general methodological issues - sage publications€¦ · conclusions about the universe would...
TRANSCRIPT
PART I
General Methodological Issues
9781412946513-Ch01 1/9/08 2:58 PM Page 27
9781412946513-Ch01 1/9/08 2:58 PM Page 28
Measures of the PersonalityFactors Found Recurrently in
Human Lexicons
Gerard Saucier
How can attributes of personality best beorganized and measured? Answers to thiscrucial scientific question provide the foun-dation not only for personality tests, but alsofor much research on personality and individ-ual differences. Studies of natural languagesprovide an important source of answers. Inthis chapter I review the approach used insuch lexical studies of personality attributes,as well as basic findings and major measuresassociated with these studies.
Lexical measures of personality factors areused primarily in research settings. Becausethe items themselves are terms from the lex-icon, they are easily embedded within lexical-study stimuli, where they provide the mostdirect representation of lexical factors. Theyhave also proven to be extremely useful tem-plates for the development of more sophisti-cated assessment instruments. Moreover,because lexical factors have a solid content-validity basis, they can be used in the valida-tion of other measures. This chapter presents anarray of measures for lexical personality fac-tors, concentrating on those measures based
most directly on lexical structures; that is,those designed to be markers of these struc-tures. Inquiries into the structure of attributeshinge strongly on how personality is defined.Therefore, the definition of personality is agood place to begin a discussion of structure.
DEFINING PERSONALITY
Definitions make one’s assumptions explicit.How one defines personality is consequential,affecting how one selects variables when study-ing personality. There is no single canonicaldefinition in current use. Personality is definedeither as (a) a set of attributes characterizingan individual, or as (b) the underlying systemthat generates the set of attributes. Funder(1997) provided a definition that includes both (a) and (b): Personality is ‘an individual’scharacteristic patterns of thought, emotion, and behavior, together with the psychologicalmechanisms – hidden or not – behind those pat-terns’ (1997: 1–2). Funder refers to a broad
1
9781412946513-Ch01 1/9/08 2:58 PM Page 29
array of attributes that simultaneously are (i)ascribed to individuals, (ii) stable over time,and (iii) psychological in nature.
But there are other ways to define person-ality. In a classic early textbook, Allport (1937)catalogued 50 distinct meanings found indefinitions of personality. These meanings canbe arrayed in a continuum ranging from one’sexternally observable manner to one’s inter-nal self. Reacting against broad omnibus def-initions of personality (e.g. Prince, 1924),Allport’s definition – ‘personality is thedynamic organization within the individualof those psychophysical systems that deter-mine his unique adjustments to his environ-ment’ (1937: 48) – highlights attributes thatare seen as residing ‘within’ the individual.
However, other ways of defining personality,consistent with a ‘biosocial’ view that Allportdeprecated, emphasize attributes that are moreexternal or that involve the effect the indi-vidual tends to have on others. These include (a) attributes of external appearance (includ-ing qualities like physical size), (b) attributesassociated with the role one assumes or thestatus one has achieved in society (e.g. pro-fessional, motherly, famous), and (c) attrib-utes of an evocative type, that involve thepattern of reactions that the individual gener-ates in others given the kind of stimulus s/heis. (e.g. charming, intimidating, boring, believ-able, lovable, respected, offensive). Such socialeffects represent a person’s social stimulusvalue (Allport, 1937: 41; based on May, 1932).
Another class with controversial status aspersonality attributes is that containinghighly evaluative terms (e.g. stupid, evil,abnormal, good). Most personality conceptsare decidedly evaluative (clearly favorable orunfavorable; Goldberg, 1982), but these aredistinct in the high ratio of the evaluative tothe descriptive component. Highly evaluativeterms are not ‘pure evaluation’; one can finddescriptive dimensions from selections con-sisting purely of such terms (Benet-Martinezand Waller, 2002), so they do have somedescriptive component.
What about patterns of belief and attitudes?Allport (1937) generally regarded attitudes
as behavioral dispositions of a specific andexternal sort, being ‘bound to an object orvalue’ (1937: 294); that is, aroused in thepresence of a specifiable class of stimuli. If,however, an attitude is ‘chronic and tempera-mental’, expressed in almost any sphere ofthe person’s behavior’ (1937: 294), as in forexample radicalism or conservatism, then forAllport it differed little from a trait. Thus,generalized attitudes – those for which it isdifficult to specify the object – can be consid-ered personality traits. Factors derived fromthe correlations among large numbers of morespecific attitudes and beliefs define traits, inthat they represent consistent patterns acrossmany attitude objects.
Values can be seen as beliefs regarding‘how one ought or ought not to behave, orabout some end-state of existence worth ornot worth attaining’ (Rokeach, 1968: 124),and, echoing Allport’s distinctions, ‘not tiedto any specific attitude object or situation’(Rokeach, 1968: 124). Super (1995) charac-terized interests as related to values, beingpreferences for classes of activities in whichindividual expect to attain their values. Interestsinvolve assessing objects according to howliked or disliked they are (rather than theirfavorability or importance more generally).Career-interest measures show even higherstability than do personality measures (Lowet al., 2005). And there are dimensions ofvariation in career-interest items that are rel-atively independent of currently popular traitdimensions (Ackerman, 1996; Ackerman andHeggestad, 1997).
Including all such additional variables,one arrives at a broader definition of person-ality: all of the relatively stable attributes,qualities, or characteristics that distinguish thebehavior, thoughts, and feelings of individu-als. Such a broad definition is close to thatproposed by Roback (1931): ‘an integrativecombination of all our cognitive (knowledge),affective (feeling), conative (volitional) andeven physical qualities’ (1931: 31–32).
Such broad definitions are not unusual.However, since Allport and Odbert (1936),personality has often been defined broadly
30 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 30
but operationalized narrowly, so that manyclasses of relevant variables are excluded.These narrow variable selections have beenachieved by either (a) starting with a fullrange of attributes of persons and then purg-ing those judged to fall in categories consid-ered unsuitable using exclusion rules (e.g.Allport and Odbert, 1936; Ashton et al.,2004; Goldberg, 1990; Norman, 1967), or (b) relying on vaguely defined ‘personalityrelevance’ ratings of judges. In effect, a nar-rower definition is being used, that personal-ity is patterns of behavior (including stableaffective tendencies but generally not pat-terns of thinking) that are believed to residewithin the individual and that cannot be dis-qualified as attitudes, temporary states, socialeffects, or social roles, or because they areoverly evaluative. Such definition-by-exclusionmakes personality into a remarkably gerry-mandered construct.
Previous research indicates that the structureof personality attributes encoded in lexiconsdepends in major ways upon the upstreamselection of variables. This is unsurprising. Ifastronomers forbade themselves from inves-tigating regions of the sky beyond that narrowband of the firmament where the most obvi-ous objects of interest (the sun, the moon, theplanets) move across the sky, astronomy’sconclusions about the universe would cer-tainly be altered. To remove the risk that weignore important phenomena and miss majordiscoveries, we need a wider view. We shouldcouple our focus on the most prototypicalattributes of personality with a simultaneous‘bigger picture’ examination of a wider rangeof psychological attributes.
PARSIMONY IN PERSONALITYMODELS
How many important traits are there? Survey-ing the scales in current personality invento-ries, one finds a bewildering variety ofconstructs. And if one turns to single wordsin modern world languages, the situation
becomes overwhelming: Allport and Odbert(1936), for example, found nearly 18,000words in Webster’s Second InternationalDictionary referring to characteristics thatmight distinguish one human being fromanother. One needs a parsimonious summaryof this vast domain of concepts.
In the field of personality the search for ascientifically compelling classification of thehuge number of personality attributes excitesincreasing interest. A classification systemat-ically divides phenomena into ordered groupsor categories; it ‘chunks’ things. A scientificclassification helps organize and integrateknowledge and research findings, providinga standard scientific nomenclature that facil-itates communication and aids in the accu-mulation of empirical findings. Becausepersonality attributes describe continua andnot categories, such a classification will nat-urally be a ‘dimensional classification’ – morelike those used for classifying colors thanlike those for classifying species.
In constructing a classification a variety ofprocedures could be used to group the phe-nomena under study. The most useful is a classof statistical methods generically referred toas factor analysis. Factor analysis can be con-sidered a variable-reduction procedure, inwhich many variables are organized by a fewfactors that summarize the interrelations amongthe variables (Goldberg and Velicer, 2006).
THE BASIS FOR THE LEXICALAPPROACH
However, prior to conducting factor analysis,one must determine which variables toinclude in the analysis. Variable selection isvitally dependent on how personality isdefined. It is also guided to some degree bythe investigator’s beliefs about the criteria for the goodness of a structural model (seeSaucier and Simonds, 2006, for a listing ofsuch criteria).
As has long been recognized (e.g. Allportand Odbert, 1936; Cattell, 1943; Goldberg,
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 31
9781412946513-Ch01 1/9/08 2:58 PM Page 31
1981; Norman, 1963), basic personalitydimensions might be discovered by studyingconceptions embedded in the natural lan-guage. The key premise of the lexicalapproach is this: The degree of representa-tion of an attribute in language has some cor-respondence with the general importance ofthe attribute in real-world transactions. Ifterms in a language are used as variables, anattribute that is represented by multiple termsin that language will likely appear as a factor.Moreover, if the factor includes terms thatare used with high frequency, the importanceof the factor is underscored.
Such factors are but a starting point forseveral reasons. The lexicon could omit orunderemphasize some scientifically impor-tant variables. Moreover, the meaning ofsingle natural-language terms can be vague,ambiguous, or context-dependent (John etal., 1988). Folk concepts of personality(Tellegen, 1993) provide basic but not exhaus-tive (necessary but not sufficient) componentsfor a science of personality attributes (Goldbergand Saucier, 1995). These components operateon the descriptive or phenotypic level, with-out implication as to what might be theunderlying biological or other causal basis.An established causal basis is an important cri-terion for the goodness of a structural model(Saucier and Simonds, 2006). Ultimately, astructural model of personality ought to alignthe descriptive level with the causal level,and there may turn out not to be perfecthomology between the two levels.
Nonetheless, lexicalized concepts – espe-cially those represented in very frequentlyused words – tend to have high social impor-tance. So variables and factors based on lex-icalized concepts have a virtual guarantee ofbeing important. Lexicalized concepts can befound in standard sources created by disinter-ested parties (e.g. linguists and lexicogra-phers), and basing variable selection on sucha source reduces the likelihood of investiga-tor bias in the selection process. And becauselexicalized concepts constitute a finite domain,one can sample them representatively and so establish content-validity benchmarks for
personality variables. For drawing conclu-sions regarding personality structure, theseconcepts thereby have a major advantageover statements and sentences: Drawing onthe generative capacity of a human language,a nearly infinite number of personality-descriptive sentences might be formed,meaning that establishing that any selectionof statements and sentences is representativewould be quite difficult.
