,
ets that reflect various combinations of the et al., 1998, for a review), yielded conclu-
two factors. The facets differ in whether sions generally consistent with those from
they are more closely related to one or the the American English research: Only five
other factor. For example, there are two fac- factors were replicable across different selec-
ets that reflect high Agreeableness and high tions of trait adjectives and across different
Conscientiousness, but they differ in which subject samples. Those five factors were sim-
of the two factors is given more prominence. ilar to the English Big Five, although the fifth
The responsibility facet represents agreeable Dutch factor emphasized Unconventionality
Conscientiousness, whereas the cooperation and Rebelliousness rather than Intellect and
facet represents conscientious Agreeableness Imagination, as found in English.
(Hofstee et al., 1997). The German taxonomy project, begun in
Bielefeld, carried out a comprehensive psy-
cholexical study of the German personality
Cross-Language and Cross-Cultural Studies
vocabulary (Angleitner, Ostendorf, & John,
The results reviewed so far suggest that the 1990). This study was explicitly based on the
Big Five structure provides a replicable rep- prototype conception and improved on ear-
resentation of the major dimensions of trait lier studies in several respects. In particular,
description in Englishthe five dimensions 10 independent judges classified all the terms
generalize across different types of samples, obtained from the dictionary as traits, states,
raters, and methodological variations when social evaluations, etc., thus providing a con-
comprehensive sets of variables are factored. tinuous measure of the prototypicality of each
Generalizability across languages and cultures term for each category and also a check on
is another important criterion for evaluating the reliability and validity of the judgments.
personality taxonomies (John, Goldberg, & The resulting German personality lexicon is
Angleitner, 1984). The existence of cultural more convenient to use than the unwieldy
universals (Goldberg, 1981) would be con- Allport and Odbert lists because prototypi-
sistent with an evolutionary perspective: If cality values are available for each term in 13
122 II. THEORETICAL PERSPECTIVES
different content categories. Thus, it is easy scientific journals, and much too often con-
to select subsets of prototypical traits, states, siderable slippage occurs in the translation
social evaluations, and so on, from the to- process. For example, because tempera-
tal pool for further studies. Angleitner and mental was listed as a definer of Extraver-
colleagues (1990) research has served as a sion in German, one might hypothesize an
blueprint for subsequent taxonomic efforts important cultural difference here until one
in other languages. realizes that the German trait was probably
To test the structure underlying the Ger- temperamentvoll, which has nothing to do
man trait terms, Ostendorf (1990) selected with temper but means full of life and en-
the most prototypical trait adjectives from ergy, as in vivacious. Similarly, the Italian
the German taxonomy, and his factor analy- trait term frizzante (translated as sparkling)
ses of about 450 traits yielded the clearest was not found related to intellect, as one
replication of the Big Five so far. In addition might expect, but to extraversion and prob-
to the prototypical traits representing the ably means something close to bubbly.
distillation of the German trait lexicon, Os- An initial study of GermanEnglish bi-
tendorf also included German translations of linguals, which provided support for cross-
several English personality instrumentsa language generalizability (John et al., 1984),
combined emicetic design that allows re- directly addressed the issue of translation
searchers to establish empirically the similar- equivalence. The unique advantage of the bi-
ity of indigenous (emic) factors to the factors lingual design is that sample differences can
translated from other languages and cultures be controlled and translation checks can be
(etic). Using correlational analyses, Osten- made at the level of individual items because
dorf conducted an a priori, quantitative the same subject provides descriptions in
evaluation of the fit between his emic Ger- both languages (see also Benet-Martnez &
man factors and the etic Big Five in the same John, 1998). Using a careful back-translation
sample of German subjects, and he found procedure, translation equivalence between
evidence for substantial cross-language con- English and German trait adjectives was
vergence. acceptable, with a mean correlation of .52
However, this combined emicetic across a 2-week interval between adminis-
strategy is difficult to implement and, un- trations (John et al., 1984). However, a few
fortunately, has not been used consistently translations proved to be inadequate, with
in research. Thus, researchers often reach item-translation correlations approaching
conclusions about factor similarity by eye- zero. These findings, obtained for closely re-
balling the item content of the factors in lated languages, suggest that mistranslations
the indigenous language and comparing it are even more likely to occur in monolingual
to the typical factor definitions in English. investigations of personality structure and
That leaves much leeway to the investigators lead to severe underestimations of cross-
in finding (or not finding) a factor that an- language generality.
other investigator might not have found. For These difficulties are illustrated in a
example, a Hebrew factor defined primarily study by Hofstee and colleagues (1997) who
by traits such as sophisticated, sharp, knowl- used 126 words they felt could be translated
edgeable, articulate, and impressive would and matched across previous lexical stud-
lead some researchers to conclude that they ies in English, Dutch, and German to assess
had found a clear Intellect factor, whereas factor congruence coefficients among all
Almagor and colleagues (1995) interpreted it pairs of factors in the three languages. Their
as Positive Valence. findings seemed to show considerable con-
gruence; with one exception (the Openness
factors in Dutch and English), the pairwise
Problems with Translations and Underestimating
congruence coefficients all exceeded .70.
Cross-Language Congruence
Strangely, the authors interpreted these lev-
Another methodological difficulty in cross- els of cross-language congruence as disap-
language research involves translations. Re- pointing (Hofstee et al., 1997, p. 27). We
searchers working within their indigenous are more optimistic about these findings. The
language have to translate their concepts observed levels of factor congruence can be
into English to communicate their findings in taken as absolute estimates only if one as-
4. Big Five Trait Taxonomy 123
sumes that the translations are perfectly ingly, these Italian researchers had followed
equivalent and that the factor structures in the Dutch selection procedures rather than
each language are perfectly stable. When the the German procedures, which likely would
cross-language congruence coefficients were have represented more Intellect terms in the
corrected for the imperfect reliabilities (rep- initial item pool. A second (and independent)
lication) of the within-language factor struc- team of Italian taxonomers (Di Blas & Forzi,
tures, the corrected EnglishGerman con- 1998) failed to find the same factors as the
gruence coefficients ranged from .84 to .93, first (see also De Raad, Di Blas, & Perugini,
impressive values given that they have not 1998). Given that both teams started with
been corrected for the imperfect translations the same lexical material (Italian personality
(John & Srivastava, 1999). Moreover, the descriptors), this notable lack of convergence
correspondence for the fifth factor was .93, within the same language is disconcerting
suggesting that the Intellect or Openness fac- and serves to illustrate the inherent difficul-
tor was defined almost identically in English ties in standardizing taxonomic procedures
and German. This reexamination suggests and factor-analytic decisions across cultures
that translation-based comparisons across and languages. How can we expect the Big
languages are heuristically useful but should Five to generalize across languages when two
not be interpreted in terms of absolute effect studies of the same language fail to show fac-
sizes. These results also suggest that the fifth tor generalizability?
factor in Dutch is defined differently than in Thus, apparent failures to replicate the
the other two languages, and explanations Big Five structure can be hard to interpret.
for this finding need to be sought. For example, Szirmak and De Raad (1994)
examined Hungarian personality descriptors
and found strong support for the first four
Rules for Including Trait Descriptors
of the Big Five but failed to obtain a factor
in Taxonomic Studies
resembling the fifth of the Big Five. Instead,
In all likelihood, some of the differences ob- when they forced a five-factor solution, the
served among the factor structures in these Agreeableness factor split into two factors.
three languages are due to the different inclu- Should this finding be counted as failure to
sion rules adopted by the taxonomy teams. replicate the Big Five, suggesting that Hun-
The selection criterion used by the Dutch re- garians do not differ systematically on traits
searchers favored terms related to tempera- related to imagination, creativity, and intel-
ment, excluded terms related to intellect, tal- lect? Probably not: When six factors were
ents, and capacities, and included a number rotated, an Intellect/Openness factor did
of extremely negative evaluative terms, such emerge in the Hungarian data. Again, we
as perverse, sadistic, and criminal. The Ger- suspect that this finding may be due to dif-
man team explicitly included intellect and ferences in the way the initial item pool was
talent descriptors but omitted attitudes and selected. In their review of this literature,
evaluative terms, which were included as Saucier, Hampson, and Goldberg (2000)
categories separate from traits. Finally, the conclude that Given these and other differ-
American English taxonomy included attitu- ences among studies, is it any wonder that
dinal terms such as liberal, progressive, and investigators might disagree about the evi-
provincial, along with a number of intellect dential basis for a particular structural rep-
terms. Given the diverse range of traits re- resentation? (p.23).
lated to the fifth factor, it is less surprising
that the German and English factors shared
Evidence in Non-Germanic Languages
the intellect components, whereas the Dutch
factor included some imagination-related Lexical research has now been extended to
traits (e.g., inventive, original, imaginative) a growing range of non-Germanic languag-
but otherwise emphasized unconventionality es, such as Chinese (Yang & Bond, 1990),
and was thus interpreted initially as a Re- Czech (Hrebickova & Ostendorf, 1995),
belliousness factor. Greek (Saucier, Georgiades, Tsouasis, &
One Italian trait taxonomy (Caprara & Goldberg, 2005), Hebrew (Almagor et al.,
Perugini, 1994) found a similar fifth factor in- 1995), Hungarian (Szirmak & De Raad,
terpreted as Unconventionality. Not surpris- 1994), Italian (e.g., De Raad et al., 1998),
124 II. THEORETICAL PERSPECTIVES
Polish (Szarota, 1995), Russian (Shmelyov & pothesis of an omnipresent trait structure
Pokhilko, 1993), Spanish (Benet-Martnez (p. 214). Although this cautious conclusion
& Waller, 1997), Tagalog in the Philipines falls short of an unequivocal endorsement of
(e.g., Church & Katigbak, 1989; Church, the universality of the Big Five, it nonethe-
Reyes, Katigbak, & Grimm, 1997), Turk- less offers a strong disconfirmation of the
ish (Somer & Goldberg, 1999), and others. linguistic-relativism hypothesis that many
This literature has now grown far beyond of us had expected to hold true before the
the scope of this chapter, and we thus refer lexical data became available: There are no
the reader to several in-depths reviews (e.g., data to suggest that each culture and lan-
Ashton et al., 2004; De Raad & Perugini, guage has its own, unique set of personality
2002; De Raad et al., 1998; Saucier & Gold- dimensions; at least at the level of broad trait
berg, 2001; Saucier et al., 2000). dimensions, cultures are more alike than we
Most generally, our reading of this lit- may have expected.
erature is that factors similar to the Big Five
have been found in many other languages,
but often more than five factors needed to The Big Five in Personality Questionnaires
be rotated, and sometimes two indigenous
factors corresponded to one of the Big Five. While researchers in the lexical tradition
The Big Five have been well-replicated in were accumulating evidence for the Big
Germanic languages, but the evidence for Five, the need for an integrative framework
non-Western languages and cultures tends became more pressing among researchers
to be more complex. Overall, the evidence who studied personality with questionnaire
is least compelling for the fifth factor, which scales. Joint factor analyses of questionnaires
appears in various guises, ranging from pure developed by different investigators had
Intellect (in German) to Unconventionality shown that two broad dimensions, Extraver-
and Rebelliousness (in Dutch and Italian). sion and Neuroticism, appear in one form
Moreover, when we only consider which fac- or another in most personality inventories.
tors are the most replicable, then structures Beyond these Big Two (Wiggins, 1968),
with fewer factors (such as two or three) are however, the various questionnaire-based
often even more robust than the more differ- models had shown few signs of convergence.
entiated Big Five (e.g., Saucier et al., 2005). For example, Eysenck (1991) observed that
Finally, a number of studies have suggested Where we have literally hundreds of in-
more than five factors. For example, the ventories incorporating thousands of traits,
seven-factor solutions in Spanish and English largely overlapping but also containing spe-
(see Benet-Martnez & Waller, 1997) sug- cific variance, each empirical finding is strict-
gested additional separate positive and nega- ly speaking only relevant to a specific trait.
tive self-evaluation factors. The more recent ... This is not the way to build a unified sci-
six-factor solutions, obtained in reanalyses entific discipline (p.786).
of data from several different languages
(Ashton et al., 2004), suggest an additional
Costa and McCraes Research
honestyhumility factor.
While it is too early to decide whether The situation began to change in the early
these additional factors hold sufficient prom- 1980s when Costa and McCrae were devel-
ise, two conclusions are apparent now. First, oping the NEO Personality Inventory (even-
note that these factors are indeed addi- tually published in 1985), labeled N-E-O
tional; that is, they provide evidence for the because it was designed to measure the three
generalizability of the Big Five plus one or dimensions of Neuroticism, Extraversion,
two further factors. Second, when more fac- and Openness to experience. Costa and Mc-
tors than the Big Five have been identified, Crae (1976) had begun their work with clus-
the additional factors rarely replicate across ter analyses of the 16PF (Cattell et al., 1970),
multiple studies conducted by independent which, as we described above, originated in
investigators. Thus, we agree with De Raad Cattells early lexical work. Their analyses
and colleagues (1998), who concluded that again yielded the ubiquitous Extraversion
the findings show the general contours of and Neuroticism dimensions, but also con-
the Big Five model as the best working hy- vinced Costa and McCrae of the importance
4. Big Five Trait Taxonomy 125
TABLE 4.3.Defining Facets for the Big Five Trait Domains: Three Approaches
Lexical facets (18) (Saucier & NEO-PI-R facets (30) (Costa & CPI-Big Five facets (16) (Soto &
Ostendorf, 1999) McCrae, 1992) John, 2008)
Extraversion (E) facets
E Sociability E Gregariousness E Gregariousness
E Assertiveness E Assertiveness E Assertiveness/Leadership
E Activity
E Activity/Adventurousness E Excitement-Seeking [O Adventurousness]
E Social Confidence vs. Anxiety
E Positive emotions
E Unrestraint
[A Warmth/Affection] E Warmth
Note. The NEO-PI-R facets are listed in the middle column because that instrument makes the largest number of
distinctions below the Big Five (30 facets), as compared with the 18 lexical facets and the 16 CPI facets. CPI-Big Five
facets are matched with NEO-PI-R facets on the basis of both rational judgments by the authors and correlations
between the two sets of facets in a sample of 520 adults (see Soto & John, 2008). Lexical facets are matched with
NEO-PI-R facets on the basis of rational judgments by the authors. Some facets (e.g., CPI Adventurousness) are listed
once under their primary Big Five domain (e.g., Openness) and again in brackets under another Big Five domain if
their best-matching facet appears there (e.g., next to NEO Excitement-Seeking, which is an Extraversion facet on the
NEO-PI-R but also has a substantial secondary correlation with Openness). Note that the Warmth facet belongs to the
Extraversion domain in the NEO-PI-R, whereas a very similar Warmth/Affection facet belongs to the Agreeableness
domain in the lexical approach.
