Reinterpreting the Myers-Briggs Type Indicator

7/28/2019 Reinterpreting the Myers-Briggs Type Indicator

1/25

Reinterpreting the Myers-Briggs TypeIndicator From the Perspective of theFive-Factor Model of Personality

Robert R. M cCrae an d Paul T. C osta , Jr.Gerontology Research Center

N ationa l Institute on A gm g, NIHABSTRACT The M yers-Bnggs Type Indicator (M BT I, Myers & McCauUey,1985) was evaluated from the perspectives of Jung's theory of psychologicaltypes and thefive-factormodel of personahty as measured by self-reports andpeer ratings on the NEO Personality Inventory (NEO-PI, Costa & McCrae ,1985b) Data were provided by 267 men and 201 women ages 19 to 93 Con-sistent with earlier research and evaluations, there was no support for the viewthat the MBTI measures truly dichotomous preferences or qualitatively distmcttypes, instead, the instrument meastires four relatively independent dimensionsThe interpretation of the Judging-Pferceivmg index was also called into ques-tion The data suggest that Jun g's theory iseither incorrect or madequately op-erationalized by the MBTI and cannot provide a sound basis for interpreting itHowever, correlational analyses showed that the four M BT I indices did m easureaspects of four of the five major dimensions of normal personality The five-factor model provides an alternative basis for interpreting MBTI findings withina broader, more comm only shared conceptual frameworkThe Myers-Bnggs Type Indicator (MBTI, Myers & McCaulley, 1985) isunusual among personality assessment devices for three reasons It isbased on one of the classic statements of personality theory, it purportsto measure types rather than traits or other continuous vanables, and it is


2/25

18 McCrae a n d Costawidely used to explam individuals' personality charactenstics not onlyto professionals but also to the individuals themselves and to theirfne nds, fam ilies, and co-workers These distinctions have made theMBTI a popular personality instrument for organizational and lndustnalpsychologists (e g , Hirsh, 1985) and for individuals who wish to under-stand themselves better

Personality psychologists, however, have generally been less enthu-siastic about the MBTI Stacker and Ross (1964a, 1964b) conducted ex-tensive and sophisticated analyses that led them to a cntical evaluationof both the typology and the scales themselves, and their concerns havebeen echoed by more recent cntics Theonsts complain that the Jungianconcepts that are supposed to underlie the MBTI have been distorted(Coan, 1978, Comrey, 1983) Psychom etncians are troubled by the con-ception of psychological types (Mendelsohn, Weiss, & Feimer, 1982)and the limited evidence that the MBTI measures anything other thanquasi-normally distributed personality traits (DeVito, 1985, Hicks,1984)

Even cntical reviewers, however, see promise in the instrument, andIts continued populanty, as well as empincal literature to da te, suggeststhat It IS effective at some level In the present article we assess the MBTIfrom the perspective of Jungian theory, we then offer an alternative con-ceptualization of the scales ofthe MBTI m terms of a taxonomy ofp er-sonality traits widely accepted in mainstream personality psychology,thefive-factormodel (Digman & Inouye, 1986, McCrae & C osta, 1987,Norman, 1963)

Jung ' s Types a n d the MBTI S ca l esIn Psychological Types, Jung (1923/1971) reviewed the history of psy-chological typologies from classical literature and poetry through thewntmgs of W illiam James as a basis for his own formulations His cen-tral distinction was between introverted and extraverted attitudes, w hichrepresent fundamental onentations to either the objective or the subjec-tive world Further, he postulated that individuals relate to the worldthrough two sets of opposed functions the rational (or judging) functionsof thinking and feeling, and the irrational (or perceiving) functions of


3/25

MBTI a n d NEO-PI 19ment of an inventory for assessing types Much of his descnption con-cerns the unconscious life of the individual, which is not directlyaccessible to self-report The subjective world of introverts is populatedby archetypes and images that are not easily communicated, Jung wrotethat introverted feeling "manifests itself for the most part negatively" (p387) and that the introverted sensing type "is uncommonly inaccessibleto objective understfuidmg, and he usually fares no better in understand-ing himself" (p 397) Descnptions of attitudes and functions sometimesseem to overlapfor exam ple, sensing types often seem to resemble ex-travertsand all classifications are complicated by the intrusion of un-conscious elements of the opposing function when the dominant, con-scious function IS overdeveloped Finally, Jung 's descnptions of whatmight be considered superficial but objectively observable charactensticsoften include traits that do not empincally covary Jung descnbed extra-verts as "open, sociable, jovial, or at least fnendly and approachablecharacter s" (p 33 0) , but also as morally conventional (p 334) andtough-mmded in Jam es's sense (p 307) Decades of research on the di-mension of extraversion show that these attnbutes simply do not coherein a single factor (Guilford, 1977, Guilford & Guilford, 1934) TheMBTI m anual itself contradicts Jung by using tough-mmded to descnbethe thinking function rather than the extraverted attitude (Myers &McCaulley, 1985, p 13)

Faced with these difficulties, Myers and Bnggs created an instrumentby elaborating on the most easily assessed and distinctive traits sug-gested by Jun g's wn tmgs and their own observations of individuals theyconsidered exemplars of different types and by relying heavily on tradi-tional psychometnc procedures (pnncipally item-scale correlations)Their work produced a set of internally consistent and relatively uncor-related mdices measunng Extraversion-Introversion (EI), Sensing-In-tuition (SN), and Thm king-Feeling (TF) They also added a Judg m ent-Fterception (JP) index that indicates, in conjunction with the EI prefer-ence, whether the rational or irrational function is dominant Each of theindices is dichotomized at a theoretically fixed zero point to show a pref-