The lexical-study paradigm gives specialimportance to one other demanding criterion.Cross-cultural generalizability can be used tojudge among competitor structures. Structuralmodels derived within one limited popula-tion, or a limited sample from that population,are prone to reflect the unique patterns foundwithin that population or sample. Althoughculture-specific patterns are certainly inter-esting, models that transfer well – acrosspopulations, languages, and socioculturalsettings – better satisfy scientific standards ofreplicability and generalizability.
We can apply this criterion in either alenient or a stringent way. The lenient way isto export a set of variables (e.g. those in asingle personality inventory) for use in otherpopulations, and then examine whether thesepreselected variables (after translation, ifnecessary) generate the same factor structurein each new language or culture (as inRolland et al., 1998; Rossier et al., 2005). Ifthe inventory’s scales generate similar factorsacross populations, one might argue (as inMcCrae and Costa, 1997) that the structure iswidely generalizable. A more stringent test isto identify the most salient and importantpersonality concepts within each linguistic/cul-tural context, derive an indigenous factorstructure from those variables, and then exam-ine how much this new structure corre-sponds to previously proposed structures. A structure that met this demanding test in anylanguage could be considered more truly ubiq-uitous and universal than a merely ‘translat-able’ structure.
The lexical approach involves such anindigenous research strategy. Analyses arecarried out separately within each language,
32 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 32
using a representative set of native-languagedescriptors, rather than merely importingselections of variables from other languages(e.g. English).
The following review will detail the struc-tures that have emerged from lexical studiesof some 16 languages, and that appear mostreplicable. These structures involve alterna-tively one, two, three, five, six, and sevenfactors. In all cases, measures of lexicallyderived factors will be described in conjunc-tion with the structure.
What if we were constrained to only one factor?
Several lexical studies have reported evi-dence about factor solutions containing onlyone factor (Boies et al., 2001; Di Blas andForzi, 1999; Goldberg and Somer, 2000;Saucier, 1997, 2003b; Saucier et al., 2006).Findings have been quite consistent. Thesingle factor contrasts a heterogeneous mixof desirable attributes at one pole with a mixof undesirable attributes at the other pole.This unrotated factor can be labelled ‘evalu-ation’. A more specific interpretation, whichfits reported findings from lexical studies,would be ‘virtues’ versus ‘bad character’.
Evaluation is the first factor to emerge inthe cognitions of young children. Whereasolder children employ more differentiated traitconcepts, younger children typically rely onglobal, evaluative inference (Alvarez et al.,2001). One can refer also to a classic findingin cognitive psychology: In judgments aboutthe meanings of diverse objects in a widearray of cultural settings, a global evaluationfactor (good vs. bad) was found recurrentlyto be the first and largest factor (Osgood,1962; Osgood et al., 1975). Osgood hypoth-esized that the ubiquity of this evaluativefactor was related to basic evolutionary prin-ciples: Our forebears would not have sur-vived if they had not become adapted at a verybasic level to any signals of good versus badobjects or events – those to approach versusthose to avoid, those leading to pleasure versus
those leading to pain (e.g. ‘Can I eat it or willit eat me?’). This motivational dimension –what is liked and approached, as opposed towhat is disliked and avoided – provides onepossible theoretical account for the one-factor model. There is no widely used meas-ure of this ‘Big One’ factor. Indeed, the factorhas had relatively little attention in personal-ity studies. This contrasts strikingly with thesituation in the field of cognitive abilitieswhere a one-factor taxonomy has long beendominant (Carroll, 1993).
For measuring a general evaluation factor,several research measures are available.Saucier (1994b) developed an adjectivalmarker scale for the single ‘general evalua-tion’ (Ge) factor. This scale was intended to berelatively orthogonal to four non-evaluativedimensions derived in the same study. Thecontent at the favorable pole was character-ized as largely a combination of likeability,good judgment, and perceived maturity.Constituent terms and psychometric indicesare provided in Table 1.1, both for the longer24-adjective scale (Ge-24) and a briefer 12-item subset (Ge-12). An alternative meas-ure was developed specifically to minimizecorrelations with the octant scales for theNon-Evaluative Personality Circumplex(NEPC) (Saucier et al., 2001; described laterunder three-factor models). Terms andindices for this scale (NEPC-E) are also pre-sented in the table. As part of a study of thestructure of English type-nouns, Saucier(2003b) used an economical ten-adjectivemarker scale for the one broad factor(derived from the Big One factor in the lexi-cal study of Saucier, 1997) labeled ‘sociallydesirable qualities’. As another alternative,one could employ terms from the bipolarscales recommended by Osgood et al., 1975,table 4:18), among which good–bad, pleas-ant–unpleasant, nice–awful, and beautiful–uglyproved the most ubiquitously useful across awide range of cultural settings. Table 1.1 pres-ents such a set. A characteristic of Osgood’sitems is that they can be used to describeinanimate objects as well as animals orpeople, because they use terms (e.g. pleasant,
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 33
9781412946513-Ch01 1/9/08 2:58 PM Page 33
34 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 34
beautiful) without strong and specific moral/ethical connotations.
Several psychometric indices are includedin Table 1.1 and succeeding tables. Two ref-erence internal consistency: coefficient alphaand the mean inter-item correlation. Two are relevant to unidimensionality: (a) thestandard deviation of the inter-item correla-tions, which decreases as unidimensionalityincreases, and (b) the ratio of variance betweenthe first and second unrotated factors fromthe scale items, which becomes more lop-sided as unidimensionality increases. Finally,the table includes the scale mean, wherescores are the average response on a 1-to-7multipoint rating scale, as well as the scalestandard deviation and the skewness statistic(where values less than −1 or greater than +1indicate extreme negative or positive skew,respectively). Comparing the five alternativemarker scales with respect to these indices, it appears that the Ge-24 and Ge-12 scalesare superior, as they combine strong internalconsistency and unidimensionality withsomewhat less skewness than the other measures.
If dimensions of psychopathology are con-strained to be only one, that dimension wouldrepresent general maladjustment. Generalmaladjustment is probably strongly related tothe evaluation factor in personality. One dif-ference is that studies of psyhopathologyunderstandably pay little attention to favor-able qualities. Abnormal psychology tends tocontrast varieties of dysfunction with themere absence of dysfunction (i.e. normality).
The big two
Two-factor solutions from lexical studiesalso suggest a consistent pattern: One factorincludes attributes associated with positivelyvalued dynamic qualities and individual ascen-dancy, whereas the other factor includes attrib-utes associated with socialization, socialpropriety, solidarity, and community cohesion(Caprara et al., 1997; Di Blas and Forzi, 1999;Digman, 1997; Goldberg and Somer, 2000;
Hrebíckov· et al., 1999; Paulhus and John,1998; Saucier, 1997, 2003b; Saucier et al.,2005, 2006; Shweder, 1972; White, 1980).These two factors may be aligned with someof the other sets of dual personological con-structs reviewed by Digman (1997) and byPaulhus and John (1998), including Hogan’s(1983) distinction between ‘getting ahead’(dynamism) and ‘getting along’ (social propri-ety). They seem also to resemble higher-order factors of the Big Five (DeYoung,2006; Digman, 1997).
To date, this two-factor structure appearsto be as ubiquitous across languages and cul-tures as the one-factor structure. Moreover,like the one-factor structure and unlike struc-tures described later, it appears to be relativelyimpervious to variable-selection effects.These two factors seem to appear whetherthere is a relatively restricted or inclusiveselection of variables (Saucier, 1997), andwhether one studies adjectives or type-nouns(Saucier, 2003b) or even more diverse com-binations of variable types (De Raad andBarelds, 2006; Saucier et al., 2006). Not yetknown is the extent to which the two-factorswill be robust across even broader selectionsof variables (e.g. those that also include vari-ables representing beliefs, attitudes, values,and interests). If both the one- and two-factorstructures eventually turn out to be universal,the latter has a clear advantage, because twofactors provide more information than one.
No consensual theory is as yet associatedwith the Big Two, but Paulhus and John(1998) reviewed a number of theories associ-ated with two-factor structures of personality.De Young has specifically proposed that thetwo higher-order factors, which he labels‘stability’ and ‘plasticity’, are related respec-tively to individual differences in serotoninand in dopamine functioning. These two fac-tors might alternatively stem from the opera-tion of basic human motivations that operatein the observer: ‘social propriety’ might ref-erence the degree to which an observed personis safe versus dangerous or hazardous (i.e. pun-ishing) for others, whereas ‘dynamism’ mightreference the degree to which an observed
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 35
9781412946513-Ch01 1/9/08 2:58 PM Page 35
person is stimulating versus boring (i.e. reward-ing) for others. Studies are needed to evaluatethese hypotheses.
There are as yet no widely used standardmeasures of the Big Two. Measures of theinterpersonal circumplex (e.g. Wiggins et al.,1988) will not serve, because its two dimen-sions are too narrow – omitting contribu-tions, for example, of openness/intellect,conscientiousness, and emotional stability.The same is true of Eysenck’s older ‘BigTwo’ – extraversion and neuroticism – whichobviously leave out contributions of a differ-ent combination of three Big Five factors.
As markers of the Big Two in lexical stud-ies in newly studied languages, Saucier hasused a relatively brief collection of Englishadjectives derived from the two-factor struc-ture in an English lexical analysis (Saucier,1997). Constituent terms and psychometricindices for these ‘initial approximation’scales are provided in Table 1.1.
There may be strong homology betweenstructures in the domains of personality andpsychopathology at the two-factor level. A favored two-dimensional model for psy-chopathology separates externalizing andinternalizing disorders, conceived as two cor-related factors (e.g. Krueger and Markon,2006). A reasonable hypothesis is that exter-nalizing disorders represent low social pro-priety (morality) whereas internalizingdisorders have a stronger relation to lowdynamism. More studies are needed to estab-lish homologies between domains at the two-factor level. Just as the single evaluativefactor is a higher-order combination of thefavorable poles of the Big Two, the singlepsychopathology factor (i.e. maladjustment)is a higher-order combination of the external-izing and internalizing dimensions.
Personality descriptors in three-dimensional space
In three-factor solutions, studies of most lan-guages of European origin (plus those inTurkish, Korean, and Chinese) have produced
factors corresponding to extraversion, agree-ableness, and conscientiousness. This structurewas not observed in Filipino, French, Greek,or Maasai studies. Still, this three-factor struc-ture does appear readily in a large subset oflanguages, and in more languages than theBig Five (De Raad and Peabody, 2005).