4. Big Five Trait Taxonomy 127
fer in their preferences for factor labels even factor analysis could be performed on more
when the factor content is quite similar. The reliable, aggregated observer judgments. The
fact that the labels differ does not necessar- varimax rotated factor loadings, shown in
ily mean that the factors are different, too. Table 4.4, provide a compelling confirma-
Thus, there may be more commonality than tion of the initial prototypes. All but one
meets the eye. item loaded on its hypothesized factor in the
A prototype approach may help identify expected direction, and most of the loadings
these commonalities across studies (John, were substantial.
1990). Natural categories such as the Big Note that the items defining each of
Five typically have fuzzy and partially over- the factors cover a broad range of content.
lapping definitions (Rosch, 1978), but they For example, Factor I includes traits such
can still be extremely useful when defined in as active, adventurous, assertive, dominant,
terms of prototypical exemplars. Similarly, energetic, enthusiastic, outgoing, sociable,
the Big Five may be defined with proto- and show-off. In light of this substantial
typical traits that occur consistently across bandwidth, the heterogeneity of the previ-
studies. One way to integrate the various ous factor labels is understood more easily
interpretations of the factors is to concep- different investigators have focused on dif-
tually map the five dimensions into a com- ferent components, or facets, of the total
mon language. Human judges were used to range of meaning subsumed by each factor.
abstract the common elements in these find- In this study, the Extraversion factor includes
ings (John, 1989, 1990), and the 300 terms at least five distinguishable components: ac-
included in the Adjective Check List (ACL; tivity level (active, energetic), dominance
Gough & Heilbrun, 1983) served as the (assertive, forceful, bossy), sociability (out-
standard language. going, sociable, talkative), expressiveness
(adventurous, outspoken, noisy, show-off),
and positive emotionality (enthusiastic,
Conceptually Derived Prototype Descriptions of
spunky). Note that these five components
the Big Five and Their Validation in Observer Data
are similar to five of the six facets Costa and
A set of 10 judges first developed a detailed McCrae (1992) included in their definition
understanding of the Big Five by reviewing of the Extraversion domain: activity, asser-
the factor solutions and interpretations of all tiveness, gregariousness, excitement-seeking,
major Big Five articles published by 1988. and positive emotions (see Table 4.3)and
The judges then independently sorted each four of the five have been identified in a lexi-
of the 300 items in the ACL into one of the cal study that empirically identified facets
Big-Five domains or into a sixth other cat- across two languages (Saucier & Ostendorf
egory, with substantial interjudge agreement. (1999), shown on the left-hand side of Table
In all, 112 of the 300 ACL terms were as- 4.3. Costa and McCraes sixth Extraversion
signed to one of the Big Five with 90% or facet, warmth, is here considered a compo-
better agreement. These terms form a rela- nent of Factor II; all 10 judges interpreted
tively narrow, or core, definition of the past research to imply that warmth is part
five factors because they include only those of Agreeableness, and the empirical loading
traits that appeared consistently across stud- of .82 confirmed this interpretation, just as
ies. As with any rationally constructed mea- the lexical facet of warmth/affection appears
sure, the validity of these categorizations was on Agreeableness. In addition to warmth (af-
tested empirically in a factor analysis of the fectionate, gentle, warm), Factor II covers
112 terms. Whereas most Big Five research themes such as tender-mindedness (sensi-
has been based on college students self- and tive, kind, soft-hearted, sympathetic), altru-
peer ratings, this study used observer data: ism (generous, helping, praising), and trust
psychologists rated 140 men and 140 wom- (trusting, forgiving), as contrasted with hos-
en who had participated in groups of 1015 tility, criticality, and distrust; again, note the
in one of the assessment weekends at the convergence with Costa and McCraes fac-
Institute of Personality and Social Research ets. More generally, the adjectival definitions
(IPSR, formerly IPAR) at Berkeley (John, of the Big Five in Table 4.4 seem to capture
1990). Because each subject had been de- the prototypical traits found in other studies
scribed on the ACL by 10 staff members, a and the facets shown in Table 4.3.
128
TABLE 4.4. Big Five Prototypes: Most Central Trait Adjectives Selected Consensually by Expert Judges and Their Factor Loadings in Personality Ratings by 10
Psychologists Serving as Observers
Extraversion Agreeableness Conscientiousness Neuroticism Openness
Low High Low High Low High Low High Low High
.83 Quiet .85 Talkative .52 Fault- .87 Sympathetic .58 Careless .80 Organized .39 Stable .73 Tense .74 .76 Wide
.80 Reserved .83 Assertive finding .85 Kind .53 Disorderly .80 Thorough .35 Calm .72 Anxious Commonplace interests
.75 Shy .82 Active .48 Cold .85 Appreciative .50 Frivolous .78 Planful .21 .72 Nervous .73 Narrow .76 Imaginative
.71 Silent .82 Energetic .45 Unfriendly .84 Affectionate .49 .78 Efficient Contented .71 Moody interests .72 Intelligent
.67 .82 Outgoing .45 .84 Soft-hearted Irresponsible .73 Responsible .71 Worrying .67 Simple .73 Original
Withdrawn .80 Outspoken Quarrelsome .82 Warm .40 Slipshot .72 Reliable .68 Touchy .55 Shallow .68 Insightful
.66 Retiring .79 Dominant .45 Hard- .81 Generous .39 .70 Dependable .64 Fearful .47 .64 Curious
.73 Forceful hearted .78 Trusting Undependable .68 Conscientious .63 High-strung Unintelligent .59
.73 Enthusiastic .38 Unkind .77 Helpful .37 Forgetful .66 Precise .63 Self-pitying Sophisticated
.68 Show-off .33 Cruel .77 Forgiving .66 Practical .60 .59 Artistic
.68 Sociable .31 Stern .74 Pleasant .65 Deliberate Temperamental .59 Clever
.64 Spunky .28 Thankless .73 Good- .46 Painstaking .59 Unstable .58 Inventive
.64 Adventurous .24 Stingy natured .26 Cautious .58 Self-punishing .56 Sharp-witted
.62 Noisy .73 Friendly .54 Despondent .55 Ingenious
.58 Bossy .72 Cooperative .51 Emotional .45 Witty
.67 Gentle .45 Resourceful
.66 Unselfish .37 Wise
.56 Praising
.51 Sensitive
Note. Based on John (1990). These items were assigned to one Big Five domain by at least 90% of the judges and thus capture the most prototypical (or central) content of each Big Five
domain. The factor loadings, shown here only for the expected factor, were obtained in a sample of 140 males and 140 females, each of whom had been described by 10 psychologists
serving as observers during an assessment weekend at the Institute of Personality and Social Research at the University of California at Berkeley (see also John, 1989).
4. Big Five Trait Taxonomy 129
phrases based on the trait adjectives known for Neuroticism, .48 for Agreeableness, and
to be prototypical markers of the Big Five .47 for Conscientiousness, averaging .55.
(John, 1989, 1990). One or two prototypi- The sizes of these convergent correlations
cal trait adjectives served as the item core to are even more impressive given that the (ab-
which elaborative, clarifying, or contextual solute) hetero-trait, hetero-method discrimi-
information was added. For example, the nant correlations averaged .09, and 19 of the
Openness adjective original became the BFI 20 correlations were below .20, with only
item Is original, comes up with new ideas one reaching .21 (indicating that individu-
and the Conscientiousness adjective perse- als who described themselves as high in neu-
vering served as the basis for the item Perse- roticism were rated by their peers as slightly
veres until the task is finished. Thus the BFI more disagreeable).
items (which are reprinted here in Appen-
dix 4.1) retain the advantages of adjectival
items (brevity and simplicity) while avoiding Measurement: Comparing Three
some of their pitfalls (ambiguous or multiple Big Five Instruments
meanings and salient desirability). Indeed,
DeYoung (2006, p.1140) hypothesized that So far, we have discussed Goldbergs (1992)
with their more contextualized trait content, TDA, Costa and McCraes (1992) NEO ques-
the BFI items should elicit higher interrater tionnaires, and the BFI. In addition, a variety
agreement than single-adjective items, and of other measures are available to assess the
found that pairwise interrater agreement was Big Five in English and other languages (see
indeed somewhat higher for the BFI. De Raad & Perugini, 2002). Many of them
Although the BFI scales include only were developed for specific research applica-
eight to ten items, they do not sacrifice ei- tions. Digman (e.g., 1989) constructed sev-
ther content coverage or good psychomet- eral different adjective sets to study teacher
ric properties. For example, the nine-item ratings of personality in children and adoles-
Agreeableness scale includes items related to cents, and Wigginss (1995) scales were de-
at least five of the six facets postulated by scribed above. Big Five scales have also been
Costa and McCrae (1992)namely, trust constructed using items from existing instru-
(trusting, forgiving), altruism (helpful and ments. For example, scales were developed
unselfish), compliance (not quarrelsome), to measure the Big Five in adolescents using
modesty (not faultfinding with others), and personality ratings on the California Child
tender-mindedness (considerate and kind). Q-sort obtained from their mothers (John et
In U.S. and Canadian samples, the alpha al., 1994). Measelle, John, Ablow, Cowan,
reliabilities of the BFI scales range from and Cowan (2005) developed scales to mea-
.75 to .90 and average above .80. Three- sure the Big Five with a puppet interview in
month testretest reliabilities range from .80 children ages 47. In behavior genetic re-
to .90, with a mean of .85 (Rammstedt & search, Loehlin, McCrae, Costa, and John
John, 2005; 2007). In a middle-age sample, (1998) used Big Five scales constructed from
Hampson and Goldberg (2006) found a the ACL (Gough & Heilbrun, 1983) and the
mean testretest stability of .74, with stabil- CPI (Gough, 1987); for the latter, we (Soto
ity correlations of .79 for Extraversion and & John, 2008) recently developed new Big
Openness and about .70 for Agreeableness, Five domain and facet scales. As shown in
Conscientiousness, and Neuroticism. Valid- Table 4.1, another broadband personality in-
ity evidence includes substantial convergent ventory that provides scores for the Big Five
and divergent relations with other Big Five is the Hogan Personality Inventory (Hogan,
instruments as well as with peer ratings 1986). The availability of so many different
(Rammstedt & John, 2005, 2007). DeYoung instruments to measure the Big Five makes
(2006) analyzed a large community data set clear that there is no single instrument that
with BFI self-reports and BFI ratings by three represents the gold standard.
peers; however, he did not report validity cor-
relations between self-reports and the aggre-
Comparing the TDA, NEO-FFI, and BFI
gated peer ratings. We therefore reanalyzed
these data and found validity correlations of In general, the NEO questionnaires represent
.67 for Extraversion, .60 for Openness, .52 the best-validated Big Five measures in the
4. Big Five Trait Taxonomy 131
TABLE 4.5. Reliability and Convergent Validity Coefficients for Three Short Big Five Measures:
BigFive Inventory, NEO Five-Factor Inventory, and Trait Descriptive Adjectives
Measures Extraversion Agreeableness Conscientiousness Neuroticism Openness Mean
Internal consistency
BFI .86 .79 .82 .87 .83 .83
NEO-FFI .82 .75 .82 .87 .76 .81
TDA .88 .84 .84 .83 .83 .84
Mean .85 .80 .83 .85 .81 .83
Standardized convergent validity coefficients from CFA (controlling for acquiescence factors)
BFITDA .99 .91 .91 .84 .97 .95
BFINEO-FFI .83 .98 .95 .93 .90 .93
TDANEO-FFI .76 .84 .87 .78 .74 .80
Mean .92 .93 .92 .86 .91 .91
Note. N = 829 (see John & Soto, 2007). BFI, Big Five Inventory (John et al., 1991); TDA, Trait Descriptive Adjec-
tives (Goldberg, 1992; 40-item mini-marker version, Saucier, 1994); NEO-FFI, NEO Five-Factor Inventory (Costa &
McCrae, 1992); CFA, confirmatory factor analysis. Means are shown in bold. All means computed with Fisher r-to-Z
transformations.