4/25

20 McCrae and Costavance over Jung's largely untested speculations However one chooses toevaluate the instrument, it is crucial to realize that it is not lsomorphicwith the theory on which it is based In consequence, using the MBTI totest Jungian theory, or using the results of such tests to validate theMBTI, must be viewed with caution

Continuous Scoring an d Typ ologiesThe MBTI is offered as a "type indicator" on the assumption that it canclassify individuals into 1 of 16 qualitatively different types, formed bycombination of the four dichotomous preferences In some respects, thevalidity of this typology is the central question in the evaluation of theinstrument If qualitatively distinct types cannot be demonstrated, theneither Jung 's theory is wrong, or the MBTI fails to operationahze it ade-quately In the absence of evidence for the typology, the instrument be-comes merely a series of scales whose information is reduced, ratherthan increased, by dichotomous classifications, the charactenstics thatset It apart from countless other personahty instruments vanish, and itmust be evaluated and used m a more traditional context

MBTI theory specifies three levels at which distinctive typologicalcharacteristics should be seen in qualitative differences between op-posed preferences, in the identification of a dominant function throughthe use ofthe EI and JP scales, and in statistical interactions among pref-erences on external cntenaThe first issue concerns the validity of dichotomizing preference

scores Although most trait psychologists adopt the language of types indiscussing individuals (contrasting, e g , introverts and extraverts), thisIS generally done as a convenient way of saying "above average on a nor-mally distnbuted trait" or "below average " Jung himself appears toadopt this position m some of his wntings, admitting that there are in-termediate positions between pure introversion and pure extraversion, mwhich individuals are "infiuenced as much from within as from without"(1923/1971, p 516) The authors of the MBTI, however, have adoptedthe interpretation that types are mutually exclusive groups of people, and


5/25

MBTI and NEO-PI 21An alternative interpretation of the nature of types points to a moresubtle form of discontinuity, in which external correlates (or regressioncoefficients) vary for the two types (H icks , 1984, Mendelsohn et al ,1982) The manual provides some studies showing such dispanties in therelations between external cn tena and opposite preferences For exam-

ple, when grade-point average is plotted against EI scores, there appearsto be a jump at the zero point, w ith introverts showing higher grades Theeffect, however, is extremely subtle, and the authors admit that "for suchsmall differences to be visible, samples of 4,000-5,000 are needed"(Myers & McCaulley, 1985, p 158) DeVito (1985) concludes that theevidence provided by the manual on this issue is "w eak "A second typological charactenstic that must be addressed concernsthe JP index This preference is intended to show which ofth e two typesof functions^judging or perceiving is favored m dealing with the ex-ternal world, m combination with the EI preference, it determines thedominant and auxiliary functions specified in Jung's theory Individualsclassified as ENTJ (l e , Extraverted, Intuitive, Thinking, and Judging)emphasize the judging function of Thinking in the external world, as Ex-traverts, they would have Thinking as the dominant function and Intui-tion as the auxiliary INTJs (Introverted, Intuitive, Thm hng, and Judg-ing individuals) would also show a preference for Thinking m dealingwith the external world, but because they are Introverts, this would betheir auxiliary function The distinction between auxiliary and dominantfunctions is important in Jungian theory, and forms the major rationalefor the JP index It is therefore surpnsing that so few studies have at-tempted to validate it Myers and McCaulley noted that "according totheory, the dominant function will show a clearer preference [l e , ahigher preference score] than will the auxiliary" (1985, p 58 ), but ad-mitted that "scores for the dominant are greater than those for the auxil-iary in only about half the types" (p 60) as one would expect fromchance


6/25

22 McCiae and Costainteractions there is no evidentiary basis for a 16-box matnx of distincttypes" (p 13) Hicks (1984) herself failed to demonstrate predicted in-teractions, and Stncker and Ross (1964a) found no evidence of interac-tion effects m analyses of academic aptitude and performance measuresThey recommended further study of this question using "other kinds ofvanables, particularly those from the personality sphere" (p 69) '

Finally, it might be noted that scale development apparently disre-garded the hypothesized typologicd structure If the types are qualita-tively distinct, one might need to create different scales to measureThinkmg-Feeling (and Sensation-Intuition) m extraverts and in intro-verts Instead, the indices were developed as four independent scales,with Item selection pooled across all different types

Ezpenmental an d Conelational Approaches toScale ValidationM ost studies on the validity of the MBTI have focused on individual m-dices rather than the full typology Because of its explicit theore tical ba-sis, some researchers have adopted an expenmental , hypothet ico-deduc-tive approach to validation of the MBTI indices (J G Carlson, 1985,DeVito , 1985) For example, R Carlson (1980) conducted a study mwhich students were asked to wnte let ters mtroducmg themselves to animagined foreign correspondent As hypothesized, intuitive types weremo re likely to make references to the imag ined other, sensing types w eremore likely to provide physical descnptions of themselves

An alternative to the expenmental approach in personality research isthe correlational study, in which a new scale is re la ted to establ i shedme astires It is possible to test specific hyp othe ses if alternative m ea sure sof the sam e or theoretically related cons tructs are u sed In the case of theM B TI , the problem is that few other m strum ents attempt to m easure Jun-gian construc ts Aside from those m the Jungian Type Survey (Wheel-w ng ht, W heelw ngh t, & Buehler, 1964), which do correlate with the cor-