Peabody (1987; Peabody and Goldberg,1989) demonstrated that the unrotated-factorversions of this Big Three can be labeled asevaluation, assertive versus unassertive (oraggressive vs. accommodating), and tightversus loose (or impulse control vs. impulseexpression). The first two of these are themost ubiquitous, as they rotate into the socialpropriety (morality) and dynamism factorsthat make up the Big Two. The Big Threedoes not replicate in all lexical studies simplybecause a tight–loose factor does not neces-sarily appear third, but rather sometimesfourth or later, in the sequence of unrotatedfactors.
For the rotated versions of these threedimensions, scales for the first three of the BigFive – that is, for extraversion, agreeable-ness, and conscientiousness (see Table 1.3) –will function reasonably well. But the unro-tated versions are also of interest, becausethey concentrate social-desirability respond-ing in only one of the factors (i.e. evaluation)and thus remove it from the other two. Thiswas demonstrated by Saucier et al. (2001),who likewise showed that these unrotatedfactors are quite similar in English and inGerman. This set of factors includes oneevaluative factor and two non-evaluative fac-tors, and the latter were presented as a non-evaluative circumplex (cf. Di Blas et al.,2000). Saucier et al. (2001) provided psycho-metric indices for the octant scales taken sep-arately. These scales produce unusuallysymmetric (non-skewed) distributions buttend to be multidimensional and only moder-ately homogeneous.
An additional three-factor model is theaffective-meaning dimensions of Osgoodand colleagues, which have a quasi-lexicalbasis, being drawn from ratings of a widevariety of objects and entities. The most
36 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 36
ubiquitous bipolar-scale markers for activityand potency across cultures (Osgood et al.,1975, table 4:18) appear to be strong–weak,big–little/small, and heavy–light (for potency),and fast–slow, young–old, active–passive, andalive–dead (for activity). Although the threeOsgood dimensions are known to apply wellacross a very broad range of target entities,activity and potency have not provided a par-ticularly good account of lexical factors.
Another three-factor alternative is evidentin the convergence between models ofEysenck (Eysenck and Eysenck, 1975), ofTellegen and colleagues (Tellegen, in press;Clark and Watson, 1999), and of Rothbart(Rothbart and Bates, 1998), all of whichshare an emphasis on affect and on biologicalbases of temperament. One factor is extraver-sion, approach, or positive emotionality. A second is neuroticism, negative affectivity,or negative emotionality. A third is psychoti-cism (which might be better labeled as somecombination of psychopathy and impulsivesensation seeking), constraint (labeled by theopposite pole), or effortful control (labeledby the opposite pole). Although this model isprominent in contemporary psychology, it isyet to be reported from a lexical study, per-haps because it tends to omit content fromagreeableness, a large and prominent con-stituent of the personality lexicon.
Saucier (1997) found that, for Englishadjectives, this structure was as robust acrossvariable selections as were the one- and two-factor structures described previously.However, that remains the only demonstra-tion of this sort. Saucier’s (2003b) study ofthe structure of English type-nouns failed toconfirm this three-factor structure, althoughit did confirm the Big One and Big Two. Andthe three factors did not appear in two recentlexical studies with more inclusive selectionsof variables (Saucier et al., 2005; Saucier etal., 2006). A conclusion is that they are notvery robust across variable selections.
The same variable-selection caveat per-tains to the next two structural models to be discussed. In the case of the Big Five and the Cross-Language Six, the structure
seems to be dependent on a narrow way ofoperationalizing personality (using exclusionrules). And all of these models may bedependent on the use of adjectives, to theexclusion of other word-forms. In order toincrease our understanding of the contingen-cies between variable selection and obtainedstructure, all lexical studies should ideallycompare results from a conventional, narrowvariable selection with that from a moreinclusive selection of variables (as in Saucier,1997; Goldberg and Somer, 2000; Saucier et al., 2006).
At this point the reader may be interestedin how the one-, two-, and three-factor levelsare related. Table 1.2 provides the correla-tions among all of the adjective marker scalesdescribed in this chapter; some of the highercorrelations are affected by item overlapbetween marker sets at different levels. The general evaluation factor, regardless ofthe scale for it, is related to both S (socialpropriety) and D (dynamism) but more to Sthan to D.
Regularities at the five-factor level
The Big Five factors are extraversion, agree-ableness, conscientiousness, emotional sta-bility, and intellect/imagination. Lexicalstudies in Germanic and Slavic languages(including English) have been supportive ofthe Big Five, and so has a study in Turkish(Goldberg and Somer, 2000). But studies inItalian (De Raad et al., 1997) and Hungarian(Szirmak and De Raad, 1994) found nocounterpart to the intellect factor in five-factor solutions. Extraction of additional fac-tors was necessary to find a factor related tointellect. In a study of modern Greek(Saucier et al., 2005), there was no intellector imagination factor (intellect terms weremore associated with a factor emphasizingcourage and self-confidence).
Several lexical studies have included a rel-atively broad selection of variables, eachincluding many terms that could be classifiedas referring to emotions and moods or as
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 37
9781412946513-Ch01 1/9/08 2:58 PM Page 37
38 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
Tabl
e 1.
2In
terc
orre
lati
ons
Am
ong
Adj
ecti
ve M
arke
r Sc
ales
For
Str
uctu
res
of O
ne t
o Se
ven
Fact
ors
(N=
533,
deci
mal
poi
nts
omit
ted)
1.G
e-24
2.G
e-12
953.
NEP
C-E
6260
4.SD
Q55
5452
5.O
sgoo
d E
5051
4568
6.S
4949
4775
587.
D36
3340
4341
048.
TL05
03−0
405
−07
00−2
29.
TALU
−04
−08
−06
−14
−20
−35
1043
10.A
U−0
1−0
3−0
3−0
8−0
7−4
543
1158
11.A
LUT
0404
0804
04−2
157
−50
0340
12.B
5MM
-I27
2531
3330
0084
−13
0944
5713
.B5M
M-II
5253
5171
6575
27−2
5−4
6−2
904
2114
.B5M
M-II
I39
3531
5840
3620
4811
10−1
917
3115
.B5M
M-IV
4644
4850
3852
1911
06−1
200
1937
2816
.B5M
M-V
2523
3927
1907
41−1
114
2220
2016
1016
17.C
L6-H
5758
4159
4965
1009
−26
−29
−16
0861
4037
0318
.CL6
-E−3
1−2
8−2
7−2
7−1
7−0
5−3
6−2
6−5
7−3
2−1
2−3
206
−27
−43
−25
−03
19.C
L6-X
3634
3745
4114
81−1
8−0
234
5592
3624
2517
22−2
720
.CL6
-A41
4344
6549
8102
−09
−36
−51
−15
−06
6521
5509
49−0
807
21.C
L6-C
4541
3255
4040
1550
0805
−26
1434
8423
0746
−20
2023
22.C
L6-O
2926
4027
2305
51−1
617
2632
2916
0920
8702
−31
2706
0423
.ML7
-gr
1716
1823
2201
68−3
3−1
528
6279
24−0
204
0505
−03
81−0
4−0
513
24.M
L7-s
a48
4741
4537
1853
0538
3426
4520
3347
3223
−−775
4618
3041
2025
.ML7
-et
3938
4454
3970
0702
−20
−36
−05
0948
1971
0637
−21
1679
1706
0423
26.M
L7-c
fo41
4038
5042
6206
−16
−47
−42
−06
0269
2822
−02
6620
1751
35−0
309
0632
27.M
L7-c
o24
2119
3726
1911
5209
10−2
711
1875
1002
30−0
915
0675
−04
−03
2003
2228
.ML7
-ov
3028
4426
2104
45−0
718
2719
2214
1216
8402
−29
1909
1084
0435
06−0
505
29.M
L7-n
v34
3624
5853
5610
24−0
6−1
1−2
113
4345
3500
48−1
622
3649
−03
0424
3129
34−0
41
23
45
67
89
1011
1213
1415
1617
1819
2021
2223
2425
2627
28
Not
e.Eu
gene
-Spr
ingf
ield
Com
mun
ity S
ampl
e,N
=53
3.De
cim
al p
oint
s om
itted
.Bet
wee
n-se
t cor
rela
tions
abo
ve.6
0 in
mag
nitu
de a
re p
rinte
d in
bol
dfac
e ty
pe.A
ll co
rrel
atio
ns w
ithin
set
(the
sam
e le
vel,
stru
ctur
e;an
d nu
mbe
r of f
acto
rs) a
re in
ital
ics.
For B
ig F
ive,
corr
elat
ions
are
bas
ed o
n ES
CS-1
995.
Som
e sc
ales
from
diff
eren
t set
s ha
ve o
verla
ppin
g ite
ms,
whi
ch m
ay in
flate
the
corr
elat
ions
am
ong
the
mor
e cl
osel
y re
late
d sc
ales
.*T
raits
may
be
mat
ched
aga
inst
jobs
,the
rapi
es tr
aits
may
be
mat
ched
aga
inst
jobs
,the
rapi
es
9781412946513-Ch01 1/9/08 2:58 PM Page 38
being unusually highly evaluative, and two ofthese studies (Goldberg and Somer, 2000;Saucier, 1997) included terms referring tophysical appearance. In these studies, therehas been no difficulty in replicating the one-and two-factor structures reviewed earlier.But none of these analyses has found the BigFive in a five-factor solution.
Because of the long history of Big Fivemodels and because of its long salience inlexical studies, numerous measures of thelexical Big Five have been constructed.Saucier and Goldberg (2002) provide adetailed account of the some major adjectivalBig Five marker scales in English. A shortersummary is provided here.
Goldberg (1990) originally experimentedwith bipolar and cluster scoring methods formeasuring the Big Five as found in adjec-tives. Then he settled on a standard set of 100‘unipolar’ adjectives, 20 for each factor(Goldberg, 1992). Although this marker sethas been widely used, it is now judged overlylong for many purposes. This influentialmarker set became the starting point forreduced-length marker scale sets. The firstwas the Mini-Markers (Saucier, 1994a),which included only a 40-item subset of the100, those most univocally loading on eachof the five factors; there are indications thatvalidity is comparable with that for thelonger marker set (Dwight et al., 1998). Analternative subset is the Ortho-40 (Saucier,2002), differing from the Mini-Markers inhaving lower interscale correlations. Anotherproblem with the 100 unipolar adjectives,and to some degree these reduced-lengthdescendents, was the use of many negations(un- terms) (Graziano et al., 1998). Byincluding some adjectives not contained inthe 100 unipolar set, Saucier (2002) devisedan alternative 40-adjective set (the 3M40)that had fewer negations while retaininginterscale correlations as low as those fromthe Ortho-40.