(1992) values and actions facets because pre- and TDA showed the strongest overall con-
liminary BFI items, based on prototype items vergence (mean r = .80), followed closely by
related to conventionality (i.e., relevant to BFI and NEO-FFI (.77), and finally TDA and
the NEO-PI-R values facet) and behavioral NEO-FFI (mean r = .68).
flexibility (relevant to the action facet), failed To determine the extent to which the
to cohere with the other items on the BFI validity correlations simply reflect the imper-
Openness scale (John et al., 1991). Thus, the fect reliability of the scales rather than sub-
NEO-FFI Openness scale showed less con- stantive differences among the instruments,
vergence with either the TDA or the BFI than we corrected for attenuation using alpha.
those two instruments with each other (John As shown in Table 4.5, the corrected valid-
& Srivastava, 1999). ity correlations averaged .93. However, this
As a first test of cross-instrument conver- excellent overall result masks some impor-
gence, we examined the full 15 15 MTMM tant differences. Across instruments, the first
correlation matrix formed by the five fac- three of the Big Five (Extraversion, Agree-
tors crossed with the three instruments. ableness, and Conscientiousness) showed
The cross-instrument validity correlations, mean validities of about .95, suggesting very
computed between pairs of instruments and high equivalence of the reliable variance of
shown in Table 4.5, were generally substan- the three instruments. However, Neuroti-
tial in size. Across all five factors, the mean cism (.86) and Openness (.90) were notably
of the convergent validity correlations across lower. Focusing on the pairwise comparisons
instruments was .75, as compared with the between instruments, the patterns were more
much smaller discriminant correlations that differentiated. BFI and TDA (corrected mean
averaged .19. As shown in Table 4.5, BFI r = .95) shared virtually all of their reliable
4. Big Five Trait Taxonomy 133
variance, with the highest correlation for Ex- sion that contrasts Eysencks Psychoticism
traversion; only the correlation for Neuroti- with what might be called good character.
cism (.82) fell below .90. BFI and NEO-FFI The size of these intercorrelations should
showed the same substantial mean conver- also dampen some of the current enthusiasm
gence (.95); here the highest correlation was (e.g., DeYoung, 2006; Markon, Krueger, &
for Agreeableness, and again only one cor- Watson, 2005) for higher-order factors above
relation fell below .90 (for Extraversion). In the Big Five (Digman, 1997). Yes, as has
contrast, TDA and NEO-FFI shared less in been noted repeatedly (e.g., John & Srivas-
common (mean corrected r = .83); only one tava, 1999; Paulhus & John, 1998), the Big
correlation (for Conscientiousness) exceeded Five dimensions, as assessed by self and peer
.90, and those for Neuroticism and Openness observers, are not strictly orthogonal, and
did not even reach .80, suggesting that the scale intercorrelations of .26 are statistically
conceptualization of four of the Big Five di- significant. However, the size of these inter-
mensions is not fully equivalent across these correlations represents barely 10% shared
two instruments. On average, then, the BFI variancehardly enough, it would seem, to
converged much better with both TDA and support the two substantively interpreted
NEO-FFI than did TDA and NEO-FFI with superordinate factors initially reported by
each other. However, in contrast to Hamp- Digman (1997). An alternative view showed
son and Goldbergs (2006) impression, the that at least some of that covariance may be
empirical findings show that the BFI does explained in terms of self-enhancing biases in
not simply occupy an intermediate position self-reports (Paulhus & John, 1998).
between the lexically derived TDA and the
questionnaire-based NEO-FFI. Instead, the
Estimating Convergent and Discriminant Validity
pattern of convergence correlations depends
While Controlling Acquiescence
on the Big Five domain: The BFI achieved
practical equivalence with the TDA for Ex- Finally, we tested whether individual differ-
traversion (.99) but with the NEO-FFI for ences in acquiescent response style (i.e., yea-
Agreeableness (.99), was much closer to the saying vs. nay-saying) might serve to in-
NEO-FFI than the TDA for Neuroticism fluence cross-instrument validity estimates,
(.94 vs. .82), a pattern that was reversed for such as inflating convergent validity corre-
Openness (.90 vs. .95), and finally converged lations or depressing discriminant validity
equally well with both for Conscientiousness correlations (Soto, John, Gosling, & Potter,
(.97 vs. .96). 2008). We found small but systematic acqui-
escence effects in the BFI; they appear as a
small response-style factor in addition to the
Discriminant Correlations
five substantive personality factors. We have
Overall, discriminant correlations were low, therefore developed a new, content-balanced
with absolute values averaging .19 overall approach that controls acquiescence vari-
and .16 for the TDA and .20 for both the ance at the BFI item level and eliminates the
NEO-FFI and the BFI. Moreover, none of response-style factor; this approach is de-
the discriminant correlations reached .35 scribed here in Appendix 4.2. For the pres-
on any of the instruments, and the largest ent analyses, we used a bifactor approach to
correlations were .30 for the TDA, .34 for modeling an acquiescence factor, in addition
the NEO-FFI, and .31 for the BFI. Averaged to the five substantive personality factors, for
across instruments, only four of the 10 dis- each of the three instruments.
criminant correlations even exceeded .20: the These structural equation models have a
mean correlation was .26 for Agreeableness number of important properties. First, each
and Conscientiousness, .26 for Agreeable- instrument (BFI, TDA, NEO-FFI) has its
ness and Neuroticism, .26 for Conscien- own acquiescence factor, which was defined
tiousness and Neuroticism, and .25 for Ex- by setting the raw regression path from its
traversion and Neuroticism. Thus, there was acquiescence factor to each individual item
little support for Eysencks (1992) contention equal to 1. By setting all of these loadings
that Agreeableness and Conscientiousness equally, we ensured that the acquiescence
are highly correlated primary traits that factor would represent positive covariance
should be combined into a broader dimen- shared across all items on the instrument,
134 II. THEORETICAL PERSPECTIVES
rather than letting it estimate something else where acquiescence was not controlled. In-
(e.g., social desirability). By setting all of the cluding the acquiescence factors also did not
loadings equal to 1 (rather than setting the change the mean estimated discriminant cor-
factor variance equal to 1), we allow the relations, suggesting that at the scale level,
variance of the acquiescence factor to be es- the three Big Five instruments are not par-
timated and can thus compare the amount of ticularly susceptible to acquiescence effects.
acquiescence variance across instruments. Together, the findings in this section show
Second, the 15 substantive personality that the Big Five are fairly independent di-
factors (Big Five times three instruments) were mensions that can be measured by several
allowed to correlate with each other freely, instruments with impressive convergent and
both within and across each instrument, thus discriminant validity.
allowing us to estimate latent convergent
and discriminant validity correlations across
How Well Do BFI, TDA, and NEO-FFI Scales
pairs of instruments. Also, the three acquies-
Represent the Six NEO-PI-R Facets?
cence factors were allowed to correlate with
each other across instruments, thus allowing The findings in Table 4.5 suggest that some
us to test whether individual differences in pairs of scales define the intended Big Five
acquiescent responding generalize across the domain in very similar ways (e.g., the BFI
Big Five instruments. Third, the substantive and TDA scales for Extraversion) whereas
factors were not allowed to correlate with other pairs of scales do not (e.g., the TDA
the acquiescence factorsfurther ensuring and NEO-FFI scales for Neuroticism). One
that the acquiescence factors did not contain way to explicate how the three Big Five in-
any substantive personality variance. struments define each Big Five domain is to
We tested two general predictions about use the six facets included on the NEO-PI-R
acquiescent responding. First, does the for each trait domain as a shared point of
single-adjective item format of the TDA and reference and correlate them, for each Big
its longer nine-step response scale elicit more Five domain, with the scales on the three
acquiescence variance than the more contex- instruments. These correlations can then be
tualized item formats of the BFI and NEO- graphed, yielding profiles that show how
FFI, with their shorter five-step response well each particular scale represents each of
scales? Indeed, the estimated variance of the the six facets defined by the NEO-PI-R. The
acquiescence factor was .106 for the TDA, results are shown in Figure 4.2; we begin
compared with .033 for the BFI and .013 with Extraversion in the middle of the figure
for the FFI. Constraining the acquiescence to illustrate how to read and interpret these
factor variances to be equal across instru- profile graphs. In the Extraversion panel, it is
ments significantly reduced fit; 2(2) = 375, immediately obvious that the profile curves
p < .001. Second, is acquiescent responding for the BFI Extraversion scale and the TDA
instrument-specific or a broad response dis- Extraversion scale are extremely similar.
position that generalizes across these instru- Both show correlations of about .55 with the
ments? Our findings suggested considerable four center facets (i.e., activity, gregarious-
generalizability; the correlations between the ness, warmth, and positive emotions), indi-
acquiescence factors were all significant: .62 cating that they weigh these facets all about
for BFITDA, .61 for BFINEO-FFI, and .43 equally. Both scales correlate most highly
for TDAFFI. with assertiveness (about .70) and least high-
Third, what are the effects of including ly with excitement-seeking (about .35), indi-
the acquiescence factors on the estimates of cating that assertiveness is emphasized much
the convergent and discriminant correla- more in their definition of Extraversion than
tions? The results for the convergent correla- is excitement-seeking, which is peripheral,
tions with the acquiescence factors included at best. In contrast, the NEO-FFI Extraver-
(thus controlling the effects of acquiescence) sion scale puts much more emphasis on posi-
are shown in Table 4.5. They were virtually tive emotions and warmth than do BFI and
identical to those without the acquiescence TDA, and less emphasis on assertiveness,
factors and, most important, the pattern was as shown by the crossover pattern of their
very similar to that for the corrected con- facet profiles. In other words, the NEO-FFI
vergent validity correlations in Table 4.5, defines Extraversion as a somewhat different
Neuroticism Openness
0.90 0.90 (mean r = .51)
(mean r = .65)
0.80 .73
0.80
C o r r e l a ti o n
0.60 .61
C o r r e la t io n
0.60
.55
0.50 0.50 .50
0.40
0.40 .42
0.30
0.30
0.20
0.20
Impulsive- Angry Self- Vulnerability Anxiety Depression
Values Feelings Actions Fantasy Aesthetics Ideas
ness Hostility Consciousness NEO-PI-R Facets for Openness
NEO-PI-R Facets for Neuroticism
Conscientiousness Extraversion
0.90
0.90
(mean r = .63) (mean r = .59)
0.80 0.80
.69 .65
0.70 0.70
.62
C o r r e la t io n
0.60 .56
C o r r e la t io n
0.60
.58 .54
0.50
0.50
0.40
0.40
0.30
0.30
0.20
0.20
Excitement- Activity Gregarious- Warmth Positive Assertive-
Deliberation Competence Dutifulness Achieve- Self- Order
seeking ness Emotions ness
ment Striving Discipline
NEO-PI-R Facets for Extraversion
NEO-PI-R Facets for Conscientiousness
Agreeableness Agreeableness
0.90 0.70
(mean r = .51)
0.60
0.80
0.50
0.70
0.40
C o r r e la t io n
C o r r e l a t io n
0.60 .57
0.30
.46 0.10
0.40
0.00
0.30
-0.10
0.20
-0.20
Modesty Straight- Tender- Trust Compliance Altruism
forwardness mindedness Excitement- Activity Gregarious- Warmth Positive Assertive-
seeking ness Emotions ness
NEO-PI-R Facets for Agreeableness NEO-FFI NEO-PI-R Facets for Extraversion
BFI
TDA
FIGURE 4.2. How do the three most commonly used, short Big Five instruments define each of the five
broad trait domains? Profiles of the Big Five domains, as measured by the NEO Five-Factor Inventory
(NEO-FFI), Big Five Inventory (BFI), and Trait Descriptive Adjectives (TDA) across the 30 facet traits
defined by the NEO Personality InventoryRevised (NEO-PI-R). Note: For each Big Five domain, the
figure shows the convergent correlations of the scales of the three instruments, each shown as a profile
curve, with the six NEO-PI-R facets for that domain. The average correlation between the domain scales
from each Big Five instrument and the six NEO-PI-R facets is presented next to their respective profile
curve. For the NEO-FFI (but not the other two instruments), these coefficients represent partwhole cor-
relations; everything else being equal, the NEO-FFI profile curve should thus always be higher than the
curves for the other two instruments (BFI and TDA). Therefore, the absolute elevation of the three profile
curves is of less interest than their shapethat is, their relative similarity and differences. To illustrate
that the interpersonal dimensions of Extraversion and Agreeableness show some overlap (see also Figure
4.3), the sixth panel (on the lower right) shows the discriminant validity correlations of the three Agree-
ableness domain scales with the Extraversion facets from the NEO-PI-R.
135
136 II. THEORETICAL PERSPECTIVES
mixture of lower-level personality attributes This fuzzy boundary between the Extra-
than do the BFI and TDA. version and Agreeableness domain is also il-
We should comment on other features of lustrated in Figure 4.3 which shows the Big
these facet profiles. First, consider the overall Five trait version of the interpersonal circum-
elevation of these three Extraversion profiles, plex (Wiggins, 1979). This figure shows fac-
which is summarized by the mean correla- tor loadings for the Extraversion and Agree-
tion presented next to each profile line: .65 ableness scales from the NEO-FFI, BFI, and
(highest overall) for the NEO-FFI, followed TDA, which were factored along with the
by .56 for the BFI and .54 for the TDA. The six NEO-PI-R facets defining Extraversion,
NEO-FFIs highest elevation here (and every- the six defining Agreeableness, and the angry
where else in this figure) is due, in part, to hostility facet from the Neuroticism domain,
partial item overlap and cannot be interpreted because it is also highly negatively related to
unambiguouslyafter all, the 60 items that Agreeableness (as indicated by its substantial
make up the 12-item NEO-FFI scales are also negative loading on that factor). As we noted
included in the 30 facet scales that we used to earlier, warmth and positive emotions are the
generate the correlations for Figure 4.2. Thus, Extraversion facets with the largest (positive)
whereas the absolute elevation is informative loadings on Agreeableness. The Agreeable-
for the BFI and TDA, it is less so for the NEO- ness facets such as trust and altruism have
FFI, and the shape of the profile is much more positive loadings on Extraversion, whereas
critical, like the crossover for assertiveness. modesty and compliance have negative load-
Second, these graphs also tell us something ings. Finally, the locations of the TDA, BFI,
about the facets that define the profiles. Note and NEO-FFI scales are also compatible
that all but one facet had correlations with all with our earlier observations. BFI and TDA
three Extraversion scales above the .50 line, Extraversion are almost in exactly the same
suggesting that they are all substantially rel- spot in this two-dimensional space, and cer-
evant to the way Extraversion is conceptual- tainly closer to assertiveness than is NEO-
ized on all three instruments. The finding that FFI Extraversion, which is rotated to the
excitement-seeking had, by far, the lowest right toward Agreeableness and thus closer
correlations with all three Extraversion scales to positive emotions and warmth. Both TDA
suggests that it is a relatively peripheral facet and BFI Agreeableness are located very close
within the Extraversion domain. Apparently, to the altruism facet, which turned out to be
the NEO-PI-R facets are not all equal in their so central to all three Agreeableness scales in
centrality to their Big Five domains. Figure 4.2.