1 The dependent vanables for which interaction effects should be found must, ofcourse, be different from the MBTI categonzmg vanables If MBTI mdices are


7/25

MBTI a n d NEO-PI 23responding MBTI indices, there are few alternative measures of sensmgversus mtuition or thinking versus feeling

Correlational research may also be exploratory, seeking an interpreta-tion ofth e scale from a pattern of convergent and discnminant relationsThere is no lack of correlational data on the MBTI, as the manual (Myers& McCaulley, 1985) amply attests What is needed, however, is a mean-ingful framework in which to organize and interpret the many corre-lates a set of standard dimensions by reference to which the MBTI in-dices can be understood Ideally, this set would systematically samplethe full range ofpersonality traits, providing the basis for a comprehen-sive assessment of the MBTIOne classification of traitsthefive-factormodel seems to providea comprehensive taxonomy Beginning with the work of Allport and Od-bert (1936), several studies have attempted to specify exhaustively therange of personality traits by examining English language trait names,on the assumption that native speakers would have evolved words for allimportant individual differences Independently, Block (1961) and hiscolleagues sought to provide a universal, clinically based language fordescnbing all important aspects of personality, and they developed theCalifornia Q-Set (CQS) Research from both these traditions has con-verged on thefive-factormodel ofpersonality (Goldberg, 1981, McCrae,Costa, & Busch, 1986, Nonnan, 1963), and the same factors have alsobeen identified in studies of several standard personality inventories(Costa & McCrae, 1985a, McCrae & Costa, 1985a, 1987)Thefive-factormodel is not based on any single theory ofpersonality,but has been shown to encompass scales that operationalize a ntimber oftheoretical perspectives (McCrae & Costa, 1985a, 1985b) The modelhas been recovered m ratings (Tlipes & Chnstal, 1961) as well as self-reports (McCrae & Costa, 1985c), in German (Amelang & Borkenau,1982) as well as Enghsh, m children (Digman & Inouye, 1986) as wellas adults Although not universally recognized as a comprehensive tax-onomy (Waller & Ben-Porath, 1987), an mcreasmg number of person-


8/25

24 McCiae a n d Costaence is seen in imaginativeness, aesthetic sensitivity, depth of feeling,cunosity, and need for vanety Agreeableness encompasses sympathy,trust, cooperation, and altruism. Conscientiousness includes organiza-tion, persistence, scrupulousness, and need for achievement

Those familiar with the content and correlates of the MBTI indicesmay note some conceptual similanties, beyond the obvious alignment ofthe two Extraversion scales (Dachowski, 1987) In particular, Intuitivesseem to be open to expenence whereas sensing types are closed. Feelingtypes seem to be agreeable and Thinking types antagonistic, and Judgingtypes seem high and Pfcrceivmg types low m Conscientiousness ^ Notethat these correspondences do not necessanly denve directly from Jun-gian theory Jung (1923/1971) does descnbe both extraverted and intro-verted intuitive types as "continually scenting out new possibilities" (p400) in the outer and inner worlds, respectively, so the link of SN withOpenness seems justified However, as we have pointed out, Jung's con-ception of extraversion is more diffuse than that embodied in the E xtra-version factor Moreover, it would be difficult to justify the associationof the feeling function with Agreeableness Fbr Jung, thinking is an in-tellectual activity m which judgments are based on the rational applica-tion of pnnciples,^ e//g is the assignment of value (acceptance or rejec-tion) to objects of expenence An individual whose first reaction to eachexpenence was a judgment of rejection (contempt, m istrust, or hatred)without a logical basis would be classified by Jung as a feeling type , butwould probably score very low on the MBTI Feeling preference As de-scnbed by Jung, the thinking and feeling functions seem to refer to em-phases on different modes of expenence, and might be related more di-rectly to facets of Openness (see Coan, 1974)

In the present research, we first examine charactenstics of the MBTItypology, evaluating the utility of the JP scale in indicating dominantfunction and searching for interaction effects on personality vanablesWe then consider relations between the MBTI continuous scales and thefive factors ofpersonality as measured by the NEO Personality Inventory(NEO-PI), usmg both self-reports and peer ratings ofpersonality In ad-dition, we test the hypothesis that the TF scale is positively related to

2 Similarly, Conley (1985) associated Sensing versus Intuition with a Conse rva-


9/25

MBTI a n d NEO-PI 25openness to Feelings, and negatively related to openness to Ideas Theseanalyses should provide a basis for understanding the MBTI scoreswithin the framework of a widely accepted personality taxonomy

METHODS u b j e c t s

Participants in the Baltim ore Longitudinal S tudy of Ag mg (BLSA , Shock etal , 1984) and their spouses provided data for the study BLSA volunteersare a group of predominantly white, community-dwelling individuals whohave agreed to return for periodic biomedical and psychological testingM ost have at least a college degre e, and many have advanced degrees Co m -pansons with a national sample (Costa et al , 1986) suggest that they aresomewhat higher than the general population in Openness to Expe nen ce, butdo not differ m arkedly m Extraversion or Neuroticism The 267 men whoprovided complete data on both the NEO-PI and MBTI ranged in age from21 to 93 (M = 62 7 ), the 201 women ranged in age from 19 to 93 (M =58 9)

A subsample of 70 men and 35 women had previously participated m apeer ratmg study M ean ratings on the NE O-PI were provided by from oneto four fnends and ne ighb ors, details are given elsewhere (McC rae & C osta ,1987)