Constituent terms and psychometricindices are provided in Table 1.3 for theMini-Markers for peer-ratings as well as self-ratings. The 569 peer-ratings are averaged
ratings from three well-acquainted peersnominated by each of the 569 persons whoprovided self-ratings, who were described bythe three peers. The scales scored from peerratings sometimes have higher internal con-sistency – specifically for agreeableness,conscientiousness, and emotional stability(the Big Five factors most highly associatedwith the broader social propriety/moralityfactor). When aggregated, peer ratings havethe potential for psychometric propertiessuperior to what self-ratings can provide(Hofstee, 1994). Correlations between selfand aggregated peer ratings were 0.66, 0.45,0.50, 0.41, and 0.49, respectively, for extra-version, agreeableness, conscientiousness,emotional stability, and intellect/imagination.
Big Five scales are also available from theitems of the International Personality ItemPool (Goldberg, 1999). Goldberg used the100 markers as orienting points for selectingitems for 20-item scales (in the IPIP-100)and 10-item scales (in the IPIP-50), with aneye to maximizing internal consistency whilebalancing the number of forward- andreverse-keyed items (Saucier and Goldberg,2002). Donnellan et al. (2006) recentlydeveloped a ‘mini-IPIP’ questionnaire byshortening the IPIP-50 to only 20 items.These IPIP scales can be expected to meas-ure factors similar to the lexical ones cap-tured by the 100 markers; however, they areone step removed from the lexical studies(Goldberg, 1990, 1992) that led to the 100markers, and they do not share method variance with adjective scales. Thus, they are not lexical-factor measures by a strict criterion.
The same can be said for the NEOPersonality Inventory (Costa and McCrae,1985, 1992b), described in another chapter inthis volume, as well as its short form, theNEO Five Factor Inventory (NEO-FFI). It isworth noting, however, that the developmentof the agreeableness and conscientiousnessdomain scales for the NEO measures wasstrongly influenced by earlier lexical meas-ures of the corresponding Big Five factors(McCrae and Costa, 1985).
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 39
9781412946513-Ch01 1/9/08 2:58 PM Page 39
40 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
Tabl
e 1.
3Ps
ycho
met
ric
Indi
ces
for
Mar
ker
Scal
es f
or t
he B
ig F
ive
No.
of
Coef
ficie
nt
% o
f Var
ianc
eM
arke
r sca
leite
ms
Sam
ple
rete
st r
2-ye
arAl
pha
rM
ean
rSD
of
r1st
– 2nd
Fact
or
Mea
nSD
Skew
Min
i-m
arke
rs (
self
-rat
ings
)I -
Extr
aver
sion
8ES
CS-1
993
0.83
0.83
0.38
0.13
47 –
13
4.10
0.56
−0.0
2II-
Agr
eeab
lene
ss8
ESCS
-199
30.
700.
790.
340.
0842
– 1
25.
740.
65−0
.91
III-C
onsc
ient
ious
ness
8ES
CS-1
993
0.78
0.83
0.38
0.11
46 –
12
5.30
0.84
−0.7
0IV
-Em
otio
nal s
tabi
lity
8ES
CS-1
993
0.73
0.76
0.28
0.12
39 –
14
4.59
0.88
−0.1
4V-
Inte
llect
/imag
inat
ion
8ES
CS-1
993
0.77
0.79
0.33
0.13
42 –
17
5.17
0.82
−0.3
8M
ini-
mar
kers
(se
lf-r
atin
gs)
I -Ex
trav
ersi
on8
ESCS
-199
8b0.
860.
420.
1350
– 1
23.
470.
80−0
.36
II- A
gree
able
ness
8ES
CS-1
998b
0.81
0.36
0.11
45 –
13
4.30
0.53
−0.7
6III
-Con
scie
ntio
usne
ss8
ESCS
-199
8b0.
860.
430.
1451
– 1
14.
080.
66−0
.86
IV-E
mot
iona
l sta
bilit
y8
ESCS
-199
8b0.
800.
330.
1342
– 1
63.
670.
70−0
.38
V-In
telle
ct/im
agin
atio
n8
ESCS
-199
8b0.
830.
380.
1446
– 1
83.
850.
64−0
.49
Min
i-m
arke
rs (
peer
-rat
ings
)I -
Extr
aver
sion
8ES
CS-1
998b
0.86
0.43
0.13
51 –
12
3.70
0.66
−0.5
4II-
Agr
eeab
lene
ss8
ESCS
-199
8b0.
870.
460.
0853
– 1
24.
280.
50−0
.95
III-C
onsc
ient
ious
ness
8ES
CS-1
998b
0.89
0.49
0.13
57 –
10
4.12
0.61
−0.9
6IV
-Em
otio
nal s
tabi
lity
8ES
CS-1
998b
0.84
0.40
0.15
48 –
17
3.57
0.63
−0.2
6V-
Inte
llect
/imag
inat
ion
8ES
CS-1
998b
0.81
0.36
0.17
44 –
20
3.98
0.49
−0.4
3
Not
eES
CS –
Eug
ene-
Sprin
gfie
ld C
omm
unity
Sam
ple
(199
3,N
=11
25;1
998b
and
199
8b s
ampl
es,N
=56
9).E
SCS-
1998
b us
ed a
1-5
ratin
g sc
ale
rath
er th
an 1
–7.%
of v
aria
nce
figur
es b
ased
on
a pr
inci
pal-a
xes
anal
ysis
of a
ll ite
ms
in th
e sc
ale.
See
sour
ce a
rtic
les
(Sau
cier
(199
4a,2
002)
or a
utho
r’s w
eb-p
ages
for l
ist o
f Min
i-mar
ker i
tem
s an
d fo
r ind
ices
bas
ed o
n ot
her s
ampl
es.S
ome
com
para
ble
indi
ces
for t
he 1
00 m
arke
rs,O
rtho
-40,
and
3M40
are
ava
ilabl
e in
pre
viou
s pu
blic
atio
ns (S
auci
er,2
002;
Sauc
ier a
nd G
oldb
erg,
2002
)
9781412946513-Ch01 1/9/08 2:58 PM Page 40
Two other lexically influenced question-naires deserve mention. One is the Big FiveInventory (BFI) (Benet-Martinez and John,1998). This measure has 44 short phraseitems. The content and positions for the fivefactors on this instrument were clearly influ-enced by both the Big Five adjective scalesand by the NEO inventory. For example, inBig Five measures one factor is intellect orimagination, whereas the correspondingNEO domain is labeled openness to experi-ence. In the BFI, the corresponding scale haselements of all three kinds of content, and sooverall represents a sort of compromise.Hendriks and her colleagues developed theFive Factor Personality Inventory (FFPI)(Hendriks et al., 1999), a 100-item Big Fiveinventory using an IPIP-style item formatthat has been translated and used in numer-ous languages (Hendriks et al., 2003). TheFFPI was constructed based in large part onresults of Dutch lexical studies, especiallythe innovative study of Hofstee and De Raad(1991). The BFI and FFPI are useful Big Fivemeasures, although not lexical-factor meas-ures by a strict criterion.
Even shorter measures of the Big Fivehave begun to appear. Gosling et al. (2003)developed a ten-item Big Five measure thatshowed adequate retest reliability and ade-quate convergence both with other Big Fivemeasures and between self and observer rat-ings. Major sources for the items wereGoldberg’s marker sets and the BFI.
As these examples of Big Five measuresillustrate, measures of lexical personalityfactors tend to be provisional and are usedprimarily in research, but they have also pro-vided a useful template for the developmentof more sophisticated assessment instruments.
Lexical six-factor models
Ashton et al. (2004) have presented evidencethat many of the lexical studies conducted todate yield a consistent pattern in six-factor solu-tions. Although the structure was first detectedin studies of Korean (Hahn et al., 1999) and
French (Boies et al., 2001), it has also appearedto a recognizable degree in Dutch, German,Hungarian, Italian, and Polish. This structureseems less bound to the Germanic and Slaviclanguage families than the Big Five.
Empirically, the extraversion, conscien-tiousness, and intellect factors in this six-factor model differ relatively little fromcorresponding factors in the Big Five.Emotionality is more related to (low) emo-tional stability than to any other Big Fivefactor. The other two factors emerge largelyout of the interstitial areas between Big Fivefactors: agreeableness from big five agree-ableness and emotional stability, and hon-esty/humility from big five agreeablenessand conscientiousness. However, as Table 1.2indicates, emotionality and honesty in partic-ular tend to have relations to more than twoBig Five factors.
Evidence to date indicates that the replic-ability of the six-factor structure across lan-guages probably exceeds that for the BigFive. Moreover, this ‘Cross-Language Six’might be considered superior because it pro-vides more information than the Big Five.In the first reported ‘horse races’ betweenthe models, replication comparisons in lexi-cal study of modern Greek (Saucier et al.,2005) and of the language of the Maasai(Saucier et al., 2006), the six-factor modelseemed about equally as replicable as the BigFive. In neither study, however, were five-or six-factor models nearly as well repli-cated as were one- and two-factor models.
Other measures focused on in this chapterhave included adjectives as items, and onemight employ adjective markers to indexthese six factors. The best approach would beto utilize as many as possible of the adjec-tives that Ashton et al. present in their ‘sum-mary of the six-factor solutions’ (2004: 363)in various languages. Table 1.4 presents theconstituent terms and psychometric indicesfor a set of marker scales so constructed;large subsets of these terms (in translation)have been used as marker scales for theCross-Language Six in two previous lexicalstudies (Saucier et al., 2005, 2006).
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 41
9781412946513-Ch01 1/9/08 2:58 PM Page 41
42 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
Tabl
e 1.
4Ps
ycho
met
ric
Indi
ces
for
Adj
ecti
ve M
arke
r Sc
ales
for
the
Cro
ss-L
angu
age
Six
rwith
HEX
ACO
-PI
Mar
ker S
cale
No.
of
Coef
f. %
of V
ar.
Item
sSa
mpl
eAl
pha
Mea
n r
SD o
f r1st
– 2nd
Fac.
Mea
n SD
Sk
ewH
EX
AC
OHo
nest
y/Hu
mili
ty (H
)10
ESCS
0.71
0.24
0.12
33 –
13
5.94
0.58
−0
.69
0.40
0.10
0.02
0.25
0.22
−0.0
8Em
otio
nalit
y (E
)11
ESCS
0.65
0.14
0.15
23 –
18
3.80
0.67
0.17
0.05
0.57
−0.1
7−0
.02
−0.1
5−0
.13
Extr
aver
sion
(X)
8ES
CS0.
750.
280.
1137
– 1
54.
890.
89−0
.21
−0.0
10.
080.
700.
000.
120.
08Ag
reea
blen
ess
(A)
9ES
CS0.
760.
290.
1138
– 1
25.
250.