This point is even more apparent in Although two-dimensional plots of fac-
the Agreeableness panel on the lower left. tor loadings, such as those in Figure 4.3,
The BFI, TDA, and NEO-FFI Agreeableness have long been used in trait taxonomic and
scales all correlate about .70 with the altruism especially circumplex work, they seem to
facet, showing remarkable Agreement about offer less specific information than the fac-
the centrality of this facet to this domain. In et profiles in Figure 4.2. For Openness, for
contrast, all three correlated less than .30 example, the biggest difference involves the
with modesty, suggesting that this facet is value facet, which seems barely represented
peripheral to the Agreeableness domain. In on either TDA or BFI. In contrast, for Neu-
terms of profile similarity, BFI and NEO-FFI roticism it is the underweighting of depres-
are much closer to each other than either is sion, anxiety, and vulnerability and the rela-
to the TDA, especially for trust and compli- tive overweighting of hostility that make the
ance, which seem relatively underrepresented TDA scale so different from both the BFI and
on the TDA. Finally, consider the sixth panel NEO-FFI Neuroticism scales.
(on the lower right), which is the one panel
that shows discriminant validity correlations,
Big Five Measurement:
relating the three Agreeableness scales with
Conclusions and Limitations
the facets from the NEO-PI-R Extraversion
domain. Here the Agreeableness scales show One of the limitations of the findings pre-
impressive profile similarity, indicating that sented here is that we did not examine ex-
all three correlate above .50 with warmth ternal (or predictive) validity. However, both
and about .40 with positive emotions. the NEO questionnaires and the BFI have
4. Big Five Trait Taxonomy 137
%XTRAVERSION
4$! % &&) %
"&) %
!SSERTIVENESS 0OSITIVE %MOTIONS
!CTIVITY
'REGARIOUSNESS
%XCITEMENT
7ARMTH
SEEKING
4RUST
!LTRUISM
!GREEABLENESS
4$! !
"&) !
4ENDER
MINDEDNESS &&) !
!NGRY (OSTILITY .
#OMPLIANCE
3TRAIGHT
FORWARDNESS
-ODESTY
n
n
n n
FIGURE 4.3. The interpersonal circumplex formed by the Big Five domains of Extraversion and Agree-
ableness. The Extraversion and Agreeableness domain scales from three Big Five instruments (NEO-FFI,
BFI, and TDA) were factored along with the six NEO-PI-R facets defining Extraversion, the six defining
Agreeableness, and the angry hostility facet from the Neuroticism domain (because it is also highly re-
lated to Agreeableness). FFI, NEO Five-Factor Inventory; BFI, Big Five Inventory; TDA, Trait Descriptive
Adjectives; E, Extraversion; A, Agreeableness; N, Neuroticism.
been shown to predict peer ratings; and ini- not at a premium, participants are well edu-
tial evidence is now available for the TDA cated and test-savvy, and the research ques-
scales (DeYoung, 2006). Future research tion calls for the assessment of multiple facets
needs to study the comparative validity of all for the Big Five, then the full 240-item NEO-
three instruments using peer ratings and oth- PI-R would be most useful. Otherwise, the
er external criteria. One of the advantages of 44-item BFI would seem to offer a measure
the BFI is its efficiency, taking only about 5 of the core attributes of the Big Five that is at
minutes of administration time, compared least as efficient and easily understood as the
with about 15 minutes for the NEO-FFI and 60-item NEO-FFI and the 100-item TDA.
the 100-item TDA. Moreover, the BFI items At this point, we cannot recommend the use
are shorter and easier to understand than the of even shorter instruments, with as few as
NEO-FFI items (Benet-Martnez & John, 10 items (e.g., Gosling, Rentfrow, & Swann,
1998; Soto et al., 2008). The 100 (or 40) ad- 2003; Rammstedt & John, 2007), unless a
jectives on the TDA are even shorter; how- researcher encounters truly exceptional cir-
ever, single-trait adjectives can be ambiguous cumstances, such as the need to measure
in their meanings. the Big Five as part of a national phone sur-
When should researchers use each of vey. From our perspective, the gains in time
these instruments? When participant time is achieved by moving from a measure such as
138 II. THEORETICAL PERSPECTIVES
the BFI (i.e., 5 minutes of subject time) to an More could be said about the many
even shorter measure can rarely compensate shortcomings of the traditional labels, but
for the potential losses in reliability and va- better labels are hard to come by. The un-
lidity one has to risk with such minimalist surpassed advantage of the traditional labels
measurement. is that they are commonly known and used,
thus preventing Babel from taking over the
literature on the Big Five. Moreover, before
Factor Names, Numbers, or Initials: any new names are devised, the definition of
WhichShall We Use? the factors in terms of facets or components
needs to be elaborated and sharpened. At this
Problems with the English Factor Labels
point, it seems premature to settle the scope
Now that we have considered both the his- and theoretical interpretation of the factors
tory of the Big Five and their measurement, it by devising new names.
is time to revisit the names or labels assigned
to the factors (see Table 4.2). Although the
Preliminary Definitions
constructs that will eventually replace the
current Big Five may be different from what Because the traditional labels are so eas-
we know now, labels are important because ily misunderstood, we provide short defini-
they imply particular interpretations and tions of the five dimensions in Table 4.2 (cf.,
thus influence the directions that theorizing Costa & McCrae, 1992; John, 1990; Telle-
might take. Normans (1963) factor labels gen, 1985). Briefly, Extraversion implies an
have been frequently used, but Norman of- energetic approach toward the social and
fered little theoretical rationale for their se- material world and includes traits such as
lection. Normans labels differ vastly in their sociability, activity, assertiveness, and posi-
breadth or inclusiveness (Hampson, Gold- tive emotionality. Agreeableness contrasts a
berg, & John, 1987); in particular, Conscien- prosocial and communal orientation toward
tiousness and Culture are much too narrow others with antagonism and includes traits
to capture the enormous breadth of these such as altruism, tender-mindedness, trust,
two dimensions. Moreover, researchers have and modesty. Conscientiousness describes
abandoned Culture as a label for Factor V, socially prescribed impulse control that fa-
in favor of Intellect or Imagination (Saucier cilitates task- and goal-directed behavior,
& Goldberg, 1996a) or Openness to Expe- such as thinking before acting, delaying
rience (McCrae & Costa, 1985b). Neither gratification, following norms and rules, and
label is truly satisfactory, however, because planning, organizing, and prioritizing tasks.
Intellect is too narrow and Openness, while Neuroticism contrasts emotional stability
broad enough, is somewhat vague. and even-temperedness with negative emo-
Agreeableness is another problematic tionality, such as feeling anxious, nervous,
label. For one, it refers to the behavioral ten- sad, and tense. Finally, Openness to Expe-
dency to agree with others, thus incorrectly rience (vs. closed-mindedness) describes the
implying submissiveness, which is more breadth, depth, originality, and complexity
closely related to the introverted pole of Fac- of an individuals mental and experiential
tor I. Agreeableness is also too detached, too life.
neutral a label for a factor supposed to cap- The numbering convention from I to V,
ture intensely affective characteristics, such favored by Saucier and Goldberg (1996b)
as love, compassion, and sympathy. Freud and Hofstee and colleagues (1997), is useful
viewed love and work as central; following because it reflects the relative size of the fac-
this lead, we could call Factor II simply Love tors in lexical studies. Factors I and II, which
(Peabody & Goldberg, 1989). primarily summarize traits of an interper-
However, Work is too narrow a label sonal nature, tend to account for the larg-
for Factor III. Even Conscientiousness is too est percentage of variance in personality rat-
narrow because it omits a central component ings, followed by Factor III, whereas the last
that Peabody and Goldberg (1989) called two factors are the smallest, by far, in lexi-
favorable impulse control. Thus, Respon- cal studies (De Raad et al., 1998). However,
sibility or even Degree of Socialization (see the Roman numerals are hard to remember,
Gough, 1987) might be labels more appro- and the order of the factors is not invariant
priate for Factor III than Conscientiousness. across studies. Thus, we favor the mnemonic
4. Big Five Trait Taxonomy 139
though these concepts are cognitive, or, more 1993). Thus, like most structural models it
appropriately, mental in nature, they are provides an account of personality that is
clearly different from IQ. Second, limited- primarily descriptive rather than explanato-
domain scales measuring concepts such as ry, emphasizes regularities in behavior rather
Absorption, Fantasy Proneness, Need for than inferred dynamic and developmental
Cognition, Private Self-Consciousness, Inde- processes, and focuses on variables rather
pendence, and Autonomy would be difficult than on individuals or types of individuals
to subsume under Extraversion, Neuroti- (cf. John & Robins, 1993, 1998). Nonethe-
cism, or Conscientiousness. Indeed, the fifth less, the Big Five taxonomy of trait terms
factor is necessary because individual differ- provides a conceptual foundation that helps
ences in intellectual and creative functioning us examine these theoretical issues. In this
underlie artistic interests and performances, section, we begin with the hierarchical struc-
inventions and innovation, and even humor. ture defined by the Big Five and then review
Individual differences in these domains of how the Big Five predict important life out-
human behavior and experience cannot be comes, how they develop, how they combine
neglected by personality psychologists. into personality types, and how different re-
Finally, the matches between the Big searchers view their conceptual status.
Five and other constructs noted in Table 4.1
should be considered with a healthy dose of
Hierarchy, Levels of Abstraction, and the Big Five
skepticism. Some of these correspondences
are indeed based on solid research findings, A frequent objection to the Big Five is that
but others are conceptually derived and await five dimensions cannot possibly capture all
empirical confirmation. These matches reflect of the variation in human personality (e.g.,
broad similarities, ignoring some important, Block, 1995; McAdams, 1992; Mershon &
implicative, and useful differences among the Gorsuch, 1988), and that they are much too
concepts proposed by different investigators. broad. However, the objection that five di-
Nonetheless, at this stage in the field, we are mensions are too few overlooks the fact that
more impressed by the newly apparent simi- personality can be conceptualized at differ-
larities than by the continuing differences ent levels of abstraction or breadth. Indeed,
among the various models. Indeed, the Big many trait domains are hierarchically struc-
Five are useful primarily because of their in- tured (Hampson, John, & Goldberg, 1986).
tegrative and heuristic value, a value that be- The advantage of categories as broad
comes apparent in Table 4.1. The availability as the Big Five is their enormous bandwidth.
of a taxonomy, even one that is as broad and Their disadvantage, of course, is their low fi-
incomplete as the Big Five, permits the com- delity. In any hierarchical representation, one
parison and potential integration of dimen- always loses information as one moves up the
sions that, by their names alone, would seem hierarchical levels. For example, categorizing
entirely disparate. something as a guppy is more informative
than categorizing it as a fish, which in turn
is more informative than categorizing it as
Critical Issues a vertebrate. Or, in psychometric terms, one
and Theoretical Perspectives necessarily loses item information as one ag-
gregates items into scales, and one loses scale
Like any scientific model, the Big Five tax- information as one aggregates scales into fac-
onomy has limitations. Critics have argued tors (John, Hampson, & Goldberg, 1991).
that the Big Five does not provide a complete The Big Five dimensions represent a
theory of personality (e.g., Block, 1995; rather broad level in the hierarchy of per-
Eysenck, 1997; McAdams, 1992; Pervin, sonality descriptors. In that sense, they are
1994), and we agree. In contrast to McCrae to personality what the categories plant
and Costas (1996; see also McCrae & Cos- and animal are to the world of biological
ta, Chapter 5, this volume) five-factor theory, objectsextremely useful for some initial
the Big Five taxonomy was never intended as rough distinctions but of less value for pre-
a comprehensive personality theory; it was dicting specific behaviors of a particular in-
developed to account for the structural re- dividual. The hierarchical level a researcher
lations among personality traits (Goldberg, selects depends on the descriptive and pre-
4. Big Five Trait Taxonomy 141
dictive tasks to be addressed (Hampson et Eysenck (1991) challenged the field, arguing
al., 1986). In principle, the number of spe- that Little is known about the social rel-
cific distinctions one can make in the descrip- evance and importance of openness, agree-
tion of an individual is infinite, limited only ableness, and conscientiousness. ... What is
by ones objectives. lacking is a series of large-scale studies which
Norman, Goldberg, McCrae and Costa, would flesh out such possibilities (p.785).
and Hogan all recognized that there was Over the past two decades, however,
a need in personality, just as in biology, researchers have taken up the task of iden-
to have a system in which different levels tifying the particular Big Five dimensions
of generality or inclusion are recognized that predict particular life outcomes in such
(Simpson, 1961, p.12). A complete trait tax- fundamental domains as physical and men-
onomy must include middle-level categories, tal health, work, and relationships (see Table
such as assertiveness, orderliness, and cre- 4.2 for examples). This research is based on
ativity, and even narrower descriptors, such the assumption that personal factors (such
as talkative, punctual, and musical (John as the individuals traits) and environmental
et al., 1991). At this point, Costa and Mc- factors (such as aspects of a job or a rela-
Craes (1992) 30 facets, shown in Table 4.4, tionship partner) interact to jointly produce
represent the most widely used and empiri- behavioral and experiential outcomes that
cally validated model. Soto and John (2008) accumulate over the individuals lifespan
developed 16 facets (see Table 4.4) from the (e.g., Caspi & Bem, 1990; Scarr & McCart-
CPI item pool that tend to be broader than ney, 1984). In other words, personality traits
Costa and McCraes facets. Hofstee and col- are important because they influence the way
leagues (1992) circumplex-based AB5C ap- individuals interact with particular environ-
proach defines 45 facets as unique, pairwise ments. As we review below, traits influence
combinations or blends of the Big Five how individuals construe and interpret the
factors (e.g., Poise as a blend of high Extra- personal meaning a particular environment
version and low Neuroticism); measures of or situation has for them (e.g., how they in-
these facets have now developed as part of terpret a potential health risk), and to which
Goldbergs collaborative and Web-based In- aspects of the environment they attend (e.g.,
ternational Personality Item Pool (IPIP) proj- a doctors prescribed treatment regimen). In
ect. Saucier and Ostendorf (1999) provide addition to these cognitive processes of per-
a thoughtful discussion of the fundamental ceiving and attending to the environment,
issues in developing empirically based fac- traits also influence the way individuals select
ets and present 18 facets (see Table 4.4) that both social and nonsocial environments (e.g.,
show initial cross-language generalizability. college classes, jobs, places to live, relation-
Although Table 4.4 shows some promising ship partners, even music) and how they then
convergences, the three approaches differ modify those environments (e.g., their bed-
substantially in the number and nature of the rooms). It is through their systematic inter-
facets they propose, indicating that further action with environmental affordances and
conceptual and empirical work is needed to risks that traits are hypothesized to influence
achieve a consensual specification of the Big the behavioral, emotional, social, and mate-
Five factors at lower level of abstraction. rial life outcomes of the individual.