MeasuresForm G of the MBTI (Myers & McCaulley, 1985) was used in this studyThis form consists of 126 questions, of which only 94 are actually scoredMost Items offer a forced-choice between two respo nse s, although somehave more response options, and respondents are occasionally allowed toendorse two or m ore responses Separate scon ng keys are provided for eachpreference (e g , both E and I, both S and N ), although m ost items are sim-ply reverse scored for the oppo sing function Because the oppo sing prefer-ence scores are alm ost completely lpsative , they are not analyzed separatelyhere Preference is determined by the greater of the two preference scores(with provisions for breaking ties), and a four-letter code (e g , INTJ) sum -


10/25


11/25

MBTI and NEO-PI 27RESULTS

Although the absence of normative information on continuous scoresmakes companson difficult, the present sample seems to resemble otheradult samples m the distnbution of MBTI scores Fbr purposes of com-panson , subjects were classified into four-letter code types Men clus-tered m the ISTJ (27 7%) and ESTJ (18 0%) types, whereas womenwere somewhat more evenly distnbuted among the 16 types Normativedata from a national sample (Tables 4 12 and 4 13 m Myers & Mc-Caulley, 1985) show a similar pattem Age, sex, and years of educationwere correlated with the four continuous scales, only three correlationswere greater than 20 age with JP (r = - 28 ), gender with TF (r =39), and education with SN (r = 23) Older subjects preferred judging,women preferred feeling, and better-educated individuals preferred in-tuition

Jungian theory suggests that a balance between opposing functionsshould be developed in later life, so it might be hypothesized that oldermen and women would show less extreme scores on the SN and TFscales However, F tests for the equality of vanance for subjects aboveand below age 65 showed no significant differencesa nullfindingrep-licated by the data m Appendix C of the MBTI manual given for ayounger group of adults These cross-sectional studies provide no sup-port for the Jungian theory of adult personality development EitherJung's theory is incorrect on this point, or the M BTI indices do not ade-quately operationalize Jungian constmcts

An alyses of the Ty pologyAccording to MBT I theory, individuals classified as EJ or IP have a dom-inant judging function (l e , T or F), whereas those who are classified asEP or IJ have a dominant perceivmg function (S or N) The dominantfunction IS the most highly developed, so the individual should show theclearest preference for it Strength of preference for judging can be as-sessed by the distance (in either the thinking or feeling direction) of thecontinuous TF score from the neutral point of 100, similarly, strength of


12/25

42653548

72804668

28 McCiae and CostaTabto lCross-Tabulation otHypothes ized Dommant Funct ion by ObservedStronger Preference

Hypothesized dominant functionObserved stronger preference Judging PerceivingMen"JudgingPerceivingWomen"JudgingPerceivingNote Data from 8 men and 4 women with tied preference scores are omitteda Continuity-adjusted chi-square = 1 37, nsb Continuity-adjusted chi-square = 0 01, ns

square analysis showed no significant association for either men orwomen It is possible that these results are due to the fact that the rangefor the SN score is greater than the range for the TF score, mcreasmg theprobability that all individuals will have a higher preference score on theperceivmg function To examine this possibility, both SN and TF scoreswere standardized to have a mean of 0 and a standard deviation of 1, thehigher absolute value was taken to indicate a more clearly preferred func-tion Again, chi-square analysis failed to show a significant associationwith the hypothesized dommant function in either men or women Thesedata provide no support for the MBTI system as a means of assessing theJungian dominant function

To examine the configural effects ofthe MBTI classifications on nor-mal personality dimensions, multivanate analyses of vanance were per-formed on the five NEO-PI factor scores using the four MBTI dichoto-mies as classifying vanables A total of 55 two-, three-, and four-wayinteractions were thus examined Overall F tests were significant for themain effects andforhe EI x JP interaction, but not for any of the 10other two-, three-, and four-way interaction terms Inspection ofthe uni-vanate F tests for the EI x JP interaction showed one significant effectNEO-PI Neuroticism was particularly high among individuals whoshowed both I and P preferences However, when examined withm sex,


13/25

MBTI a n d NEO-PI 29Table 2M ea n Leve ls of NEO-PI Factor Sco res for MBTI Preferences

NEO-PI factorNeuroticismExtra-

versionOpennessAgreeablenessConscien-

tiousness

E48 655 75 0 450 34 8 7

N 214

I50 4*44504949

254

S49 44*** 48 8218

45 149 049 9

264

MBTIN49 7

50 5

preferenceT47

4856 9*** 4950 648 5

204

4650

297

77693

F52 8***51 1*5 1 554 4***47 6***

J4848494951

171 357

98154

P5151545042

111

89**Q * * *2 7 * * *

Note NEO-PI = NEO Personahty Inventory MBTI = M yers-Bnggs Type IndicatorE = Extraverted, I = Introverted, S = Sensing, N = Intuitive T = Thinking, F =Feeling, J = Judging, P = Perceiving*p< 05**p< 01**V< 001the four-letter type codes of the MBTI simply summarize four indepen-dent main effects

Analyses were then repeated using contrasting M BTI preferences andgender as classifying vanables Because there were no Preference xGender interactions, only the mam effects for preference need be consid-ered, mean values for the NEO-PI factors (expressed as T scores) aregiven m Table 2 These findings show that individuals categonzed as In-troverts by the MBTI score slightly higher in NEO-PI Neuroticism andmuch lower in NEO-PI Extraversion, Intuitive types score higher inOpenness to Expenence , Feeling types score higher m Neuroticism, Ex-traversion, and Agreeableness, and lower m Conscientiousness, and Psr-ceivmg types score higher m NEO-PI Extraversion and Openness, andlower m Conscientiousness


