Recent Assessments of the Myers-Briggs Type Indicator

Journal of Personality Assessment

ISSN: 0022-3891 (Print) 1532-7752 (Online) Journal homepage: http://www.tandfonline.com/loi/hjpa20

John G. Carlson

To cite this article: John G. Carlson (1985) Recent Assessments of the Myers-Briggs Type Indicator, Journal of Personality Assessment, 49:4, 356-365, DOI: 10.1207/s15327752jpa4904_3

To link to this article: http://dx.doi.org/10.1207/s15327752jpa4904_3

Published online: 10 Jun 2010.

Submit your article to this journal

Article views: 199

View related articles

Citing articles: 61 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=hjpa20

Download by: [KSU Kent State University] Date: 15 June 2016, At: 04:54 Journal of Personality Assessment, 1985,49, 4

Recent Assessments of the Myers-Briggs Type Indicator JOHN G. CARLSON University of Hawaii Abstract: The present paper focuses on approximately two dozen recent published studies that examined reliability and validity of the Myers-Briggs Type Indicator (MBTI) in clinical, counseling, and research settings. Several assessments of split-half and test-retest reliability of the standard Form E and shorter Form G of the Inventory have Yielded generally satisfactory correlations far all four scales. A larger number of studies of construct validity of the MdTI have yielded support for research hypotheses in situations ranging from correlations of the MBTI with a personality inventory, to couples problems in a counseling setting, to line judgment in groups, and others, Therefore, the applications of the MBTl have been broad, although somewhat unsystematic, and with generally favorable validity assessment. Continued attempts to validate the instrument in a variety of settings are needed. In the words of Isabel Briggs Myers, ward, towards other people or external the Myers-Briggs Type Indicator (MBTI) objects in the "attitude" labeled extra- is a test designed "to implement Jung's version. By contrast, thelibido may flow theory of type" (1962, p. 1). Therefore, inward, towards one's own conscious like the projective teckniques, the MBTI self, defining the attitude, i~troversion. is closely allied with psychodynamic The distinction essentially determines thought, at least in its original concep- dlle'9 firimary direction for mental function. However, unlike the projectives, tioning, either predorrlinately upon the the MBTI is a highly quantified device external world qs a source of f;lctual with distinct methods of administration, material (the extraverted personality) or objective scoring, and for which data upon the internal woild as a source of exists concerningnorms, reliability, and ideas and concepts (the introverted per- validity. sonality). Following a brief description of the Besides these two fundamental atti- theoretical underpinnings of the MBTI, tudes, each of us is said to perform men- this paper will review very recent as- tal "fu~ullctions" along two dimensions. sessments of reliability and validity of The first dimension, perceiving, describes the scales within the Indicator. Some- two polar opposites in terms of the ways what more extensive reviews of the ear- that information is received. In sensa- lier literature relevant to the MBTI may tion, data takes the form of acts or sense be found in Myers (1962) and Carlyn impresgions, that is, information from (1977). The: present paper focuses on the five senses. In intuition, one's per- published studies from 1975 that exam- ceptions are indirect, distorted by the ined reliability and validity of the MBTI unc~n~eious,and the sources of one's in clinical, counseling, and research set- information are not very clear, as in tin@. '%humches" or the "sixth sense." The in- In one view, Carl Jung's theory of tuitive-dominant personality will tend types (1923) is based upon distinctions to rely more upon the latter mode for with respect to the sources from which recoipt of significant data; just as the Downloaded by [KSU Kent State University] at 04:54 15 