& Friedman, Chapter 31, this volume). For and organizational problems that can lead
example, low Conscientiousness predicts the to broader adjustment problems in school
likelihood of engaging in risky behaviors settings and even relationships. Ultimately,
such as smoking, substance abuse, and poor findings linking Big Five profiles to develop-
diet and exercise habits (Bogg & Roberts, mental or adjustment problems may help us
2004; Hampson, Andrews, Barckley, Lich- identify children at risk and design appropri-
tenstein, & Lee, 2000; Trull & Sher, 1994). ate interventions, such as teaching children
Moreover, highly conscientious individuals, relevant behaviors and skills (e.g., strategies
when diagnosed with an illness, are more for delaying gratification).
likely to adhere to their treatment regimens
(Kenford et al., 2002) and have been shown
Links to Academic and Work Outcomes
to live longer lives (Danner, Snowden, &
Friesen, 2001; Friedman, Hawley, & Tuck- Industrial and organizational researchers
er, 1994; Weiss & Costa, 2005). Other Big have also rediscovered the importance of
Five dimensions are also related to health- personality traits, and a growing body of re-
related risk factors. Low Agreeableness (es- search has linked the Big Five to academic
pecially hostility) predicts cardiovascular and work achievement. Early studies showed
disease (Miller, Smith, Turner, Guijarro, & that Conscientiousness as well as Openness
Hallet, 1996). High Neuroticism predicts predict school performance as measured with
less successful coping and poorer reactions objective tests in early adolescence (John et
to illness, in part because highly neurotic in- al., 1994; Robins et al., 1994). In college,
dividuals are more likely to ruminate about Conscientiousness predicts higher academic
their situation (David & Suls, 1999; Scheier grade-point averages (Noftle & Robins,
& Carver, 1993). Individuals high in Extra- 2007; Paunonen, 2003), whereas Openness
version, on the other hand, have available predicts the total years of education complet-
more social support and close relationships ed by middle adulthood (Goldberg, Sweeney,
important for coping with illness (Berkman, Merenda, & Hughes, 1998).
Glass, Brissette, & Seeman, 2000). Beyond primary and secondary school-
ing, Conscientiousness has emerged also as a
general predictor of job performance across
Links to Psychopathology, Personality Disorders,
a wide range of jobs (for reviews, see Barrick
and Adjustment Problems
& Mount, 1991; Mount, Barrick, & Stewart,
The availability of the Big Five taxonomy has 1998). The other Big Five dimensions relate
also renewed interest in the links between to more specific aspects of job performance,
personality and psychopathology, especially such as better performance or satisfaction in
personality disorders (e.g., Costa & Widi- specific job types or positions. For example,
ger, 2002; Wiggins & Pincus, 1989); findings Agreeableness and Neuroticism predict per-
from this burgeoning literature have been formance in jobs where employees work in
reviewed by Krueger and Tackett (2006) groups, Extraversion predicts success in sales
and Widiger and Smith (Chapter 30, this and management positions, Openness pre-
volume). From a developmental perspective, dicts success in artistic jobs, and Conscien-
the Big Five dimensions may serve as risks or tiousness predicts success in conventional jobs
buffers for subsequent adjustment problems. (Barrick, Mount, & Gupta, 2003; Larson,
In adolescents, for example, both low Agree- Rottinghaus, & Borgen, 2002). Neuroticism
ableness and low Conscientiousness predict is an important predictor of job satisfaction.
delinquency and externalizing problems, Highly neurotic individuals are more likely
whereas high Neuroticism and low Consci- to experience burnout and to change jobs,
entiousness predict internalizing problems whereas more emotionally stable individuals
such as depression and anxiety (John et al., feel satisfied and committed to their organi-
1994; Measelle et al., 2005; Robins et al., zations (Thoresen, Kaplan, Barsky, Warren,
1994; 1996). Low Conscientiousness is also & de Chermont, 2003). These trait-by-job
the personality trait most strongly linked to interactions help researchers develop a more
ADHD, at least when diagnosed in adult- fine-grained understanding of how different
hood (Nigg et al., 2002); more specifically, traits are instrumental to performance and
low Conscientiousness predicts attentional satisfaction in various job environments.
4. Big Five Trait Taxonomy 143
gence and peak expression of traits, their (Digman, 1989; Graziano & Ward, 1992).
stability or change throughout the lifespan, Extending Digmans (1989) earlier work on
and the effects of traits on other aspects of Hawaiian children, Digman and Shmelyov
personal development. Some critics have (1996) examined both temperament dimen-
suggested that Big Five researchers have not sions and personality dimensions in a sample
paid enough attention to issues of personal- of Russian children. Based on analyses of
ity development in childhood and adoles- teachers ratings, they concluded that the
cence (Pervin, 1994). This criticism has some Big Five offer a useful model for describing
merit: Although the Big Five taxonomy has the structure of temperament. Studies using
influenced research on adult development free-response techniques found that the Big
and aging (e.g., Helson, Kwan, John, & Five can account for a substantial portion
Jones, 2002; McCrae & Costa, 2003; Rob- of childrens descriptions of their own and
erts, Walton, & Viechtbauer, 2006; see also others personalities (Donahue, 1994), as
Roberts, Wood, & Caspi, Chapter 14, this well as teachers and parents descriptions of
volume), there has been much less research childrens personality (Kohnstamm, Halver-
on personality structure in childhood. De- son, Mervielde, & Havill, 1998). In a large
velopmental and temperament psychologists Internet sample providing self-reports on the
have studied a number of important traits BFI, we analyzed data from children and ad-
(e.g., sociability, fearful distress, shyness, olescents, ages 1020, and found substantial
impulsivity), but many studies examine one changes in the coherence and differentiation
trait at a time, in isolation from the others, of the Big Five domains, even though a vari-
and the available research has not been inte- ant of the adult Big Five factor structure was
grated in a coherent taxonomic framework. apparent as early as age 10 when individual
Until this work is done, however, research on differences in response acquiescence were
personality development across the lifespan controlled (see Appendix 4.2; Soto et al.,
is likely to remain fragmented (Halverson, 2008). In an even younger sample, Measelle
Kohnstamm, & Martin, 1994). and colleagues (2005) found that using an
The adult personality taxonomy defined age-appropriate puppet interview task, chil-
by the Big Five can offer some promising dren as young as ages 57 were able to self-
leads. In our view, the Big Five should be report on their personality; by age 6, their
examined in developmental research for two self-reports were beginning to show evidence
reasons (John et al., 1994). Theoretically, it of coherence, longitudinal stability, and ex-
may be necessary to examine the develop- ternal validity (using ratings by adults) for
mental origins of the Big Five: Given that the Big Five domains of Extraversion, Agree-
the Big Five emerge as basic dimensions of ableness, and Conscientiousness.
personality in adulthood, researchers need Two large-scale studies suggest that the
to explain how they develop. Practically, the picture may be more complicated. One study
Big Five have proven useful as a framework tested whether the adult Big Five structure
for organizing findings on adult personality would replicate in a large and ethnically di-
in areas as diverse as behavior genetics and verse sample of adolescent boys (John et al.,
industrial psychology. Thus, extension of 1994). The California Child Q-sort provided
the Big Five into childhood and adolescence a comprehensive item pool for the descrip-
would facilitate comparisons across develop- tion of children and adolescents that was not
mental periods. derived from the adult Big Five and does not
Work on these issues has now begun, represent any particular theoretical orienta-
and researchers are drawing on existing tion. Factor analyses identified five dimen-
models of infant and child temperament to sions that corresponded closely with a priori
make connections to the Big Five dimensions scales representing the adult Big Five. How-
in adulthood (for reviews, see Caspi, Rob- ever, two additional dimensions emerged. Ir-
erts, & Shiner, 2005; De Clerq, De Fruyt, ritability involved negative affect expressed
Van Leeuwen, & Mervielde, 2006; Halv- in age-inappropriate behaviors, such as whin-
erson et al., 1994; Shiner & Caspi, 2003). ing, crying, tantrums, and being overly sensi-
Some research suggests that the Big Five may tive to teasing. Activity was defined by items
provide a good approximation of personal- involving physical activity, energy, and high
ity structure in childhood and adolescence tempo, such as running, playing, and moving
4. Big Five Trait Taxonomy 145
and reacting quickly. In several Dutch sam- ier and Goldberg (1996b) argued that their
ples of boys and girls ages 316 years, van studies of personality description do not ad-
Lieshout and Haselager (1994) also found dress issues of causality or the mechanisms
the Big Five plus two additional factorsone underlying behavior. Their interest is primar-
similar to Activity factor observed earlier in ily in the language of personality. This level
the United States (John et al., 1994), and a of self-restraint may seem dissatisfactory to
Dependency dimension defined primarily by psychologists who are more interested in per-
eagerness to please and reliance on others. sonality itself. However, the findings from the
Although these findings need to be replicated lexical approach are informative because the
and extended, they leave open the possibility lexical hypothesis is essentially a functional-
that the structure of personality traits may ist argument about the trait concepts in the
be more differentiated in childhood than in natural language. These concepts are of in-
adulthood. Specifically, the additional di- terest because language encodes the charac-
mensions may originate in temperamental teristics that are central, for cultural, social,
features of childhood personality (e.g., activ- or biological reasons, to human life and ex-
ity level) that become integrated into adult perience. Thus, Saucier and Goldberg argue
personality structure over the course of ado- that lexical studies define an agenda for per-
lescence (John et al., 1994). sonality psychologists because they highlight
These studies illustrate how the Big Five the important and meaningful psychological
can help stimulate research that connects phenomena (i.e., phenotypic characteristics)
and integrates findings across long-separate that personality psychologists should study
research traditions. They also provide initial and explain. Thus, issues such as the accu-
insights about the way in which personality racy of self and peer descriptions and the
structure may develop toward its adult form. causal origin of traits (i.e., genotypes) are left
Yet a great deal of work still lies ahead. Stud- as open questions that need to be answered
ies need to examine the antecedents of the empirically. However, important characteris-
Big Five and their relations to other aspects tics may exist that people may not be able
of personality functioning in childhood and to observe and describe verbally; if so, the
adolescence. In this way, the Big Five can agenda specified by the lexical approach may
help connect research on adult personality be incomplete and would need to be supple-
with the vast field of social development. mented by more theoretically driven ap-
proaches (Block, 1995; Tellegen, 1993).
Several theories conceptualize the Big
Theoretical Perspectives on the Big Five:
Five as relational constructs. Interpersonal
Description and Explanation
theory (Wiggins & Trapnell, 1996), empha-
Over the years, researchers have articulated a sizes the individual in relationships. The Big
number of different perspectives on the con- Five are taken to describe the relatively
ceptual status of the Big Five dimensions. Be- enduring pattern of recurrent interpersonal
cause of their lexical origin, the factors were situations that characterize a human life
initially interpreted as dimensions of trait (Sullivan, 1953, pp.110111), thus concep-
description or attribution (John et al., 1988). tualizing the Big Five as descriptive concepts.