79−0
.54
0.27
0.02
−0
.08
0.59
0.04
0.02
Cons
cien
tious
ness
(C)
9ES
CS0.
760.
260.
1135
– 1
35.
700.
72−0
.96
0.13
0.00
0.06
−0.0
10.
69−0
.16
Ope
nnes
s (O
)11
ESCS
0.76
0.23
0.14
31 –
13
5.18
0.70
−0.4
8−0
.15
−0.1
60.
33−0
.02
0.03
0.60
Not
eES
CS –
Eug
ene–
Sprin
gfie
ld c
omm
unity
sam
ple,
N=
533,
exce
pt N
=51
9 fo
r cor
rela
tions
with
HEX
ACO
-PI.
% o
f var
ianc
e fig
ures
bas
ed o
n a
prin
cipa
l-axe
s an
alys
is o
f all
item
s in
the
scal
e.Ad
ject
ives
sel
ecte
d as
mar
kers
for C
ross
-Lan
guag
e Si
x ar
e dr
awn
from
thos
e ci
ted
as m
ost r
ecur
rent
acr
oss
lang
uage
s by
Ash
ton
et a
l.(2
004)
,and
are
as
follo
ws:
(hon
esty
/hum
ility
)ho
nest
,sin
cere
,fai
r,lo
yal,
mod
est,
vs.d
ecei
tful,
hypo
criti
cal,
conc
eite
d,sl
y,gr
eedy
;(em
otio
nalit
y) a
nxio
us,f
earfu
l,vu
lner
able
,em
otio
nal,
sens
itive
,sen
timen
tal v
s.st
rong
,cou
rage
ous,
inde
-pe
nden
t,to
ugh,
inde
pend
ent,
self-
assu
red;
(ext
rave
rsio
n) ta
lkat
ive,
soci
able
,che
erfu
l,en
erge
tic v
s.qu
iet,
shy,
pass
ive,
with
draw
n;(a
gree
able
ness
) gen
tle,t
oler
ant,
peac
eful
,agr
eeab
le,g
ood-
natu
red
vs.i
rrita
ble,
argu
men
tativ
e,ag
gres
sive
,sho
rt-t
empe
red;
(con
scie
ntio
usne
ss) o
rder
ly,pr
ecis
e,ca
refu
l,se
lf-di
scip
lined
vs.
diso
rgan
ized
,laz
y,ne
glig
ent,
reck
less
,irr
espo
nsib
le;(
open
ness
)cr
eativ
e,in
telle
ctua
l,ph
iloso
phic
al,t
alen
ted,
educ
ated
,witt
y un
conv
entio
nal v
s.un
crea
tive,
unin
telle
ctua
l,un
educ
ated
,con
vent
iona
l.Ps
ycho
met
ric in
dice
s fo
r the
HEX
ACO
Per
sona
lity
Inve
ntor
y (H
EXAC
O-P
I) ca
n be
obt
aine
d fro
m L
ee a
nd A
shto
n (2
004)
.
9781412946513-Ch01 1/9/08 2:58 PM Page 42
However, the standard way to measurethese six factors is with a questionnairecalled the HEXACO Personality Inventory(HEXACO-PI) (Lee and Ashton, 2004). Inthis inventory, each of the six factors has foursubscales measuring facets of the six factors.Psychometric indices for the six higher-order scales are presented elsewhere (Leeand Ashton, 2004). Correlations betweenHEXACO scales and adjective markers forthe Cross-Language Six are also included in Table 1.4. The correlations indicate goodlexically based content validity for theHEXACO scales, with one exception: theconvergence of lexical and questionnairehonesty/humility (r = 0.40) is rather weak.This is probably due to the H scale’s use offairly specific contextualized items. Generally,the questionnaire scale appears less suffusedwith evaluation and agreeableness than thelexical version.
This six-factor model may be found onlyin the adjective domain. Saucier (2003b)found that the structure of type-nouns inEnglish yielded six factors very similar tothose found in studies of Dutch (De Raad andHoskens, 1990) and German (Henss, 1998).However, these six factors – liveliness,antagonism, malignancy/cowardice, mas-culinity, intellect/openness, and attractive-ness – as a set do not correspond closely tothe Cross-Language Six described here.
Seven-factor models found with awider inclusion of lexical variables
Analyses leading to the five- or six-factorstructure have involved, in effect, removal ofthe most extremely evaluative terms at anearly stage of the variable-selection process.Indeed, Allport (Allport and Odbert, 1936)and Norman (1963) both favored removal ofhighly evaluative terms. Also removed havebeen (a) terms indicating relative eccentricity(e.g. average, strange, unusual); (b) termsthat can refer to both stable and temporaryattributes (e.g. happy, tired, bored); (c) ten-dencies to affect others in a consistent way
(e.g. likeable, annoying); (d) social statusindicators (e.g. wealthy, famous); and (e)attributes of physique and health (e.g. tall,fat, sickly). When investigators have usedwider variable selections (i.e. those includingmany or all of these excluded types of vari-ables), the Big Five has not appeared readilyin five-factor solutions. Studies in Englishand Turkish, however, did find Big-Five-like factors within a seven-factor solution(Goldberg and Somer, 2000; Saucier, 1997;Tellegen and Waller, 1987).
Of the two additional factors, one wasfound in all three studies: ‘negative valence’is a factor emphasizing attributes withextremely low desirability and endorsementrates and with descriptive content involvingmorality/depravity, dangerousness, worth-lessness, peculiarity, and stupidity (cf. Benet-Martinez and Waller, 2002). Its overallthemes – extreme social impropriety, failinga threshold for social acceptability, and notbeing worthy of trust or credence – involvenon-normativeness: Does one stand outsideof social norms to a high enough degree thatone becomes liable for exclusion from thegroup? The other factor varied more acrossthe three studies and involved descriptorsindicating some kind of power to impressothers, either in the form of a ‘positive valence’factor emphasizing positive attributes (possi-bly social effects) like ‘impressive’ and ‘out-standing’ (found by Tellegen and Waller,1987) or, where attractiveness terms wereincluded, an ‘attractiveness’ factor (found byGoldberg and Somer, 2000; Saucier, 1997;also Saucier, 2003b). Whether attractivenessor negative valence fall within the domain ofpersonality can be debated, but both factorsinvolve phenomena of great interest to socialpsychologists.
The structure labeled the Big Seven wasestablished in an unpublished lexical study ofEnglish descriptors that used the method ofsampling one descriptors from one in everyfour pages of a dictionary (Tellegen and Waller,1987). The structure includes five Big-Five-like factors, except that ‘intellect/ imagina-tion’ is reconceived as ‘unconventionality’.
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 43
9781412946513-Ch01 1/9/08 2:58 PM Page 43
The two additional factors are labeled ‘posi-tive valence’ and ‘negative valence’. Scrutinyof empirical indices of replication indicatethat two attempts to replicate this structure,in studies of Spanish (Benet-Martinez andWaller, 1997) and Hebrew (Almagor et al.,1995) had only modest success, althougheach of these studies found structures of inter-est in their own right. A standard lexicalmeasure of the Big Seven in its originalEnglish-study structural form is the Inventoryof Personal Characteristics (IPC-7) (Tellegenet al., 1991).
Studies in some other languages withbroad variable-selection criteria indicate analternative seven-factor structure. The con-vergences between these studies occurred inspite of their many differences in methodol-ogy. Lexical studies in Filipino (Church et al., 1997, 1998) and Hebrew (Almagor et al., 1995) – languages from unrelated language-families and cultures – yielded ahighly convergent seven-factor structure,even though this similarity was obscured bydiscrepant labels. The English translations ofmarker adjectives for the Filipino andHebrew factors have been shown to corre-spond in a one-to-one way (Saucier, 2003a).
One of these new factors resembles thenegative valence factor just described. Twoof them resemble Big Five factors – consci-entiousness and intellect. The other three BigFive factors – extraversion, agreeableness,and emotional stability – correlate substan-tially with the remaining four factors, whichmap an affective-interpersonal domain (cf.Saucier, 1992). These four can be labeled‘gregariousness’ (or ‘liveliness’), ‘self-assur-ance’ (or ‘fortitude’), ‘even temper’ (‘toler-ant’ versus ‘temperamental’), and ‘concernfor others’ (versus ‘egotism’). Big Five extra-version is related to gregariousness and self-assurance, emotional stability to self-assuranceand even temper, and agreeableness to eventemper and concern for others.
The relation of the Multi-Language Seven(ML7) to the Cross-Language Six (CL6) isbest explained with the help of Figure 1.1. Thisfigure shows the relations between lexical
structures of one, two, five, six, and sevenfactors. It joins any factors at adjacent levelsthat have a substantial correlation (more than0.35 in magnitude) in Table 1.2. The SDQscale was used for the one-factor level. Pairsof factors correlated the most highly (above0.70) are joined by a thick and bold line. Thefigure depicts very strong relations betweenCL6 extraversion and ML7 gregariousness,between CL6 agreeableness and ML7 eventemper, between the conscientiousness fac-tors, and between CL6 openness and ML7originality/virtuosity. ML7 self-assurance isrelated to both CL6 emotionality (reverse-scored) and extraversion, whereas ML7 concern for others and negative valence(reverse-scored) are both related both to CL6 agreeableness and honesty/humility,negative valence being related also to CL6conscientiousness.
It is noteworthy that negative valence issubstantially correlated with CL6 honesty(H), conscientiousness (C), and agreeable-ness (A) (−0.36 to −0.49 with each), anotherindicator that this factor contains descriptivecontent. Unlike extraversion, emotionality,and openness, these three factors (H, C, andA) concern moral and prosocial behavior,and are clearly related to the broad socialpropriety and morality factor (and not todynamism). The aspect of social proprietyand morality uniquely captured by so-callednegative valence is normality violation; thatis, the tendency to behave in ways that areawry, askew, and violative of normal stan-dards for behavior, by way of undependabil-ity, recklessness, abusiveness, incompetence,or sheer eccentricity. The favorable pole ofthis dimension is characterized by ‘vanilla’descriptors like normal and trustworthy(Saucier, 2003a). The unfavorable pole isparticularly richly represented (in English)by type-nouns, like creep, idiot, fool, twit,crook, and deadbeat, terms whose useimplies that the target is being singled out forsocial exclusion (Saucier, 2003b). The con-tempt implied in these descriptors may not beunusual when we encounter others who vio-late the standards of what we consider normal.
44 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 44
Saucier (2003a) developed a 60-adjectivemarker set for the seven factors. Constituentterms and psychometric indices are providedin Table 1.5.