14/25


15/25

MBTI a n d NEO-PI 31Tabl*4Correlations of Peer-Sated NEO-PI Facto is With MBTI ContinuousScales in Men a n d Women

MBTI scalesMenEI (Introversion)SN (Intuition)TF (Feeling)JP (Perception)Women

EI (Introversion)SN (Intuition)TF (Feeling)JP (Perception)

N

0 507- 11

- 100 8161512

NEO-PI factorE_ 34**

0 62211

- 38*09- 0100

0

1561*** -010 4

- 0 441* -- 0618

A

1027*2 5 *100 30146**11

C

17- 15- 16- 34**- 02- 04- 39*- 55***

Note N = 10 for men, 35 for women NEO -PI = NEO Personality Inventory, MBTI= M yers-Bnggs Type Indicator N = Neuroticism, E = Extraversion, O = Opennessto Exp enen ce, A = Agreeableness, C = Conscientiousness*p< 05**p< 01***p< 001cation of the findings seen within self-reports in Tables 2 and 3 In bothmen and women, EI is significantly (negatively) correlated with peer-rated Extraversion, SN with peer-rated Openness, TF with peer-ratedAgreeableness, and JP (negatively) with peer-rated ConscientiousnessTwo other correlations are significant, but not replicated across genderMen who score m the Intuitive direction are seen by peers as being lowerin Agreeableness, and women who score m the Feeling direction are seenas being lower m Conscientiousness Note that none of the MBTI indicesIS related to peer-rated Neuroticism

The relations shown in Tables 3 and 4 were replicated when alternativemeasures of the five factors were examined in subsets of subjects TheEI index was correlated with Extraversion factors from the CQS (r =- 58, N = 168, p < 001) and the adjective checklist (r = - 62,N


16/25

32 McCrae a n d CostaScatterplots of the significant relations in Tables 3 and 4 were exam-ined visually for evidence of deviations from lineanty or discontinuity atthe transition point between contrasting types None was apparentFinally, the MBTI TF scale was correlated with two relevant facets of

Openness to Expenence that provide a test ofthe distinctive onentationsof thinking and feeling As hypothesized, a preference for feeling wasdirectly related to Openness to Feelings in both self-reports (r = 30, N= 468,p< 001), and peer ratings ( r = 2 4 ,A ^ = 1 0 5 ,p < 05), i tw asinversely related to Openness to Ideas (rs = - 16, p< 001, and - 21,p < 05, for self-reports and peer ratings , respectively) Thus, althoughthe TF scale appears chiefiy to measure aspects ofthe domain of Agree-ableness, It does show some ability to differentiate between onentationstoward thinking and feeling

DISCUSSIONThe results of the present study provide the basis for a very mixed eval-uation-or a radical reinterpretation of the MBTI Consistent withmuch previous research (especially Stncker & Ross, 1964a), the studyfound no support for the typological theory the instrument is intended toembody There was no evidence that preferences formed true dichoto-mies, the 16 types did not appear to be qualitatively distinct, becauseanalyses of then-joint effects on personality dimensions showed that only1 of 55 interactions was significant, and only m women, and, contrary tohypothesis, the theoretically dominant function was no more clearly pre-ferred than the auxiliary The Jungian prediction that opposing functionsshould be developed in later life was not confirmed using the MBTI Thecorrelates of individual scales were consistent with their item content, 'but would probably not have been predicted from Jungian theory Fronithe perspective of construct validation, the MBTI appears to have verysenous problems Weighing the evidence to date, the MBTI does notseem to be a promising instrument for m easun ng Jun g's types, thosewho embrace Jung's theory should probably avoid the MBTI

Conversely, those who have found the MBTI to be a useful instrumentfor assessing and descnbing individual differences should senously con-sider abandoning Jungian theory and some of the associated language


17/25

MBTI a n d NEO-PI 3 3impressive evidence of convergence with one of the five major dimen-sions of normal personality, whether assessed through self-reports orpeer ratings It is these convergences that probably account for the manymeaningful associations between MBTI scales and extemal cn ten a suchas occupational preferences, creativity, and educational performanceThe correspondences seen in the present study are generally consistentwith the body of literature on the correlates of the MBTI (see Myers &McCaulley, 1985, pp 175-223) For example, the SN index, which ishighly correlated with Openness to Expenence m the present study, isalso related to the Sixteen Personality Factor Questionnaire (16PF, Ca-tell, Eber, & Tatsuoka, 1970) Intelligence, Tender-Mmdedness, Imagi-nativeness, and Radicalism scales, which are known to form an opennessfactor (McC rae & Co sta, 1985b) Studies sum m anzed m the M BTImanual showing that creativity is characteristic of intuitive types areconsistent with recent evidence that most personality vanables related tocreativity measure aspects of Openness to Expenence (McCrae, 1987)Again, the JP scale seems from the current perspective to measure anaspect of Conscientiousness, and the manual shows that it is related toother measures of orderliness and self-disciplme, such as 16PF Controland Superego Strength