Subsequent research, however, has shown Wiggins and Trapnell emphasize the inter-
that the lexical factors converge with dimen- personal motives of agency and communion
sions derived in other personality research and interpret all of the Big Five dimensions
traditions, that they have external or predic- in terms of their interpersonal implications.
tive validity (as reviewed above), and that all Because Extraversion and Agreeableness are
five of them show about equal amounts of the most clearly interpersonal dimensions in
heritability (Loehlin et al., 1998). We thus the Big Five, they receive conceptual priority
need to ask how these differences should be in this model.
conceptualized (e.g., Wiggins, 1996); below Socioanalytic theory (Hogan, 1996)
we briefly summarize some of the major the- focuses on the social functions of self- and
oretical perspectives. other perceptions. According to Hogan, trait
Researchers in the lexical tradition tend concepts serve as the linguistic tools of ob-
to take an agnostic stance regarding the con- servers (1996, p.172) used to encode and
ceptual status of traits. For example, Sauc- communicate reputations. This view implies
146 II. THEORETICAL PERSPECTIVES
that traits are socially constructed to serve and characteristic adaptations. Personal-
interpersonal functions. Because trait terms ity traits are basic tendencies that refer to the
fundamentally reflect reputation, individu- abstract underlying potentials of the individ-
als who self-report their traits engage in a ual, whereas attitudes, roles, relationships,
symbolicinteractionist process of intro- and goals are characteristic adaptations that
spection (i.e., the individual considers how reflect the interactions between basic tenden-
others view him or her). Hogan emphasizes cies and environmental demands accumu-
that individuals may distort their self-reports lated over time. According to McCrae and
with self-presentational strategies; another Costa, basic tendencies remain stable across
source of distortion are self-deceptive biases the life course, whereas characteristic adap-
(cf. Paulhus & John, 1998), which do not re- tations can undergo considerable change.
flect deliberate impression management but From this perspective, then, a statement such
honestly held, though biased, beliefs about as Paul likes to go to parties because he is
the self. extraverted is not circular, as it would be if
The evolutionary perspective on the extraverted were merely a description of typ-
Big Five holds that humans have evolved ical behavior (Wiggins, 1997). Instead, the
difference-detecting mechanisms to per- concept extraverted stands in for biological
ceive individual differences that are relevant structures and processes that remain to be
to survival and reproduction (D. M. Buss, discovered. This view is similar to Allports
1996, p. 185; see also Botwin, Buss, & (1937) account of traits as neuropsychic
Shackelford, 1997). Buss views personality as structures and Eysencks view of traits as
an adaptive landscape where the Big Five biological mechanisms (Eysenck & Eysenck,
traits represent the most salient and impor- 1985).
tant dimensions of the individuals survival The idea that personality traits have a
needs. The evolutionary perspective equally biological basis is also fundamental to Gos-
emphasizes person perception and individual lings proposal that personality psychology
differences: Because people vary systemati- must be broadened to include a comparative
cally along certain trait dimensions, and be- approach to study individual differences in
cause knowledge of others traits has adap- both human and nonhuman animals (Gos-
tive value, humans have evolved a capacity ling, 2001; Gosling, Kwan, & John, 2003).
to perceive those individual differences that Although scientists are understandably re-
are central to adaptation to the social land- luctant to ascribe personality traits, emo-
scape. The Big Five summarize these central- tions, and cognitions to animals, evolution-
ly important individual differences. ary theory predicts cross-species continuities
McCrae and Costa (1996; see also not only for physical but also for behavioral
Chapter 5, this volume) view the Big Five traits; for example, Darwin (1872/1998) ar-
as causal personality dispositions. Their gued that emotions exist in both human and
five-factor theory (FFT) is a general trait nonhuman animals. A review of 19 studies
theory that provides an explanatory inter- of personality factors in 12 nonhuman spe-
pretation of the empirically derived Big Five cies showed substantial cross-species conti-
taxonomy. One central tenet of the FFT is nuity (Gosling & John, 1999). Chimpanzees
based on the finding that all of the Big Five and other primates, dogs, cats, donkeys,
dimensions have a substantial genetic basis pigs, guppies, and octopuses all showed re-
(e.g., Loehlin et al., 1998; Plomin, DeFries, liable individual differences in Extraversion
Craig, & McGuffin, 2003; see also Krueger and Neuroticism, and all but guppies and oc-
& Johnson, Chapter 10, this volume) and topuses varied in Agreeableness as well, sug-
must therefore derive, in part, from biologi- gesting that these three Big Five factors may
cal structures and processes, such as specific capture fundamental dimensions of individu-
gene loci, brain regions (e.g., the amygdala), al differences. Further evidence suggests that
neurotransmitters (e.g., dopamine), hor- elements of Openness (such as curiosity and
mones (e.g., testosterone), and so on (e.g., playfulness) are present in at least some non-
Canli, 2006; see also Canli, Chapter 11, this human animals. In contrast, only humans
volume); it is in this sense that traits are as- and our closest relatives, chimpanzees, ap-
sumed to have causal status. McCrae and pear to show systematic individual differenc-
Costa distinguish between basic tendencies es in Conscientiousness. Given the relatively
4. Big Five Trait Taxonomy 147
complex social-cognitive functions involved not yet meet this high standard. It provides
in this dimension (i.e., following norms and descriptive concepts that still need to be ex-
rules, thinking before acting, and controlling plicated theoretically, and a nomenclature
impulses), it makes sense that Conscientious- that is still rooted in the vernacular Eng-
ness may have appeared rather recently in our lish.
evolutionary history. The careful application The Big Five structure has the advantage
of ethological and experimental methodology that everybody can understand the words
and the high interobserver reliability in these that define the factors. Moreover, the natural
studies make it unlikely that these findings language is not biased in favor of any existing
merely reflect anthropomorphic projections scientific conceptions; although the atheoret-
(c.f., Gosling et al., 2003; Kwan, Gosling, & ical nature of the Big Five dimensions makes
John, 2008). Rather, these surprising cross- them less appealing to some psychologists, it
species commonalities suggest that personal- also makes them more palatable to research-
ity traits reflect, at least in part, biological ers who reject dimensions cast in a theoreti-
mechanisms that are shared by many mam- cal mold different from their own. Whatever
malian species. the inadequacies of the natural language for
In summary, researchers subscribe to a scientific systematics, broad dimensions in-
diversity of perspectives on the conceptual ferred from folk usage are not a bad place
status of the Big Five, ranging from purely to start a taxonomy. Even in the biological
descriptive concepts to biologically based taxonomy of animals, the technical sys-
causal concepts. This diversity might be tem evolved from the vernacular (Simpson,
taken to imply that researchers cannot agree 1961, pp.1213).
about the definition of the trait concept and Obviously, a system that initially derives
that the field is in disarray. It is important from the natural language does not need to
to recognize, however, that these perspec- reify such terms indefinitely. Indeed, several of
tives are not mutually exclusive. For exam- the dimensions included among the Big Five,
ple, although Saucier and Goldberg (1996b) most notably Extraversion and Neuroticism,
caution against drawing inferences about have been the target of various physiological
genotypes from lexical studies, the lexical and mechanistic explanations (e.g., Canli et
hypothesis does not preclude the possibility al., 2001; see also L. A. Clark, 2005). In re-
that the Big Five are embodied in biological search on emotion regulatory processes, the
structures and processes. In our view, what links between the Big Five, the chronic use
is a trait? is fundamentally an empirical of particular regulatory strategies, and their
question. Research in diverse areas such as emotional and social consequences are being
behavior and molecular genetics, personality articulated (John & Gross, 2007). Similarly,
stability and change, and accuracy and bias the conceptual explication of Extraversion
in self-reports and interpersonal perception and Neuroticism as persistent dispositions
will be instrumental in building and refining toward thinking and behaving in ways that
a comprehensive theoretical account of the foster, respectively, positive and negative af-
Big Five. fective experiences (e.g., Tellegen, 1985; see
also Clark & Watson, Chapter 9, this vol-
ume) promises to connect the Big Five with
Conclusions and Implications individual differences in affective functioning,
which, in turn, may be studied in more tight-
At the beginning of this chapter, we argued ly controlled laboratory settings (see Gross,
that a personality taxonomy should provide Chapter 28, this volume). At this point, the
a systematic framework for distinguishing, Big Five differentiate domains of individual
ordering, and naming the behavioral, emo- differences that have similar surface manifes-
tional, and experiential characteristics of in- tationsjust like the early animal taxonomy
dividuals. Ideally, that taxonomy would be that was transformed by better accounts of
built around principles that are causal and evolutionary processes and by the advent of
dynamic, exist at multiple levels of abstrac- new tools, such as molecular genetics. Like-
tion or hierarchy, and offer a standard no- wise, the structures and processes underly-
menclature for scientists working in the field ing these personality trait domains are now
of personality. The Big Five taxonomy does beginning to be explicated. Explanatory and
148 II. THEORETICAL PERSPECTIVES
ter- und Persoenlichkeitsforschung (Whole No. study of personality influences on brain reactiv-
1). Bern, Switzerland: A. Francke. ity to emotional stimuli. Behavioral Neurosci-
Becker, W. C. (1960). The matching of behavior ence, 115, 3342.
rating and questionnaire personality factors. Caprara, G. V., & Perugini, M. (1994). Personal-
Psychological Bulletin, 57, 201212. ity described by adjectives: The generalizability
Belsky, J., Jaffee, S. R., Caspi, A., Moffitt, T., & of the Big Five to the Italian lexical context.
Silva, P. A. (2003). Intergenerational relation- European Journal of Personality, 8, 357369.
ships in young adulthood and their life course, Caspi, A., & Bem, D. J. (1990). Personality conti-
mental health, and personality correlates. Jour- nuity and change across the life course. In L. A.
nal of Family Psychology, 17, 460471. Pervin (Ed.), Handbook of personality: Theory
Benet-Martnez, V., & John, O. P. (1998). Los and research (pp.549575). New York: Guil-
Cinco Grandes across cultures and ethnic ford Press.
groups: Multitraitmultimethod analyses of the Caspi, A., Roberts, B. W., & Shiner, R. L. (2005).
Big Five in Spanish and English. Journal of Per- Personality development: Stability and change.
sonality and Social Psychology, 75, 729750. Annual Review of Psychology, 56, 453484.
Benet-Martnez, V., & Waller, N. G. (1997). Fur- Cattell, R. B. (1943). The description of person-
ther evidence for the cross-cultural generality ality: Basic traits resolved into clusters. Jour-
of the Big Seven Factor model: Indigenous and nal of Abnormal and Social Psychology, 38,
imported Spanish personality constructs. Jour- 476506.
nal of Personality, 65, 567598. Cattell, R. B. (1945a). The description of person-
Berkman, L. F., Glass, T., Brissette, I., & Seeman, ality: Principles and findings in a factor analysis.
T. E. (2000). From social integration to health. American Journal of Psychology, 58, 6990.
Social Science and Medicine, 51, 843857. Cattell, R. B. (1945b). The principal trait clusters
Block, J. (1995). A contrarian view of the five- for describing personality. Psychological Bulle-
factor approach to personality description. tin, 42, 129161.
Psychological Bulletin, 117, 187215. Cattell, R. B. (1990). Advances in Cattellian per-
Block, J. H., & Block, J. (1980). The role of ego- sonality theory. In L. A. Pervin (Ed.), Hand-
control and ego-resiliency in the organization book of personality: Theory and research
of behavior. In W. A. Collins (Ed.), Minne- (pp.101110). New York: Guilford Press.
sota Symposia on Child Psychology (Vol. 13, Cattell, R. B., Eber, H. W., & Tatsuoka, M. M.
pp.39101). Hillsdale, NJ: Erlbaum. (1970). Handbook for the Sixteen Personality
Bogg, T., & Roberts, B. W. (2004). Conscientious- Factor Questionnaire (16PF). Champaign, IL:
ness and health behaviors: A meta-analysis. IPAT.
Psychological Bulletin, 130, 887919. Chaplin, W. F., John, O. P., & Goldberg, L. R.
Borgatta, E. F. (1964). The structure of personal- (1988). Conceptions of states and traits: Di-
ity characteristics. Behavioral Science, 9, 817. mensional attributes with ideals as prototypes.
Botwin, M. D., & Buss, D. M. (1989). Structure Journal of Personality and Social Psychology,
of actreport data: Is the five-factor model of 54, 541557.
personality recaptured? Journal of Personality Chaplin, W. F., Phillips, J. B., Brown, J. D., Clan-
and Social Psychology, 56, 9881001. ton, N. R., & Stein, J. L. (2000). Handshak-
Botwin, M. D., Buss, D. M., & Shackelford, T. K. ing, gender, personality, and first impressions.
(1997). Personality and mate preferences: Five Journal of Personality and Social Psychology,
factors in mate selection and marital satisfac- 79, 110117.
tion. Journal of Personality, 65, 107136. Church, A. T., & Katigbak, M. S. (1989). Internal,
Burisch, M. (1984). Approaches to personality in- external, and self-report structure of personal-
ventory construction. American Psychologist, ity in a non-Western culture: An investigation
39, 214227. of cross-language and cross-cultural generaliz-
Buss, A. H., & Plomin, R. (1975). A temperament ability. Journal of Personality and Social Psy-
theory of personality development. New York: chology, 57, 857872.
Wiley. Church, A. T., Reyes, J. A. S., Katigbak, M. S.,
Buss, D. M. (1996). Social adaptation and five & Grimm, S. D. (1997). Filipino personality
major factors of personality. In J. S. Wiggins structure and the Big Five model: A lexical ap-
(Ed.), The five-factor model of personality: proach. Journal of Personality, 65, 477528.
Theoretical perspectives (pp. 180207). New Clark, J., Boccaccini, M. T., Caillouet, B., & Chap-
York: Guilford Press. lin, W. F. (2007). Five factor model personal-
Canli, T. (Ed.). (2006). Biology of personality ity traits, jury selection, and case outcomes in
and individual differences. New York: Guilford criminal and civil cases. Criminal Justice and
Press. Behavior, 34, 641660.
Canli, T., Zhao, Z., Desmond, J. E., Kang, E., Clark, L. A. (2005). Temperament as a unify-
Gross, J., & Gabrieli, J. D. (2001). An fMRI ing basis for personality and psychopathol-
150 II. THEORETICAL PERSPECTIVES
mental psychology: The unification of psychol- s pecies review. Current Directions in Psycho-
ogy and the possibility of a paradigm. Jour- logical Science, 8, 6975.
nal of Personality and Social Psychology, 73, Gosling, S. D., Ko, S. J., Mannarelli, T., & Mor-
12241237. ris, M. E. (2002). A room with a cue: Personal-
Eysenck, H. J., & Eysenck, M. W. (1985). Per- ity judgments based on offices and bedrooms.
sonality and individual differences: A natural Journal of Personality and Social Psychology,
science approach. New York: Plenum Press. 82, 379398.