An integrative framework forstructures of one to seven lexicalfactors
How are the structures (and measures) ofone, two, three, five, six, and seven lexicalfactors related to one another? Some answersmight be found by examining the intercorre-lations of the scales measuring their factors(Table 1.2). This table leads to a clear pictureof the relation of the one- and two-factorstructures with each other and with the five-,six-, and seven-factor structures. The generalevaluation factor bifurcates into S and D.Social propriety divides into agreeableness,conscientiousness, and emotional stability(Big Five), or into honesty, agreeableness,and conscientiousness (Cross-LanguageSix), or into even temper, concern for others,conscientiousness, and (reversed) normalityviolation (ML7). Dynamism divides intoextraversion and intellect/imagination (BigFive), into extraversion, openness, and (low)emotionality (CL6), or into gregariousness,self-assurance, and originality/virtuosity(ML7). Thus, the Big Two is a sensiblehigher-order organization for each of thesethree structures. However, it is far more diffi-cult to give a simple description of how thefive-, six-, and seven-factor structures relateto each other. Indicating complexity, inFigure 2.1 the lines joining levels of five, six, and seven factors have several crossinglines.
A hierarchical structural representationcombining both broader and narrower con-structs will provide the best compromisebetween parsimony and accuracy. The broadlevels, with wider bandwidth constructs,offers higher efficiency (i.e. parsimony). Thenarrower levels offer higher fidelity (i.e. pre-dictive accuracy). Given the differences incovariation structure between languages, it
seems appropriate to defer such studies oflexically derived facets until a consensualhierarchical structure at the broad levels isbetter defined.
Belief, value, and attitude factors asadditions to the dimensionalclassification
As the above review indicates, the Big Fiveand Cross-Language Six are structureswhose appearance seems contingent on a rel-atively narrow selection of variables, andthus on an operational definition of personal-ity that has many exclusion clauses. TheMulti-Language Seven may be contingent ona more inclusive variable selection and defi-nition of personality. Structures with one ortwo broad factors seem less dependent on thevariable selection on the definition of person-ality. However, none of the previous lexicalstudies of personality has included a substan-tial representation of belief, value, and attitudevariables. Would including such variables leadto additional factors?
Unfortunately, the research literature onthe structure of beliefs, values, and attitudeshas been poorly developed. To rectify this sit-uation, Saucier (2000) used a lexical ration-ale, extracting from a large dictionary allEnglish nouns ending in ‘-ism,’ such termspostulated to represent many of the mostimportant beliefs and attitudes. From the def-initions of these terms, Saucier developed389 questionnaire items, which he adminis-tered to a large sample of college students,who indicated their extent of agreement witheach item. Analyses revealed four broaddimensions of beliefs and attitudes, whichwere replicated in a follow-up sample, andlater in Romanian (Krauss, 2006). The studyyielded important increments to knowledgeabout belief/attitude dispositions: Two of thefour factors are little represented in previousmeasures.
Saucier’s (2000) four ‘isms’ factors arelabeled as traditional religiousness (α), sub-jective spirituality (δ), unmitigated self-interest
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 45
9781412946513-Ch01 1/9/08 2:58 PM Page 45
46 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
Tabl
e 1.
5Ps
ycho
met
ric
Indi
ces
for
Adj
ecti
ve M
arke
r Sc
ales
for
the
Mul
ti-L
angu
age
Seve
nCo
effic
ient
%
of V
aria
nce
Mar
ker S
cale
No.
of I
tem
s Sa
mpl
eAl
pha
Mea
nr
SD o
fr1st
– 2nd
Fact
orM
ean
SDSk
ewG
rega
rious
ness
(GR)
8ES
CS0.
780.
300.
1340
– 1
44.
070.
960.
06Se
lf-As
sura
nce
(SA)
9ES
CS0.
740.
250.
1134
– 1
35.
210.
78−0
.63
Even
Tem
per (
ET)
8ES
CS0.
810.
350.
1545
– 1
24.
601.
06−0
.29
Conc
ern
for O
ther
s (C
FO)
8ES
CS0.
700.
250.
1235
– 1
75.
460.
71−0
.41
Cons
cien
tious
ness
(CO
)9
ESCS
0.72
0.23
0.16
35 –
13
4.84
0.81
−0.4
4O
rigin
ality
/Virt
uosi
ty (O
V)9
ESCS
0.73
0.25
0.12
34 –
14
4.98
0.79
−0.3
2N
egat
ive
Vale
nce
or
Nor
mal
ity V
iola
tion
(NV)
9ES
CS0.
760.
270.
1036
– 1
56.
390.
64−1
.72
Not
e.N
= 5
92.%
of v
aria
nce
figur
es b
ased
on
a pr
inci
pal-a
xes
anal
ysis
of a
ll ite
ms
in th
e sc
ale
9781412946513-Ch01 1/9/08 2:58 PM Page 46
(β), and protection of civil institutions (γ).These four dimensions showed low correla-tions with markers for the Big Five (Saucier,2000), indicating that adding the stable dispositions underlying beliefs and attitudesin a dimensional classification of personal-ity will result in at least four additional factors. Contemporary personality invento-ries, however, include scales for constructslike self-transcendence, traditionalism, andopenness to experience, which are morehighly related to belief, value, and attitudedispositions.
Saucier (2006) has developed a brief set ofmarker items for these four factors, the itemsbeing based mostly on dictionary definitionsfrom the earlier study (Saucier, 2000). Table 1.6provides psychometric indices for thesescales. In addition to reasonable internal con-sistency (a from 0.69 to 0.79), the scalesshow impressive retest stability (r from 0.64to 0.85) across nearly four years, and reteststability for tradition-oriented religiousness(0.85) well exceeds that for the typical per-sonality measure. Table 1.7 provides the cor-relations between the four isms scales andlexical marker scales reviewed earlier in this chapter. The highest r is 0.31 (betweensubjective spirituality and loose vs. tight)and, consistent with these being factors addi-tional to those in the lexical marker scales,there are few correlations above 0.20 in magnitude.
As for values, Renner (2003a, 2003b)developed a questionnaire from a lexicalstudy of German, and found four factors incommon across adjective and noun variableselections. These factors were labeled salva-tion, profit, intellectualism, and balance; thefirst two may correspond to tradition-ori-ented religiousness and unmitigated self-interest from Saucier (2000). The sameauthor completed a similar project using theNorthern Sotho language from South Africa(Renner et al., 2003). Again, factors relatedto salvation and profit (though differentlylabeled), plus three additional factors, werefound. Renner’s program of studies makesclear that values can be studied by the lexical
approach, and that lexical value factors prob-ably have some relation to lexical isms fac-tors. These studies promise to lead eventuallyto lexically based measures of values.Analogous lexical studies of interests wouldbe very useful.
CONCLUSIONS
Lexical studies of personality attributes nar-rowly defined have now reached a stagemature enough that key aspects of their struc-ture are becoming evident in the recurrentfindings from these studies. However, the‘personality’ represented in most of thesestudies is a considerably narrower phenome-non than personality as it is typically defined,and the structure of personality attributesencoded in lexicons depends in major waysupon the upstream selection of variables.Therefore, personality psychology shouldcouple the focus that it already has, on themost prototypical attributes of personality,with a simultaneous ‘bigger picture’ exami-nation of all psychological attributes onwhich there are stable individual differences.
Recurrent aspects of the factors at the topof the personality-attribute hierarchy – theone- and two-factor levels – are already quiteclear. Beneath this top level, findings seemmore dependent on variable selection. Givenrelatively narrow variable selection proce-dures, the Big Five emerges readily from somelanguages (mainly those having origins innorthern Europe) while the Cross-LanguageSix emerges readily from an apparently evenwider range of languages. Given more inclu-sive procedures, studies to date are too few topermit firm conclusions. More studies areneeded, and the direct measures of lexicon-derived personality factors reviewed in thischapter are a vital tool for these studies.These measures help facilitate the search forwhat is recurrent and ubiquitous (and what is not) in the personality tendencies that dif-ferentiate humans, as sedimented in humanlexicons.
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 47
9781412946513-Ch01 1/9/08 2:58 PM Page 47
48 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
Tabl
e 1.
6Ps
ycho
met
ric
Indi
ces
for
Dic
tion
ary-
Base
d M
arke
r Sc
ales
for
the
Fou
r Is
ms
Fact
ors
No.
of
4-ye
ar
Coef
ficie
nt
Mea
n In
ter-
SD o
f Int
er-
% o
f Var
ianc
eM
arke
r Sca
leIte
ms
Sam
ple
Rete
st r
Alph
aite
m r
item
r 1st
– 2nd
Fact
orM
ean
SDSk
ewTr
aditi
on-O
rient
ed R
elig
ious
ness
(TR;
a)6
ESCS
0.85
0.79
0.39
0.11
50 –
14
2.94
0.90
−0.1
0U
nmiti
gate
d Se
lf-In
tere
st (U
SI;fl
)8
ESCS
0.65
0.70
0.23
0.08
33 –
15
1.72
0.50
0.77
Prot
ectio
n of
Civ
il In
stitu
tions
(PCI
;?)
8ES
CS0.
640.
690.
220.
1033
– 1
64.
280.
46−1
.05
Subj
ectiv
e Sp
iritu
ality
(SS;
d)6
ESCS
0.77
0.75
0.33
0.08
45 –
15
2.94
0.81
−0.1
5
Not
eN
= 7
03,e
xcep
t ret
est r
,bas
ed o
n N
= 6
52.%
of v
aria
nce
figur
es b
ased
on
a pr
inci
pal-a
xes
anal
ysis
of a
ll ite
ms
in th
e sc
ale
9781412946513-Ch01 1/9/08 2:58 PM Page 48
ACKNOWLEDGMENTS
Work on this article was supported by GrantMH-49227 from the National Institute ofMental Health, US Public Health Service. I am grateful to Lewis R. Goldberg for edit-ing suggestions. Correspondence regardingthis article may be addressed to: GerardSaucier, Department of Psychology, 1227University of Oregon, Eugene, OR 97403,USA (E-mail: [email protected]
REFERENCES
Ackerman, P.L. (1996) ‘A theory of adult intel-lectual development: Process, personality,
interests, and knowledge’, Intelligence, 22:227–57.
Ackerman, P.L. and Heggestad, E.D. (1997)‘Intelligence, personality, and interests:Evidence for overlapping traits’, PsychologicalBulletin, 121: 218–45.
Allport, G.W. (1937) Personality: A Psycholo-gical Interpretation. New York: Holt.
Allport, G.W. and Odbert, H.S. (1936) ‘Traitnames: A psycho-lexical study’, PsychologicalMonographs, 4(1, Whole No. 211):
Almagor, M., Tellegen, A. and Waller, N. (1995)‘The Big Seven model: A cross-cultural repli-cation and further exploration of the basicdimensions of natural language of traitdescriptions’, Journal of Personality andSocial Psychology, 69: 300–7.