The view of the M BTI indices provided by the five-factor model ex-plains some anomalous correlations reported in the manual For exam-ple, one might suppose that the Thinking Introversion scale of the Om-nibus Pfersonality Inventory (OPI, Heist, McConnell, W ebster, & Yonge,1963) would be related to Thinking and to Introversion, m fact, it is vir-tually unrelated (rs = 10) to either the EI or the TF index However, itIS strongly related (r = 53) to the SN index These relations make senseif Thinking Introversion is viewed as a form of thoughtfulness or open-ness to ideas The Thinking Introversion factor of the Guilford-Zimmer-man Temperament Survey (Guilford, Z imm erman, & Guilford, 1976) isknown to load on a broader Openness to Experience factor (Costa &McCrae, 1985a)

Similarly, a reinterpretation o fthe MBTI scales can make better senseof some of the reported correlates of the TF index TF is positively re-


18/25

34 McCiae a n d C ostastmct of Agreeableness The TF index emphasizes a cognitive style as-sociated with Agreeableness, the Stem scales represent interpersonal a s-pects of the same broad domain

The five-factor model is purely d esc np tive , it does not exp lain theongins ofpersonality nor the mechanisms that account for individual dif-ferences Can Jungian psychology provide a theoretical basis for any ofthe factors'' In particular, can individual differences in Openness andAgreeableness, respectively, be attnbuted to different styles of perceiv-ing and judging " A brief examination of this possibility raises somedoubts Although openness is related to perception m the broad sense oftaking m information, the contrast between sensation and intuition doesnot seem to capture the essence of openness Open individuals use theirfive senses as much or more than closed individuals, they differ not mthe type of information they prefer, but rather m the quantity Similarly,agreeable people differ from antagonistic people not simply in a prefer-ence for feeling over logic, but in a preference for warm feelmgs overcold logic Other psychologists, however, may perceive deeper connec-tions between Jungian theory and thefive-factormodel, further explora-tion of this issue would be welcome

The dimensional model offered here as an altemative to a Jungian ty-pology may seem deficient in another respect In their cntique of the ear-lier three-dimensional NEO trait model, Whitboume and Weinstock(1986) argued that "What is still lacking is a set of predictions con-ceming the interrelationships among the three dunensions and differentcombinations of scores on the three personality factors" (pp 224-2 25)The M BTI types attempt to satisfy the need to consider the whole pattemof traits and summanze personality m a single comprehensive GestaltIn fact, however, there is no evidence that MBTI types represent uniqueconfigurations, they merely summanze four additive main effects INTJsdiffer from ENTJs only m the ways that mtroverts differ from extraverts,EI does not interact with the other indices to form qualitatively new com-binationsThe fact that personality dimensions do not appear to m teract to form

distmct types of persons does not mean that traits do not mteract with


19/25

MBTI a n d NEO-PI 35So m e Im pl ica t ions for the In terpreta t ion

of th e MBTIThere are num erous books, chapters, and pam phlets devoted to an expla-nation of the MBTI types for use by counselors, personnel psycholo-gists, educators, and laypersons They descnbe the 16 cells individuallyor in subgroups (e g , Introverted Thinkers) The descnptions are basedin part on Jungian theory, in part on the item content of the indices, andin part on empm cal research or personal expenence with the types Howwell these descnptions square with known correlates of the indices canbe roughly gauged by substituting the corresponding factor names fromTable 3 for the MBTI codes Fbr example, the ENFJ type is , m terms ofthe five-factor model, extraverted, open, agreeable, and conscientiousMyers and McCaulley's (1985, p 21) descnption of individuals of thistype appears to fit extraverted, agreeable, and conscientious individuals(although there is httle m the descnption suggesting openness) Most ofthe descnptions provided m the manual seem to be reasonably good bythis cntenon

However, the accompanying assertions about the dominance of partic-ular preferences in inner and outer life are based solely on Jungian theoryand on the use of the JP and EI scales to determine the dominant func-tion, and are not supported by data There is no good evidence that theJP scale has any beanng at all on the relative importance of thinking orperceiving However, MBTI users might wish to reconceptualize the JPscale as an mdex of preference for structured versus spontaneous living ,this view IS supported by the item content and the extemal correlatesThe use of dichotomous preference scores is also questionable, asmany reviewers have noted Some authors (e g , K eirsey & Bates , 1978)have taken the reality of the dichotonues so senously that they estimatetheir num bers m the populationsurely an unwarranted reification Inmterpretmg scores, Myers (1980) advised individuals to explore closelyrelated types (differmg in one or two letters) to see if a better fit could befound In fact, however, most individuals will probably accept at facevalue whatever descnption is provided This system errs in two ways by


20/25

36 McCrae an d Costaor so of the distnbution on each index as an "x " Data on the retest sta-bility of types (Table 10 7 of the manual) show that such a policy couldsubstantially increase reliability

Merits of Alternative Instruments a n dApproaches

Most personality inventones report scores in terms of continuous sca les,using all available information MBTI users have this option, by attend-ing to the preference scores or continuous scores A com panson of theMBTI w ith the NEO -PI, however, suggests some ways m which M BTIusers might benefit from the supplementary use of other personality in-struments Most conspicuous is the lack of a Neuroticism factor in theMBTI Its absence is understandable on two counts first, because emo-tional instability versus adjustment did not enter into Jung's definitionsof the types, and second, because the authors of the test were apparentlyphilosophically committed to a position which saw each type as equallyvaluable and positive (Myers with M yers, 1980)a view that is difficultto hold with regard to Neuroticism ' Although it makes interpretation ofresults palatable to most respondents, this approach also omits informa-tion that may be cmcial to employers, co-workers, counselors, and theindividuals themselves For many, if not most, applications, some mea-sure of Neuroticism would be useful