Fiske, D. W. (1949). Consistency of the factorial Gosling, S. D., Kwan, V. S. Y., & John, O. P.
structures of personality ratings from different (2003). A dogs got personality: A cross-species
sources. Journal of Abnormal and Social Psy- comparative approach to evaluating personali-
chology, 44, 329344. ty judgments. Journal of Personality and Social
Friedman, H. S., Hawley, P. H., & Tucker, J. S. Psychology, 85, 11611169.
(1994). Personality, health, and longevity. Cur- Gosling, S. D., Rentfrow, P. J., & Swann, W. B.,
rent Directions in Psychological Science, 3, Jr. (2003). A very brief measure of the Big Five
3741. personality domains. Journal of Research in
Goldberg, L. R. (1981). Language and individual Personality, 37, 504528.
differences: The search for universals in per- Gough, H. G. (1979). A creative personal-
sonality lexicons. In L. Wheeler (Ed.), Review ity scale for the Adjective Check List. Jour-
of personality and social psychology (Vol. 2, nal of Personality and Social Psychology, 37,
pp.141165). Beverly Hills: Sage. 19381405.
Goldberg, L. R. (1982). From ace to zombie: Gough, H. G. (1987). The California Psychologi-
Some explorations in the language of personal- cal Inventory administrators guide. Palo Alto,
ity. In C. D. Spielberger & J. N. Butcher (Eds.), CA: Consulting Psychologists Press.
Advances in personality assessment (Vol. 1, Gough, H. G., & Heilbrun, A. B., Jr. (1983). The
pp.203234). Hillsdale, NJ: Erlbaum. Adjective Check List manual. Palo Alto, CA:
Goldberg, L. R. (1990). An alternative descrip- Consulting Psychologists Press.
tion of personality: The Big-Five factor struc- Gould, S. J. (1981). The mismeasure of man. New
ture. Journal of Personality and Social Psychol- York: Norton.
ogy, 59, 12161229. Graziano, W. G., & Eisenberg, N. (1997). Agree-
Goldberg, L. R. (1992). The development of ableness: A dimension of personality. In R. Ho-
markers for the Big-Five factor structure. Psy- gan, J. A. Johnson, & S. R. Briggs (Eds.), Hand-
chological Assessment, 4, 2642. book of personality psychology (pp.795824).
Goldberg, L. R. (1993). The structure of pheno- San Diego, CA: Academic Press.
typic personality traits. American Psychologist, Graziano, W. G., & Ward, D. (1992). Probing the
48, 2634. Big Five in adolescence: Personality and adjust-
Goldberg, L. R., & Kilkowski, J. M. (1985). ment during a developmental transition. Jour-
The prediction of semantic consistency in self- nal of Personality, 60, 425439.
descriptions: Characteristics of persons and of Guilford, J. P. (1975). Factors and factors of per-
terms that affect the consistency of responses sonality. Psychological Bulletin, 82, 802814.
to synonym and antonym pairs. Journal of Per- Halverson, C. F., Kohnstamm, G. A., & Mar-
sonality and Social Psychology, 48, 8298. tin R. P. (1994). The developing structure of
Goldberg, L. R., & Rosolack, T. K. (1994). The temperament and personality from infancy to
Big Five factor structure as an integrative frame- adulthood. Hillsdale, NJ: Erlbaum.
work: An empirical comparison with Eysencks Hampson, S. E., Andrews, J. A., Barckley, M.,
P-E-N model. In C. F. Halverson, G. A. Kohn- Lichtenstein, E., & Lee, M. E. (2000). Consci-
stamm, & R. P. Martin (Eds.), The developing entiousness, perceived risk, and risk-reduction
structure of temperament and personality from behaviors: A preliminary study. Health Psy-
infancy to adulthood (pp.735). Hillsdale, NJ: chology, 19, 247252.
Erlbaum. Hampson, S. E., & Goldberg, L. R. (2006). A first
Goldberg, L. R., Sweeney, D., Merenda, P. F., & large cohort study of personality trait stability
Hughes, J.E. (1998). Demographic variables over the 40 years between elementary school
and personality: The effects of gender, age, and midlife. Journal of Personality and Social
education, and ethnic/racial status on self- Psychology, 91, 763779.
descriptions of personality attributes. Person- Hampson, S. E., Goldberg, L. R., & John,
ality and Individual Differences, 24, 393403. O. P. (1987). Category-breadth and social-
Gosling, S. D. (2001). From mice to men: What desirability values for 573 personality terms.
can we learn about personality from animal re- European Journal of Personality, 1, 241258.
search? Psychological Bulletin, 127, 4586. Hampson, S. E., John, O. P., & Goldberg, L. R.
Gosling, S. D., & John, O. P. (1999). Personal- (1986). Category breadth and hierarchical
ity dimensions in nonhuman animals: A cross- structure in personality: Studies of asymmetries
152 II. THEORETICAL PERSPECTIVES
(Vol. 6, pp. 616). Tilburg, The Netherlands: McCrae, R. R., & Costa, P. T., Jr. (1985a). Updat-
Tilburg University Press. ing Normans adequate taxonomy: Intelligence
John, O. P., & Soto, C. J. (2007). The importance and personality dimensions in natural language
of being valid: Reliability and the process of and in questionnaires. Journal of Personality
construct validation. In R. W. Robins, R. C. and Social Psychology, 49, 710721.
Fraley, & R. F. Krueger (Eds.), Handbook of McCrae, R. R., & Costa, P. T., Jr. (1985b). Open-
research methods in personality psychology ness to experience. In R. Hogan & W. H. Jones
(pp.461494). New York: Guilford Press. (Eds.), Perspectives in personality (Vol. 1,
John, O. P., & Srivastava, S. (1999). The Big Five pp.145172). Greenwich, CT: JAI Press.
trait taxonomy: History, measurement, and McCrae, R. R., & Costa, P. T., Jr. (1985c). Com-
theoretical perspectives. In L. A. Pervin & O. parison of EPI and psychoticism scales with
P. John (Eds.), Handbook of personality: The- measures of the five-factor model of personal-
ory and research (2nd ed., pp.102138). New ity. Personality and Individual Differences, 6,
York: Guilford Press. 587597.
Karney, B. R., & Bradbury, T. N. (1995). Assessing McCrae, R. R., & Costa, P. T., Jr. (1987). Vali-
longitudinal change in marriage: An introduc- dation of the five-factor model of personality
tion to the analysis of growth curves. Journal across instruments and observers. Journal of
of Marriage and the Family, 57, 10911108. Personality and Social Psychology, 52, 8190.
Kenford, S. L., Smith, S. S., Wetter, D. W., Jorenby, McCrae, R. R., & Costa, P. T., Jr. (1996). Toward
D. E., Fiore, M. C., & Baker, T. B. (2002). Pre- a new generation of personality theories: Theo-
dicting relapse back to smoking: Contrasting retical contexts for the five-factor model. In J.
affective and physical models of dependence. S. Wiggins (Ed.), The five-factor model of per-
Journal of Consulting and Clinical Psychology, sonality: Theoretical perspectives (pp.5187).
70, 216227. New York: Guilford Press.
Klages, L. (1932). The science of character. Lon- McCrae, R. R., & Costa, P. T., Jr. (2003). Per-
don: Allen & Unwin. sonality in adulthood: A five-factor theory per-
Kohnstamm, G. A., Halverson, C. F., Mervielde, spective (2nd ed.). New York: Guilford Press.
I., & Havill, V. L. (Eds.). (1998). Parental de- McCrae, R. R., & John, O. P. (1992). An intro-
scriptions of child personality: Developmental duction to the five-factor model and its applica-
antecedents of the Big Five? Mahwah, NJ: Er- tions. Journal of Personality, 60, 175215.
lbaum. Measelle, J. R., John, O. P., Ablow, J. C., Cowan,
Krueger, R. F., & Tackett, J. L. (Eds.). (2006). P. A., & Cowan, C. P (2005). Can children pro-
Personality and psychopathology. New York: vide coherent, stable, and valid self-reports on
Guilford Press. the Big Five dimensions? A longitudinal study
Kwan, V. S. Y., Gosling, S. D., & John, O. P. from ages 5 to 7. Journal of Personality and
(2008). Anthropomorphism as a special case of Social Psychology, 89, 90106.
social perception: A cross-species comparative Mehl, M. R., Gosling, S. D., & Pennebaker, J.
approach and a new empirical paradigm. So- W. (2006). Personality in its natural habitat:
cial Cognition, 26, 129142. Manifestations and implicit folk theories of
Larson, L. M., Rottinghaus, P. J., & Borgen, F. H. personality in daily life. Journal of Personality
(2002). Meta-analyses of Big Six interests and and Social Psychology, 90, 862877.
Big Five personality factors. Journal of Voca- Mershon, B., & Gorsuch, R. L. (1988). Number
tional Behavior, 61, 217239. of factors in the personality sphere: Does in-
Loehlin, J. C., McCrae, R. R., Costa, P. T., Jr., & crease in factors increase predictability of real-
John, O. P. (1998). Heritabilities of common life criteria? Journal of Personality and Social
and measure-specific components of the Big Psychology, 55, 675680.
Five personality factors. Journal of Research in Miller, T. Q., Smith, T. W., Turner, C. W., Gui-
Personality, 32, 431453. jarro, M. L., & Hallet, A. J. (1996). A meta-
Markon, K. E, Krueger, R. F., & Watson, D. analytic review of research on hostility and
(2005). Delineating the structure of normal physical health. Psychological Bulletin, 119,
and abnormal personality: An integrative hier- 322348.
archical approach. Journal of Personality and Mount, M. K., Barrick, M. R., & Stewart, G. L.
Social Psychology, 88, 139157. (1998). Five-factor model of personality and
McAdams, D. P. (1992). The five-factor model in performance in jobs involving interpersonal in-
personality: A critical appraisal. Journal of Per- teractions. Human Performance, 11, 145165.
sonality, 60, 329361. Myers, I. B., & McCaulley, M. H. (1985). Man-
McCrae, R. R. (1996). Social consequences of ual: A guide to the development and use of the
experiential openness. Psychological Bulletin, MyersBriggs Type Indicator. Palo Alto, CA:
120, 323337. Consulting Psychologists Press.
154 II. THEORETICAL PERSPECTIVES
Newcomb, A. F., Bukowski, W. M., & Pattee, postgenomic era. Washington, DC: American
L. (1993). Childrens peer relations: A meta- Psychological Association.
analytic review of popular, rejected, neglected, Rammstedt, B., & John, O. P. (2005). Short ver-
controversial, and average sociometric status. sion of the Big Five Inventory (BFI-K): Develop-
Psychological Bulletin, 113, 99128. ment and validation of an economic inventory
Nigg, J. T., John, O. P., Blaskey, L. G., Huang- for assessment of the five factors of personality.
Pollock, C. L., Willcutt, E. G., Hinshaw, S. P., Diagnostica, 51, 195206.
et al. (2002). Big Five dimensions and ADHD Rammstedt, B., & John, O. P. (2007). Measur-
symptoms: Link between personality traits and ing personality in one minute or less: A 10-item
clinical syndromes. Journal of Personality and short version of the Big Five Inventory in Eng-
Social Psychology, 83, 451469. lish and German. Journal of Research in Per-
Noftle, E. E., & Robins, R. W. (2007). Person- sonality, 41, 203212.
ality predictors of Academic Outcomes: Big Rentfrow, P. J., & Gosling, S. D. (2006). Message
Five correlates of GPA and SAT scores. Jour- in a ballad: The role of music preferences in in-
nal of Personality and Social Psychology, 93, terpersonal perception. Psychological Science,
116130. 17, 236242.
Norman, W. T. (1963). Toward an adequate tax- Revelle, W. (1987). Personality and motivation:
onomy of personality attributes: Replicated Sources of inefficiency in cognitive perfor-
factor structure in peer nomination personality mance. Journal of Research in Personality, 21,
ratings. Journal of Abnormal and Social Psy- 436452.
chology, 66, 574583. Roberts, B. W., & DelVecchio, W. F. (2000). The
Norman, W. T. (1967). 2,800 personality trait rank-order consistency of personality traits
descriptors: Normative operating characteris- from childhood to old age: A quantitative re-
tics for a university population. Unpublished view of longitudinal studies. Psychological Bul-
paper, Department of Psychology, University of letin, 126, 325.
Michigan. Roberts, B. W., Walton, K. E., & Viechtbauer, W.
Nowakowska, M. (1973). The limitations of the (2006). Patterns of mean-level change in per-
factor-analytic approach to psychology with sonality traits across the life course: A meta-
special application to Cattells research strat- analysis of longitudinal studies. Psychological
egy. Theory and Decision, 4, 109139. Bulletin, 132, 125.
Ostendorf, F. (1990). Sprache und Persoen- Robins, R. W., Caspi, A., & Moffitt, T. E. (2002).
lichkeitsstruktur: Zur Validitaet des Fuenf- Its not just who youre with, its who you are:
Faktoren-Modells der Persoenlichkeit [Lan- Personality and relationship experiences across
guage and personality structure: On the validity multiple relationships. Journal of Personality,
of the five factor model of personality]. Regens- 70, 925964.
burg, Germany: S. Roderer Verlag. Robins, R. W., John, O. P., & Caspi, A. (1994).