June 2016 information is derived, the ways that sensgtion-dominant individual prefers people perceive information, and the to utilize the basic senses for receipt of ways that information is dealt with in information and not ta gp beyond the reaching conclusions. In Jung's view, objective, empirical world for facts. psychic energy, or libido, may flow out- The other dimension of aental func- tioning in Jung's th~oryof types is the This is to express my sincere thanks to my dear friend judging function, the function that proc- and colleague, Dr. Ruth Sherman, for my first ~ntroduc- tlon tp the MBTl and assistance during preparation of esses in some way the the manuscript. matian that is perceived, irrespective of JOHN G. CARLSON its source. The thinking-dominant type ing (TF function) is said to be dominant prefers to analyze, synthesize, and de- in one's dealings with the world. Thus, termine the truth or falseness of infor- with this additional dimension, 16 pos- mation in an impersonal fashion. By sible personality types are generated by contrast, the feeling-dominant type combinations of preferences. evaluates incoming data in terms of Form F of the Indicator, apparently their goodness or badness, a process the most commonly used version, con- that is personal, subjective, and non- tains 166 forced-choice items. Form G,a analytical (but nonetheless rational, in more recent innovation, makes use of Jung's terms). 126 items and appears to be a reliable In any individual, one of the polar and faster alternative to Form F (Cars- opposites of each function (sensing vs. kadon, 1979). Each scale of either form intuition, thinking vs. feeling) and one is a subset of items designed 'Yo re- of the two attitudes (extraversion vs. in- flect.. . a habitual choice between oppo- troversion) are possible. Therefore, in sites, analogaus to right- or left-handed- this view there are eight possible combi- ness" (Myers, 1962, p. 2). nations or personality types in terms of For purposes of computing correla- mental attitudes and f~~nctions. tions and for regression analyses of the The Myers-Briggs formulation essen- MBTI, a continuous score may be com- tially parallels Jung's theary in terms of puted (or found in a table in Myers, these three dimensions of personality 1962) that converts type and preference plus a fourth dimension that was "impli- score into a single value. But more cit" in Jung's system (Matoon, 1981). commonky, preference scores on the An individual's type is assessed by means NBTI are expressed as types in terms of of the MBTI which uses scales or indi- the closest pole on a given scale. ces fot determining an individual's "preferences" for a given attitude or function. Reliability and Validity Thus: ad the MBTB: 1975-83 the EL index is designed to reflect whether The following is a selective discussion the person is an extrsvert or an introvert of recent attempts to establish various in the sense intended by Jung.. . . The SN types of reliability and validity of the index is designed t~ reflect the person's MBTI. The recent volpminous litera- prefrrenoe as betwegn two opposite ways ture on the scale (one bibliography lists of perceiving, i.e., whether he relies pri- approximately 700 referenc:~~)reflects marily on the familiar process of sensing. . . or primarily on the less obvious process largely successful efforts to apply it in a of intuition.. . . The TF index is designed large variety of educational, clinical, to reflect the person's prefmence as be- counseling, business, and research set- tween tfwo opposite ways of judging, i.e., tings. Perhaps the widest usage of the wh&ther he relies primarily upon think- test is in educational areas, several re- ing.. . or primarily upon feeling (Myers, views of which are contained in Cars- 1962, -pp. 1-2). kadon (1981). The present review fo- In addition to the El, SN, and TF cuses on studies in clinical, counseling, indices, the MBTI introduces the, JP and research settings. (judging-perceiving) index. While not a