Further, the MBTI does not give comprehensive information on thefour domains it does sample The TF scale, for exam ple, encom passesthe tough- versus tender-minded aspect of Agreeableness, but has no di-rect measures of interpersonal aspects such as trust, altmism, and co-operativeness All four indices give only a broad, global picture , withoutany distinctions of traits within each domain Other personality mea-sures give more detail For example, the NEO-PI provides informationon SIX specific facets of ExtraversionWarmth, Greganousness, Asser-tiveness , Activity, Excitement Seefang, and Positive Emotions This isespecially important for understanding individuals who score in the av-erage range on overall extraversion Individuals who are good leaders butrelatively sober and senous may score the same as others who are cheer-


21/25

MBTI a n d NEO-PI 3 7ful and jov ial , but would rather let others make decisions Such differ-ences would be important in a variety of applied settings, and show thepotential value of supplementing the MBTI with other measures

There are two ways in which the users of other psychological tests canbenefit from experience with the MBTI First, translation of the MBTImdices into the common language of thefive-factormodel makes it pos-sible to assimilate the extensive empincal literature on the MBTI Forexample, dozens of occupations have been classified with respect to thefrequency of MBTI preferences (Myers & McCaulley, 1985) Intuitivetypes are disproportionately found in such occupations as artist, wnter,and psychologist, whereas sensing types choose vocations such asfarmer, mechanic, or clencal worker When the SN index is interpretedas a measure of Openness to Experience, these data provide valuable rep-lications of other work linking Openness to vocational preferences(Costa, McCrae, & Holland, 1984)

At another level, those who develop and employ personahty measurescan leam a good deal from the strategies of application that have beenused w ith the MBTI Decades of expenence show that it is possible andsometimes useful to descnbe personality m lay terms, to teach individ-uals the nature and value of individual differences, and to increase un-derstanding among people who interact by shanng information on per-sonahty and explaining differences m styles Adapting such approachesto instm ments that include measures of ostensibly undesirable traits suchas Neuroticism will require tact and skill, but promises to be valuable inthe long mn Regardless of the mstmment used, information given tolaypersons should refiect current knowledge m personality psychologyIf the MBTI IS used, evidence to date suggests that it may be better toabandon its Jungian framework and reinterpret the MBTI in terms ofthefive-factor model

SEFERENCES


22/25

38 McCiae an d CostaBlock, J (1961) The Q-sortmethod in personality assessment and psychiatric researchSpringfield, IL Charles C ThomasBlock, J , & Ozer, D J (1982) Two types of psychologists Remarks on the Mendel-sohn , Weiss, and Feimer contnbution Journal qfPersonality and Social Psychology,

42, 1171-1181Carlson, J G (1985) Recent assessments of the Myers-Bnggs Type Indicator Jburna/cf Personality Assessment, 49, 356-365

Carlso n, R (1980) Studies of Jungian typology II Representations of the personalworld Journal qf Personality and Social Psychology, 38 , 801 - 810Cattell, R B , Eber, H W , & Tatsuoka, M M (1970) The handb ook for the Sixteen

Personality Factor Questionnaire Cham paign, IL Institute for Ftersonality and Abil-ity TestingCoan, R W (1974) The optimal persona lity New York Columbia Umversity PressCoan, R W (1978) [Review of Myers-Bnggs Type Indicator] In O K Buros (Ed ),The eighth mental measurements yearbook (Vol 1 , pp 97 3- 97 5) Highland Park,NJ Gryphon PressComrey, A L (1983) An evaluation ofthe Myers-Bnggs Type Indicator Academic Piy-

chology Bulletin, S, 115-129Conley, J J (1985) A personality theory of adulthood and aging In R H o g a n & W H

Jones (Eds), Perspectives in personality (Vol 1, pp 81-1 15) Greenwich, CT JAIPressCosta, P T , Jr , & M cCrae, R R (1985a) Concurrent validation after 20 years Impli-

caUons ofpersonality Stability for Its assessment In J N Butcher & C D Spielber-ger (Eds ), Advances in personality assessment (Vol 4, pp 31 -5 4) Hillsdale, NJLawrence ErlbaumCosta, P T , Jr , & McCrae, R R (1985b) The NEO Personality Inventory m anualOdessa, FL Psychological Assessment ResourcesCosta, P T , J r , & McC rae, R R (1988) Pfersonality m adulthood A six-year longitu-dinal study of self-reports and spouse ratings on the NEO Personality Inventory Jour-

nal qfPersonality and Social Psychology, 54, 853-863Costa, P T , Jr , McCrae, R R , & Holland, J L (1984) ftrsonality and vocationalinterests m an adult sample Journal qf Applied Psychology, 69, 390-40 0Costa, P T , Jr , M cCrae, R R , Zonderman, A B , Barbano, H E , Lebowitz, B , &Larson, D M (1986) Cross-sectional studies of personality in a national sample 2

Stability in neuroticism, extraversion, and openness Psychology and Aging, 1 144-149Dachowski M M (1987) A convergence of the tender-minded and the tough-m inded''American Psychologist, 42, 886-887DeV ito, A J (1985) [Review of Myers-Bnggs Type Indicator] I n J V Mitchell, Jr

(Ed), Ninth mental measurements yearbook (Vol 2, pp 1030-1032 ) Lincoln Uni-versity of Nebraska Press