Ozer, D. J., & Benet-Martnez, V. (2006). Person- Major dimensions of personality in early ad-
ality and prediction of consequential outcomes. olescence: The Big Five and beyond. In C. F.
Annual Review of Psychology, 57, 401421. Halverson, J. A. Kohnstamm, & R. P. Martin
Paulhus, D. L., & John, O. P. (1998). Egoistic (Eds.), The developing structure of tempera-
and moralistic biases in self-perception: The ment and personality from infancy to adult-
interplay of self-deceptive styles with basic hood (pp.267291). Hillsdale, NJ: Erlbaum.
traits and motives. Journal of Personality, 66, Robins, R. W., John, O. P., Caspi, A., Moffitt,
10251060. T. E., & Stouthamer-Loeber, M. (1996). Re-
Paunonen, S. V. (2003). Big Five factors of per- silient, overcontrolled, and undercontrolled
sonality and replicated prediction of behavior. boys: Three replicable personality types? Jour-
Journal of Personality and Social Psychology, nal of Personality and Social Psychology, 70,
84, 411424. 157171.
Peabody, D., & Goldberg, L. R. (1989). Some de- Rosch, E. (1978). Principles of categorization. In
terminants of factor structures from personali- E. Rosch & B. B. Lloyd (Eds.), Cognition and
ty-trait descriptors. Journal of Personality and categorization (pp.2748). Hillsdale, NJ: Erl-
Social Psychology, 57, 552567. baum.
Pervin, L. A. (1994). A critical analysis of cur- Rentfrow, P. J., & Gosling, S. D. (2006). Message
rent trait theory. Psychological Inquiry, 5, in a ballad: The role of music preferences in in-
103113. terpersonal perception. Psychological Science,
Pervin, L. A., Cervone, D., & John, O. P. (2005). 17, 236242.
Personality: Theory and research (9th ed.). Saucier, G. (1992). Openness versus Intellect:
Hoboken, NJ: Wiley. Much ado about nothing. European Journal of
Plomin, R., DeFries, J. C., Craig, I. W., & McGuf- Personality, 6, 386391.
fin, P. (Eds.). (2003). Behavioral genetics in the Saucier, G. (1994). Mini-markers: A brief version
4. Big Five Trait Taxonomy 155
of Goldbergs unipolar Big-Five markers. Jour- Five domains and 16 facets using the California
nal of Personality Assessment, 63, 506516. Psychological Inventory. Manuscript submit-
Saucier, G. (1997). Effects of variable selection ted for publication.
on the factor structure of person descriptors. Soto, C. J., John, O. P., Gosling, S. D., & Potter,
Journal of Personality and Social Psychology, J. (2008). The developmental psychometrics
73, 12961312. of Big Five self-reports: Acquiescence, factor
Saucier, G., Georgiades, S., Tsouasis, I., & Gold- structure, coherence, and differentiation from
berg, L. R. (2005). Factor structure of Greek ages 10 to 20. Journal of Personality and Social
personality-descriptive adjectives. Journal Psychology, 94, 718737.
of Personality and Social Psychology, 88, Srivastava, S., John, O. P., Gosling, S. D., & Pot-
856875. ter, J. (2003). Development of personality in
Saucier, G., & Goldberg, L. R. (199a). Evidence early and middle adulthood: Set like plaster or
for the Big Five in analyses of familiar English persistent change? Journal of Personality and
personality adjectives. European Journal of Social Psychology, 84, 10411053.
Personality, 10, 6177. Sullivan, H. S. (1953). The interpersonal theory
Saucier, G., & Goldberg, L. R. (1996b). The lan- of psychiatry. New York: Norton.
guage of personality: Lexical perspectives on Szarota, P. (1995). Polska Lista Przymiotnikowa
the five-factor model. In J. S. Wiggins (Ed.), (PLP): Narzedzie do diagnozy Pieciu Wielkich
The five-factor model of personality: Theoreti- czynnikow osobowosci [Polish Adjective List:
cal perspectives (pp.2150). New York: Guil- An instrument to assess the five-factor model
ford Press. of personality]. Studia Psychologiczne, 33,
Saucier, G., & Goldberg, L. R. (2001). Lexi- 227256.
cal studies of indigenous personality factors: Szirmak, Z., & De Raad, B. (1994). Taxonomy
Premises, products, and prospects. Journal of and structure of Hungarian personality traits.
Personality, 69, 847879. European Journal of Personality, 8, 95117.
Saucier, G., Hampson, S. E., & Goldberg, L. Tellegen, A. (1982). Brief manual for the Differ-
R. (2000). Cross-language studies of lexical ential Personality Questionnaire. Unpublished
personality factors. In S. E. Hampson (Ed.), manuscript, University of Minnesota, Minne-
Advances in personality psychology (Vol. 1, apolis.
pp.136). London: Routledge. Tellegen, A. (1985). Structures of mood and per-
Saucier, G., & Ostendorf, F. (1999). Hierarchi- sonality and their relevance to assessing anxi-
cal subcomponents of the Big Five personal- ety, with an emphasis on self-report. In A. H.
ity factors: A cross-language replication. Jour- Tuma & J. D. Maser (Eds.), Anxiety and the
nal of Personality and Social Psychology, 76, anxiety disorders (pp.681716). Hillsdale, NJ:
613627. Erlbaum.
Scarr, S., & McCartney, K. (1984). How people Tellegen, A. (1993). Folk concepts and psycho
make their own environments: A theory of logical concepts of personality and per-
genotypeenvironment effects. Annual Prog- sonality disorder. Psychological Inquiry, 4,
ress in Child Psychiatry and Child Develop- 122130.
ment, 98118. Tellegen, A., Ben-Porath, Y. S., McNulty, J. L.,
Scheier, M. F., & Carver, C. S. (1993). On the Arbisi, P. A., Graham, J. R., & Kaemmer, B.
power of positive thinking: The benefits of be- (2003). MMPI-2 Restructured Clinical (RC)
ing optimistic. Current Directions in Psycho- Scales: Development, validation, and interpre-
logical Science, 2, 2630. tation. Minneapolis: University of Minnesota
Shiner, R., & Caspi, A. (2003). Personality dif- Press.
ferences in childhood and adolescence: Mea- Thoresen, C. J., Kaplan, S. A., Barsky, A. P., War-
surement, development, and consequences. ren, C. R., & de Chermont, K. (2003). The af-
Journal of Child Psychology and Psychiatry, fective underpinnings of job perceptions and
44, 232. attitudes: A meta-analytic review and integra-
Shmelyov, A. G., & Pokhilko, V. I. (1993). A tion. Psychological Bulletin, 129, 914945.
taxonomy-oriented study of Russian personal- Trapnell, P. D., & Wiggins, J. S. (1990). Exten-
ity-trait names. European Journal of Personal- sion of the Interpersonal Adjective Scales to
ity, 7, 117. include the Big Five dimensions of personality.
Simpson, G. G. (1961). Principles of animal tax- Journal of Personality and Social Psychology,
onomy. New York: Columbia University Press. 59, 781790.
Somer, O., & Goldberg, L. R. (1999). The struc- Trull, T. J., & Sher, K. J. (1994). Relationship be-
ture of Turkish trait-descriptive adjectives. tween the five-factor model of personality and
Journal of Personality and Social Psychology, Axis I disorders in a nonclinical sample. Jour-
76, 431450. nal of Abnormal Psychology, 103, 350360.
Soto, C. J., & John, O. P. (2008). Measuring Big Tupes, E. C., & Christal, R. C. (1961). Recur-
156 II. THEORETICAL PERSPECTIVES
rent personality factors based on trait ratings. Wiggins, J. S. (1968). Personality structure. In P.
Technical Report, U.S. Air Force, Lackland Air R. Farnsworth (Ed.), Annual review of psy-
Force Base, TX. chology (Vol. 19, pp.320322). Palo Alto, CA:
Tupes, E. C., & Christal, R. C. (1992). Recurrent Annual Reviews.
personality factors based on trait ratings. Jour- Wiggins, J. S. (1979). A psychological taxonomy
nal of Personality, 60, 225251. of trait-descriptive terms: The interpersonal
van Lieshout, C. F. M., & Haselager, G. J. T. domain. Journal of Personality and Social Psy-
(1994). The Big Five personality factors in Q- chology, 37, 395412.
sort descriptions of children and adolescents. Wiggins, J. S. (1995). Interpersonal Adjective
In C. F. Halverson, G. A. Kohnstamm, & R. Scales: Professional manual. Odessa, FL: Psy-
P. Martin (Eds.), The developing structure of chological Assessment Resources.
temperament and personality from infancy to Wiggins, J. S. (Ed.). (1996). The five-factor model
adulthood (pp. 293318). Hillsdale, NJ: Erl- of personality: Theoretical perspectives. New
baum. York: Guilford Press.
Vazire, S., & Gosling, S. D. (2004). e-Perceptions: Wiggins, J. S. (1997). In defense of traits. In R. Ho-
Personality impressions based on personal gan, J. A. Johnson, & S. R. Briggs (Ed.), Hand-
websites. Journal of Personality and Social Psy- book of personality psychology (pp.649679).
chology, 87, 123132. San Diego, CA: Academic Press.
Watson, D., & Clark, L. A. (1997). Extraversion Wiggins, J. S., & Pincus, A. L. (1989). Concep-
and its positive emotional core. In R. Hogan, J. tions of personality disorders and dimensions
A. Johnson, & S. R. Briggs (Eds.), Handbook of personality. Psychological Assessment, 1,
of personality psychology (pp. 767793). San 305316.
Diego, CA: Academic Press. Wiggins, J. S., & Trapnell, P. D. (1996). A dyadic
Watson, D., Hubbard, B., & Wiese, D. (2000). interactional perspective on the five-factor
General traits of personality and affectivity as model. In J. S. Wiggins (Ed.), The five-factor
predictors of satisfaction in intimate relation- model of personality: Theoretical perspectives
ships: Evidence from self- and partner-ratings. (pp.180207). New York: Guilford Press.
Journal of Personality, 68, 413449. Yang, K. S., & Bond, M. H. (1990). Exploring
Weiss, A., & Costa, P. T., Jr. (2005). Domain and implicit personality theories with indigenous or
facet personality predictors of all-cause mor- imported constructs: The Chinese case. Jour-
tality among Medicare patients aged 65100. nal of Personality and Social Psychology, 58,
Psychosomatic Medicine, 67, 724733. 10871095.
APPENDIX 4.1. Big Five Inventory Response Form and Instructions to Participants
Instructions: Here are a number of characteristics that may or may not apply to you. For example, do
you agree that you are someone who likes to spend time with others? Please write a number next to each
statement to indicate the extent to which you agree or disagree with that statement.
3
1 2 4 5
Neither agree
Disagree strongly Disagree a little Agree a little Agree strongly
nordisagree
Note. From John, Donahue, and Kentle (1991). Copyright 1991 by Oliver P. John. Reprinted by permission.
157
APPENDIX 4.2.Scoring the BFI Scales and Acquiescence Index, and Ipsatizing the BFI Items
Computing the Content-Balanced Acquiescence Index and Ipsatizing the BFI Items
To index individuals acquiescent response style (where high scores mean yea-saying and low scores
mean nay-saying), we compute their acquiescence score as their mean response across 32 BFI items
that form 16 pairs of items with opposite implications for personality (e.g., item 1 Is talkative and
item 21 Tends to be quiet). As described in Soto, John, Gosling, and Potter (2008), we devised these
16 pairs of opposite items on the basis of item content and the size of their (negative) interitem correla-
tions. Using the standard BFI item numbers (see Appendix A; see also Benet-Martnez & John, 1998;
John & Srivastava, 1999), these item pairs are 1 and 21, 6 and 16, 31 and 36, 2 and 17, 7 and 12, 27
and 42, 32 and 37, 3 and 43, 8 and 13, 18 and 33, 23 and 28, 9 and 19, 24 and 29, 34 and 39, 5 and
35, 30 and 41.
The Statistical Package for the Social Sciences (SPSS) syntax given below first computes each per-
sons acquiescence score (bfiave; the average of his or her 16 2 = 32 item responses) and response
extremeness (bfistd; the standard deviation of that persons 32 item responses). Both of these individu-
al-difference scores may be retained and used in research on individual or group differences in response-
scale use (e.g., comparing Asian Americans and European Americans). The syntax below shows how
these two scores can be used to ipsatize the full set of 44 BFI items, by removing from each item score
the individuals acquiescence score (i.e., content-balanced response mean) and then adjusting the result-
ing deviation scores by dividing them by the individuals standard deviation, resulting in person-centered
standard (or Z) scores. The syntax below assumes that the variables for the 44 BFI items are named bfi1
to bfi44 in standard order.
SPSS Syntax to Ipsatize the 44 BFI Items before Scoring the Scales
* Compute within-person response means (bfiave) and standard deviations
(bfistd).
COMPUTE bfiave = mean(bfi1, bfi6, bfi16, bfi21, bfi31, bfi36, bfi2, bfi7, bfi12,
bfi17, bfi27, bfi32, bfi37, bfi42, bfi3, bfi8, bfi13, bfi18, bfi23, bfi28, bfi33,
bfi43, bfi9, bfi19, bfi24, bfi29, bfi34, bfi39, bfi5, bfi30, bfi35, bfi41).
COMPUTE bfistd = sd(bfi1, bfi6, bfi16, bfi21, bfi31, bfi36, bfi2, bfi7, bfi12, bfi17,
bfi27, bfi32, bfi37, bfi42, bfi3, bfi8, bfi13, bfi18, bfi23, bfi28, bfi33, bfi43, bfi9,
bfi19, bfi24, bfi29, bfi34, bfi39, bfi5, bfi30, bfi35, bfi41).
EXECUTE.
Then use the ipsatized item scores to compute scale scores as mean item responses, as described above.
158