Alvarez, J.M., Ruble, D.N. and Bolger, N. (2001)‘Trait understanding or evaluative reasoning?
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 49
Table 1.7 Correlations Between Isms Factor Scales and Adjective Marker Scales ForStructures of One to Seven Factors Isms Factor ScaleAdjective Scale TR (α) USI (β) PCI (γ ) SS (δ)Ge-24 −0.09 −0.11 0.19 0.07Ge-12 −0.08 −0.10 0.18 0.08NEPC-E −0.08 −0.17 0.18 0.10SDQ −0.01 −0.07 0.25* 0.00Osgood E 0.02 −0.09 0.18 0.09S 0.05 −0.12 0.22* 0.03D −0.10 −0.02 0.07 0.08TL 0.00 −0.03 0.15 −0.31*TALU −0.17 0.10 −0.04 −0.18AU −0.16 0.10 0.01 −0.07ALUT −0.09 0.06 −0.17 0.19B5MM-I −0.03 −0.03 0.08 0.04B5MM-II 0.07 −0.14 0.22* 0.12B5MM-III 0.04 −0.02 0.22* −0.13B5MM-IV −0.07 −0.07 0.20* −0.11B5MM-V −0.15 −0.11 −0.07 0.07CL6-H 0.14 −0.16 0.24* −0.03CL6-E 0.18 −0.02 −0.02 0.11CL6-X −0.02 −0.02 0.14 0.06CL6-A 0.02 −0.05 0.20* 0.06CL6-C 0.07 −0.08 0.27* −0.14CL6-O −0.20* −0.09 −0.08 0.10ML7-gr 0.00 −0.02 0.07 0.11ML7-sa −0.10 0.02 0.11 −0.03ML7-et 0.03 −0.06 0.13 −0.04ML7-cfo 0.23* −0.07 0.16 0.16ML7-co 0.24* −0.05 0.27* −0.14ML7-ov −0.21* −0.10 −0.03 0.09ML7-nv −0.02 0.08 −0.29* 0.22*
Note Eugene–Springfield Community Sample, N = 521. All correlations.09 and above are significant, p < .05. For Big Five,correlations are based on ESCS-1995, which is closer in time to the administration of the isms measures. * correlationsover.20 in magnitude. Highest correlation in each column is underlined. TR – tradition-oriented religiousness,USI – unmitigated self-interest, PCI – protection of civil Institutions, SS – subjective spirituality.
9781412946513-Ch01 1/9/08 2:58 PM Page 49
An analysis of children’s behavioral predic-tions’, Child Development, 72: 1409–25.
Ashton, M.C., Lee, K., Perugini, M., Szarota, P.De Vries, R.E., Di Blas, L., Boies, K. and DeRaad, B. (2004) ‘A six-factor structure of personality-descriptive adjectives: Solutionsfrom psycholexical studies in seven lan-guages’, Journal of Personality and SocialPsychology, 86: 356–66.
Benet-Martínez, V. and John, O.P. (1998) ‘LosCinco Grandes across cultures and ethnicgroups: Multitrait multimethod analyses ofthe Big Five in Spanish and English’, Journalof Personality and Social Psychology, 75:729–50.
Benet-Martinez, V. and Waller, N.G. (1997)‘Further evidence for the cross-cultural gen-erality of the Big Seven factor model:Indigenous and imported Spanish personalityconstructs’, Journal of Personality, 65: 567–98.
Benet-Martinez, V. and Waller, N.G. (2002)‘From adorable to worthless: Implicit andself-report structure of highly evaluative per-sonality descriptors’, European Journal ofPersonality, 16: 1–41.
Boies, K., Lee, K., Ashton, M.C., Pascal, S. andNicol, A.A.M. (2001) ‘The structure of theFrench personality lexicon’, European Journalof Personality, 15: 277–95.
Bouchard, T.J. (1994) ‘Genes, environment,and personality’, Science, 264: 1700–1.
Caprara, G.V., Barbanelli, C. and Zimbardo,P.G. (1997) ‘Politicians’ uniquely simple per-sonalities’, Nature, 385: 493.
Carroll, J.B. (1993) Human Cognitive Abilities:A Survey of Factor-analytic Studies. NewYork: Cambridge University Press.
Cattell, R.B. (1943) ‘The description of person-ality: Basic traits resolved into clusters’,Journal of Abnormal and Social Psychology,38: 476–506.
Church, A.T., Katigbak, M.S. and Reyes, J.A.S.(1998) ‘Further exploration of Filipino per-sonality structure using the lexical approach:Do the Big Five or Big Seven dimensionsemerge?’, European Journal of Personality,12: 249–69.
Church, A.T., Reyes, J.A.S., Katigbak, M.S. andGrimm, S.D. (1997) ‘Filipino personalitystructure and the Big Five model: A lexicalapproach’, Journal of Personality, 65: 477–528.
Clark, L.A. and Watson, D. (1999) ‘Tempera-ment: A new paradigm for trait psychology’,
in L.A. Pervin and O.P. John (eds), Handbookof Personality: Theory and Research (2nd ed.,pp. 399-423). New York: Guilford Press.
Cloninger, C.R., Bayon, C. and Svrakic, D.M.(1998) ‘Measures of temperament and char-acter in mood disorders: A model of funda-mental states as personality types’, Journalof Affective Disorders, 51: 21–32.
Costa, P.T. and McCrae, R.R. (1985) The NEOPersonality Inventory Manual. Odessa, FL:Psychological Assessment Resources.
Costa, P.T. and McCrae, R.R. (1992a) ‘Fourways five factors are basic’, Personality andIndividual Differences, 13: 653–5.
Costa, P.T. and McCrae, R.R. (1992b) RevisedNEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO-FFI)Professional Manual. Odessa, FL: PsychologicalAssessment Resources.
De Raad, B. and Barelds, D. (2006) ‘A new tax-onomy of Dutch personality traits based on acomprehensive and unrestricted list of descrip-tors’, unpublished manuscript, University ofGroningen.
De Raad, B., Di Blas, L. and Perugini, M. (1998)‘Two independent Italian trait taxonomies:Comparisons with Italian and betweenItalian Germanic languages’, EuropeanJournal of Personality, 12: 19–41.
De Raad, B. and Hoskens, M. (1990) ‘Personality-descriptive nouns’, European Journal ofPersonality, 4: 131–46.
De Raad, B. and Peabody, D. (2005) ‘Cross-cul-turally recurrent personality factors: Analysesof three factors’, European Journal ofPersonality, 19: 451–74.
DeYoung, C.G. (2006) ‘Higher-order factors ofthe Big Five in a multi-informant sample’,Journal of Personality and Social Psychology.
Di Blas, L. and Forzi, M. (1998) ‘An alternativetaxonomic study of personality descriptors inthe Italian language’, European Journal ofPersonality, 12: 75–101.
Di Blas, L. and Forzi, M. (1999) ‘Refining a descriptive structure of personality attrib-utes in the Italian language: The abridgedBig Three circumplex structure’, Journal of Personality and Social Psychology, 76:451–81.
Di Blas, L., Forzi, M. and Peabody, D. (2000)‘Evaluative and descriptive dimensions fromItalian personality factors’, European Journalof Personality, 14: 279–90.
50 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 50
Digman, J.M. (1997) ‘Higher order factors ofthe Big Five’, Journal of Personality andSocial Psychology, 73: 1246–56.
Donnellan, M.B., Oswald, F.L., Baird, B.M. andLucas, R.E. (2006) ‘The Mini-IPIP Scales: Tinyyet effective measures of the Big Five factorsof personality’, Psychological Assessment,18: 192–203.
Dwight, S.A., Cummings, K.M. and Glenar, J.L.(1998) ‘Comparison of criterion-relatedvalidity coefficients for the Mini-Markers andfor Goldberg’s markers of the Big Five per-sonality factors’, Journal of PersonalityAssessment, 70: 541–50.
Eysenck, H.J. (1991) ‘Dimensions of personal-ity: 16, 5, or 3? Criteria for a taxonomic par-adigm’, Personality and Individual Differences,12: 773–90.
Eysenck, H.J. and Eysenck, S.B.G. (1975)Manual of the Eysenck PersonalityQuestionnaire. London: Hodder & Stoughton.
Funder, D.C. (1997) The Personality Puzzle.New York: Norton.
Goldberg, L.R. (1981) ‘Language and individualdifferences: The search for universals in per-sonality lexicons’, in L.W. Wheeler (ed),Review of Personality and Social Psychology(Vol. 2) Beverly Hills, CA: Sage. pp. 141–65.
Goldberg, L.R. (1982) ‘From Ace to Zombie:Some explorations in the language of per-sonality’, in C.D. Spielberger and J.N. Butcher(eds), Advances in Personality Assessment(Vol. 1). Hillsdale, NJ: Erlbaum. pp. 203–34.
Goldberg, L.R. (1990) ‘An alternative “descrip-tion of personality”: The Big-Five factorstructure’, Journal of Personality and SocialPsychology, 59: 1216–29.
Goldberg, L.R. (1992) ‘The development ofmarkers for the Big-Five factor structure’,Psychological Assessment, 4: 26–42.
Goldberg, L.R. (1993) ‘The structure of pheno-typic personality traits’, AmericanPsychologist, 48: 26–34.
Goldberg, L.R. (1999) ‘A broad-bandwidth,public-domain, personality inventory meas-uring the lower-level facets of several five-factor models’, in I. Mervielde, I. Deary, F. DeFruyt and F. Ostendorf (eds), PersonalityPsychology in Europe (Vol. 7). Tilburg, TheNetherlands: Tilburg University Press. pp. 7–28.
Goldberg, L.R. and Saucier, G. (1995) ‘So whatdo you propose we use instead? A reply toBlock’, Psychological Bulletin, 117: 221–5.
Goldberg, L.R. and Somer, O. (2000) ‘The hier-archical structure of common Turkish person-descriptive adjectives’, European Journal ofPersonality, 14: 497–531.
Goldberg, L.R. and Velicer, W.F. (2006)‘Principles of exploratory factor analysis’, inS. Strack (ed), Differentiating Normal andAbnormal Personality (second edition). NewYork: Springer. pp. 209–37.
Gosling, S.D., Rentfrow, P.J. and Swann, W.B.(2003) ‘A very brief measure of the Big Fivepersonality domains’, Journal of Research inPersonality, 37: 504–28.
Graziano, W.G., Jensen-Campbell, L.A., Steele,R.G. and Hair, E.C. (1998) ‘Unknown wordsin self-reported personality: Lethargic andprovincial in Texas’, Personality and SocialPsychology Bulletin, 24: 893–905.
Hahn D.W., Lee, K. and Ashton, M.C. (1999) ‘A factor analysis of the most fre-quently used Korean personality trait adjec-tives’, European Journal of Personality, 13:261–82.