23/25

MBTI a n d NEO-PI 3 9Goldberg, L R (1981) Language and individual differences The search for universals

m personality lexicons In L Wheeler (Ed ), Review qf personality and social psy-chology (Vol 2, pp 141-165) Beverly Hills Sage

Guilford, J P (1977) Will the real factor of extraversion-introversion please stand up ''A reply to Eysenck Psychological Bulletin, 84, 412-416Guilford, J P , & Guilford, R B (1934) An analysis of the factors in a typical test ofmtroversion-extroversion Journal qfAbnormal and Social Psychology, 28, 377-399

Guilford, J S , Zimmerm an, W S , & Guilford, J P (1976) The Guilford-ZimmermanTemperament Survey Handbook Twenty-five years qf research and application SanDiego EdITS Publishers

Heist, P A , M cConnell, T R , Webster, H , & Yonge, G D (1963) Omnibus Person-ality Inventory New York The Psychological Corporation

Hicks, L E (1984) Conceptual and empincal analysis of some assumptions of an ex-plicitly typological theory Journal of Personality and Social Psychology, 46, 1118-1131

Hicks, L E (1985) Dichotomies and typologies Summary and implications Journalqf Psycholog ical Type, 10, 11-13

Hirsh,S K (1985) Using the Myers-Bnggs Type Indicator in organizations A resourcebook Palo Alto Consulting Psychologists Press

Jung, C G (1971) Psycholog ical types (H G Baynes, Trans , revised by R F CHull) Pnnceton Pnnceton University Press (Ongm al work published 1923)

K eirsey, D , & Ba tes, M (1978) Please understand me Character and temperamenttypes Del Mar, CA Prometheus Nemesis BooksM cCrae, R R (1987) CreaUvity, divergent thinking, and openness to expenence Jour-

nal (rfPersonality and Social Psychology, 52, 1258-1265M cCrae, R R , & Costa, P T , Jr (1983) Social desirability scales More substance

than style Journal of Consulting and Clinical Psychology, 51, 882-888McC rae, R R , & Costa, P T , Jr (1985a) Companson of EPI and Psychoticism scales

with m easures of thefive-factormodel of personality Personality and Individual Dif-ferences, 6, 587-597M cCrae, R R , & Costa, P T , Jr (1985b) Openness to expenence In R Hogan & WH ]one&(Ejd&),Perspectives in personality (Vol 1, pp 145-172) Greenwich, CTJAI Press

M cCrae, R R , & Costa, P T , Jr (1985c) Updating Norm an's "adequate taxonomy"Intelligence and personality dimensions in natural language and in questionnairesJournal afPersonality and Social Psychology, 49, 710-721

M cCrae, R R , & Costa, P T , Jr (1987) Validation ofthe five-factor model ofperson-ality across instruments and observers Joumal cfPersonality and Social Psychology,5 2 , 8 1 - 9 0


24/25

40 McCrae a n d CostaMendelsohn, G A Weiss D S , & Feimer N R (1982) Conceptual and empincalanalysis of the typological implications of patterns of socialization and feminity Jour-nal ofPersonality and Social Psychology, 42, 1157-1170Myers I B (1980) Introductwnto npe Orded) Palo Alto Consulting PsychologistsPressMyers I B ,& McCaulley M H (1985) Manual A guide to the development and useofthe Myers-Bnggs Type Indicator Palo Alto Consulting Psychologists PressMyers, I B , with Myers, P B (1980) Gifts differing Palo Alto Consulting Psychol-ogists PressNorman, W T (1963) Toward an adequate taxonomy ofpersonality attributes Rep li-cated factor structure m peer nomination personality ratings Journal ofAbnormal andSocial Psychology 66 574-583Shock, N W Greulich, R C Andres, R , Arenberg, D , Costa, P T Jr , Lak atta,

E G ,&Tobin J D (1984) Normal human aging The B altimore Longitudinal Studvo/i4gmg (NIH Publication No 84-2450) Bethesda National Institutes of HealthStem, M I (1966) Volunteers for pea ce New York WileyStncker, L J , & Ross, J (1964a) An assessment of some structural properties of theJungian personality typology Journal of Abnorm al and Social Psycho logy, 6 8 6 2 -71Strieker, L J , & Ross, J (1964b) Some correlates of a Jungian personality inventoryPsychological Reports, 14, 62 3-64 3Tupes, E C , & Chnstal R E (1961) Recurrent persona lity factors based on trait rat-

ings (USAF ASD Tech Rep No 61-97) Lackland Air R)rce Base, TX U S AirForceWaller, N G , & Ben-Porath, Y S (1987) Is it time for clinical psychology to embracethe five-factormodel of personality''vAmencan Psychologist, 42, 887-8 89Weiss, D S , Mendelsohn G A , & Feimer, N R (1982) Reply to the comm ents of

Block and Ozer Journal qf Personality and Social Psychology 42 1182-1189Wheelwnght J B , Wheelwright, J H , & Buehler, H A (1964) Jungian Type SurveyThe Gray-Wheelwright test (I6th revision) San Francisco Society of Jungian Ana-lysts of Northern CaliforniaW hitboume,S K ,& Weinstock, C S (1986) Adult development (2nd ed) New YorkPraeger

Manuscript received November 23, 1987 revised August 8, 1988


25/25

Reinterpreting the Myers-Briggs Type Indicator

Documents

Transcript of Reinterpreting the Myers-Briggs Type Indicator