Hendriks, A.A.J., Hofstee, W.K.B. and De Raad,B. (1999) ‘The Five-Factor PersonalityInventory (FFPI)’, Personality and IndividualDifferences, 27: 307–25.
Hendriks, A.A.J., Perugini, M., Angleitner, A.,Ostendorf, F., Johnson, J.A., De Fruyt, F.,Hrebícková, M., Kreitler, S., Murakami, T.,Bratko, D., Conner, M., Nagy, J., Rodriguez-Fornells, A. and Ruisel, I. (2003) ‘The Five-Factor Personality Inventory: Cross-culturalgeneralizability across 13 countries’,European Journal of Personality, 17: 347–73.
Henss, R. (1998) ‘Type nouns and the fivefactor model of personality description’,European Journal of Personality, 12: 57–71.
Hofstee, W.K.B. (1994) ‘Who should own thedefinition of personality?’, European Journalof Personality, 8: 149–62.
Hofstee, W.K.B. and De Raad, B. (1991)‘Persoonlijkheidsstructuur: de AB-sub-5C-taxonomie van Nederlandse eigenschapster-men [Personality structure: The Abridged BigFive-Dimensional Circumplex (AB5C) struc-ture of Dutch trait adjectives’, NederlandsTijdschrift voor de Psychologie en haarGrensgebieden, 46: 262–74.
Hogan, R. (1983) ‘A socioanalytic theory ofpersonality’, in M.M. Page (ed), NebraskaSymposium on Motivation. Lincoln: Universityof Nebraska Press. pp. 336–55.
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 51
9781412946513-Ch01 1/9/08 2:58 PM Page 51
Hrebícková, M., Ostendorf, F., Osecká, L. andCermák, I. (1999) ‘Taxonomy and structure ofCzech personality-relevant verbs’, in I.Mervielde, I.J. Deary, F. De Fruyt and F.Ostendorf (eds), Personality Psychology inEurope (Vol. 7). Tilburg, The Netherlands:Tilburg University Press. pp. 51–65.
John, O.P. (1990) ‘The “Big Five” factor taxon-omy: Dimensions of personality in the natu-ral language and in questionnaires’ in L.A. Pervin (ed), Handbook of Personality:Theory and Research. New York: Guilford.pp. 66–100.
John, O.P., Angleitner, A. and Ostendorf, F.(1988) ‘The lexical approach to personality: Ahistorical review of trait taxonomic research’,European Journal of Personality, 2: 171–203.
Krauss, S. (2006) ‘Does ideology transcend cul-ture? A preliminary examination in Romania’,Journal of Personality, 74: 1219–56.
Krueger, R.F. and Markon, K.E. (2006)‘Reinterpreting comorbidity: A model-basedapproach to understanding and classifyingpsychopathology’, Annual Review of ClinicalPsychology, 2: 111–33.
Lee, K. and Ashton, M.C. (2004) ‘Psychometricproperties of the HEXACO PersonalityInventory’, Multivariate Behavioral Research,39: 329–58.
Low, K.S.D., Yoon, M., Roberts, B.W., Rounds,J. (2005) ‘Stability of vocational interestsfrom early adolescence to middle adulthood:A quantitative review of longitudinal stud-ies’, Psychological Bulletin, 131: 713–37.
May, M.A. (1932) ‘The foundations of person-ality’, in P.S. Achilles (ed), Psychology atWork. New York: McGraw-Hill.
McCrae, R.R. and Costa, P.T. (1985) ‘UpdatingNorman’s “adequate taxonomy”: Intelligenceand personality dimensions in natural lan-guage and in questionnaires’, Journal ofPersonality and Social Psychology, 49: 710–21.
McCrae, R.R. and Costa, P.T. (1997) ‘Personalitytrait structure as a human universal’,American Psychologist, 52: 509–16.
Norman, W.T. (1963) ‘Toward an adequate tax-onomy of personality attributes: Replicatedfactor structure in peer nomination personal-ity ratings’, Journal of Abnormal and SocialPsychology, 66: 574–83.
Norman, W.T. (1967) 2800 Personality TraitDescriptors: Normative Operating Charact-eristics for a University Population. Department
of Psychology, University of Michigan, AnnArbor.
Osgood, C.E. (1962) ‘Studies on the generalityof affective meaning systems’, AmericanPsychologist, 17: 10–28.
Osgood, C.E., May, W. and Miron, M. (1975)Cross-cultural Universals of AffectiveMeaning. Urbana: University of Illinois Press.
Paulhus, D.L. and John, O.P. (1998) ‘Egoisticand moralistic biases in self-perception: Theinterplay of self-descriptive styles with basictraits and motives’, Journal of Personality,66: 1025–60.
Peabody, D. (1987) ‘Selecting representativetrait adjectives’, Journal of Personality andSocial Psychology, 52: 59–71.
Peabody, D. and Goldberg, L.R. (1989) ‘Somedeterminants of factor structures from personality-trait descriptors’, Journal ofPersonality and Social Psychology, 57: 552–67.
Renner, W. (2003a) ‘A German value question-naire developed on a lexical basis:Construction and steps toward validation’,Review of Psychology, 10: 107–23.
Renner, W. (2003b) ‘Human values: A lexicalperspective’, Personality and IndividualDifferences, 34: 127–41.
Renner, W., Peltzer, K. and Phoswana, M.G.(2003) ‘The structure of values amongNorthern Sotho speaking people in SouthAfrica’, South African Journal of Psychology,33: 103–8.
Roback, A.A. (1931) Personality: The Crux ofSocial Intercourse. Cambridge, MA: Sci-Art.
Rokeach, M. (1968) Beliefs, Attitudes, andValues: A Theory of Organization andChange. San Francisco: Jossey-Bass.
Rolland, J.P., Parker, W.D. and Stumpf, H.(1998) ‘A psychometric examination of theFrench translations of the NEO-PI-R andNEO-FFI’, Journal of Personality Assessment,71: 269–91.
Rossier, J., Dahouru, D. and McCrae, R.R.(2005) ‘Structural and mean-level analyses ofthe five-factor model and locus of control:Further evidence from Africa’, Journal ofCross-Cultural Psychology, 36: 227–46.
Rothbart, M.K. and Bates, J.E. (1998)‘Temperament’, in W. Damon (series ed.) andN. Eisenberg (vol. ed.), Handbook of ChildPsychology: Vol. 3, Social, Emotional andPersonality Development (5th edition). NewYork: Wiley. pp. 105–76.
52 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT
9781412946513-Ch01 1/9/08 2:58 PM Page 52
Saucier, G. (1992) ‘Benchmarks: Integratingaffective and interpersonal circles with the Big-Five personality factors’, Journal of Personalityand Social Psychology, 62: 1025–35.
Saucier, G. (1994a) ‘Mini-markers: A brief ver-sion of Goldberg’s unipolar Big-Five mark-ers’, Journal of Personality Assessment, 63:506–16.
Saucier, G. (1994b) ‘Separating description andevaluation in the structure of personalityattributes’, Journal of Personality and SocialPsychology, 66: 141–54.
Saucier, G. (1997) ‘Effects of variable selectionon the factor structure of person descrip-tors’, Journal of Personality and SocialPsychology, 73: 1296–312.
Saucier, G. (2000) ‘Isms and the structure ofsocial attitudes’, Journal of Personality andSocial Psychology, 78: 366–85.
Saucier, G. (2002) ‘Orthogonal markers fororthogonal factors: The case of the Big Five’,Journal of Research in Personality, 36: 1–31.
Saucier, G. (2003a) ‘An alternative multi-lan-guage structure of personality attributes’,European Journal of Personality, 17:179–205.
Saucier, G. (2003b) ‘Factor structure of English-language personality type-nouns’, Journal ofPersonality and Social Psychology, 85:695–708.
Saucier, G. (2005) ‘Framework for integratinglexical structures of five, six, and seven fac-tors’, ORI Technical Report, 45(3). Eugene:Oregon Research Institute.
Saucier, G. (2006) ‘A brief measure of the fourfactors in the Survey of Dictionary-basedIsms (SDI)’, unpublished manuscript,University of Oregon.
Saucier, G., Georgiades, S., Tsaousis, I. andGoldberg, L.R. (2005) ‘The factor structureof Greek personality adjectives’, Journal ofPersonality and Social Psychology, 88: 856–75.
Saucier, G. and Goldberg, L.R. (2002)‘Assessing the Big Five: Applications of 10psychometric criteria to the development ofmarker scales’, in B. De Raad, and M.
Perugini (eds), Big Five Assessment. Seattle:Hogrefe & Huber. pp. 29–58.
Saucier, G., Ole-Kotikash, L. and Payne, D.L.(2006) ‘The structure of personality andcharacter attributes in the language of theMaasai’, unpublished report, University ofOregon.
Saucier, G., Ostendorf, F. and Peabody, D.(2001) ‘The non-evaluative circumplex ofpersonality adjectives’, Journal of Personality,69: 537–82.
Saucier, G. and Simonds, J. (2006) ‘The struc-ture of personality and temperament’, inD.K. Mroczek and T.D. Little (eds), Handbookof Personality Development. Mahwah, NJ:Erlbaum. pp. 109–28.
Shweder, R.A. (1972) ‘Semantic structure andpersonality assessment’, unpublished doc-toral dissertation, Harvard University.
Szirmák, Z. and De Raad, B. (1994) ‘Taxonomyand structure of Hungarian personalitytraits’, European Journal of Personality, 8:95–118.
Tellegen, A. (1993) ‘Folk concepts and psycho-logical concepts of personality and personal-ity disorder’, Psychological Inquiry, 4: 122–30.
Tellegen, A. (in press) Manual for theMultidimensional Personality Questionnaire.Minneapolis: University of Minnesota Press.
Tellegen, A., Grove, W. and Waller, N.G. (1991)‘Inventory of personal characteristics # 7’,unpublished measure, University ofMinnesota.
Tellegen, A. and Waller, N.G. (1987) ‘Re-exam-ining basic dimensions of natural languagetrait descriptors’, aper presented at the 95thAnnual Convention of the AmericanPsychological Association.
White, G.M. (1980) ‘Conceptual universals ininterpersonal language’, American Anthro-pologist, 82: 759–81.
Wiggins, J.S., Trapnell, P. and Phillips, N. (1988)‘Psychometric and geometric characteristicsof the Revised Interpersonal Adjective Scales(IAS-R)’, Multivariate Behavioral Research,23: 517–30.
MEASURES OF THE PERSONALITY FACTORS FOUND RECURRENTLY IN HUMAN LEXICONS 53
9781412946513-Ch01 1/9/08 2:58 PM Page 53