Downloaded by [KSU Kent State University] at 04:54 15 June 2016 direct application of Jung's type theory, Reliability the index does relate to Jung's notion The original reliability studies, report- that one cnd of a dirrqe~nsionof personal- ed in the Myers-Priggs Manual, yielded ity will tehd to dominate in one's every- split-half reliability coefficients (Pear- day ilvteracrtjans with the environment. son rs) commonly exceedinq .80. More The Myers-Briggs JP index assigns a recent studies, including assessment of preference ta one or the other two men- test-retest reliabilities, have also yielded tal functibps, themselves. That is, either favorable correlatians. the perceiving (SN) fuinction or the judg- Most assessments of the Indicator 358 Recent Assessment of the Myers-Briggs Type Indicator have examined characteristics of each of results. Kaiser (198 1) compared catego- the four scales separately. Carlyn's re- ry types based on the shorter form to view of the MBTI (Form F) literature those on Form G and found that type through 1975 reported tetrachoric coef- classification was unchanged in 91% to ficients for split-half reliability ranging 96% of the cases using the shorter ver- from .66 to .92 and concluded that "Es- sion. However, when scores themselves timated reliabilities of tvpe categories were examined. the onlv dimension that appear to be satisfactory;h most cases" did ,lor show significantchanges was the (Carlyn, 1977, p. 465) irrespective of the EI scale. Therefore, Kaiser concludes scales examined. Computations based that for nominal type the shorter version on continuous scores are higher, as of the Form G is congruent with the might be expected, since the conversion standard Form G for all categories; but of score to type categories reduces infor actual scores reflecting strength of formation. Carlyn (1977) also reviewed type, the shortened version of the MBTI test-retest reliability studies and con- is unreliable in at least three categories. cluded that scores of college students In a nicely conceived study, Howes appear to be "reasonably stable" over and Carskadon (1979) examined the re- time in terms of type category, while liability of Form G as a function of both older subjects'scores appear to be even time and mood changes. In each of two more stable. Again in this case, analyses administrations at five week intervals, of continuous scores have typically yield- subjects were given a bogus "Self-report ed higher coefficients, with Pearson rs Inventory (SRI)" designed actually to ranging from .69 to .83 across two produce either mood elevation, mood month periods. depression, or no mood changes (con- Since Carlyr,'~review, several other trol). Among other questionnaires, the reliability studies have appeared. How- subjects were also administered the ever, reports of test-retest reliabilities MBTI. The SRI content from first to are sparse. Steele and Kelley (1976), as second administrations for the two part of a study reviewed below, reported groups was reversed from elevation LO test-retest reliabilities of the El, TF, and depression, and vice-versa. Significant SNscales on the MBTI ranging from .86 changes in moods on a "mood adjective to .89. Carskadon (1977) reported reli- questionnaire" were obtained consistent abilities ofcontinuous scores on Form F with the SRI manipulations. However, across an eight week interval. For fe- reliabilities of test-retest continuous males, test-retest rs ranged from .73 to scores on the MBTI ranged from .78 to .87 on the four scales; for males, rs .87 and showed no significant differ- ranged from .56 to .79. In both cases, the ences as a function of the mood manipu- lowest coefficients were obtained on the lations. An additional analysis of type TFscale, a result that had been reported reliability showed that 19% of subjects by others earlier(Stricker& Ross, 1964), changed type on the EI scale, I I% on but not obtainedinarecentstudy(Cars- SN, 17% on TI;: and 16% 011 JP, and kadon, 1982). tha~the greater the preference score, the Carskadon (l979c) also reported test- lower was the likelihood of type change retest reliabilities for the short Form G upon retest. of the MBTI. Across seven week inter- Very recently, McCarley and Cars- Downloaded by [KSU Kent State University] at 04:54 15 June 2016 vals, the four scales yielded rs ranging kadon (1983) again examined reliabili- from .48 to .84 for continuous scores. ties for the four subscales of Form G. Carskadon also compared these coeffi- Across all items (five week retest peri- cients with those obtained in his study od), reliability coefficients for continu- using Form F (1977) and found no dif- ous scores were high, ranging from .77 ferences. (TFscale) to .89 (JPscale). Interesting- A study of reliability of a still shorter ly, however, the percentage of subjects version of the M BTI, consisting of the who retained their specific dichoromous first 50 items of Form G, yielded mixed type preferences across all four scales JOHN G. CARLSON 3 59 was only 47%. In other words, a subject sures "some attribute or quality (con- who had, say, an ESTJ preference on struct) that people are presumed to pos- first testing had only a 50-50 chance of sess" (Brown, 1976, p. 128). It seems maintaining the identical preference on premature to consider research on the every one of the four subscales upon MBTI as more than just tentative be- retesting, It was additionally found that ginnings on the establishment of the these results applied equally well to type "nomological net" (Cronbach & Meehl, of item, that is, the phrased questions or 1955) that underlies the test. Jung's the- the word pair items used in the MBTI, as ory itself is often vague and concepts well as to split-half assessments ("X- ill-defined. Moreover, translations of half' and "Y-half" forms within the the theory as well as attempts to opera- MBTI). The authors regard the former tionalize it in the form of the MBTI as important since they view word pair often leave out conceptual steps that items to be more valid than phrased would allow for clear empirical assess- questions, whereas the X-half, Y-half ment of related constructs. Neverthe- comparison is aimed simply at consis- less, the rlesearch reviewed here can be tency of the test. seen as important in the early stages of Finally, the reliability of a Spanish development of useful measurement of translation of the MBTI, Form G, was Jung's constructs as translated by the recently reported by Levy and Padilla MBTI. The degree to which attempts to (1982), Beyond the fact that the test was empirically validate the test are success- in Spanish, the population of Puerto ful will no doubt influence the likeli- Ricans from which the subjects were hood that the nomological net will gain drawn is also unusual in research with wider acc~eptance(Brown, 1976). the MBTJ, Test-retest reliabilities were A brief summary of evidence with re- high, ranging from .79 (TF scale) to '$9 gard to content and predictive validity (JP scale). The; r values may have been of the MBTI was included in the paper positively i~fluencedby the short, two by Carlyn (1977). In addition, approxi- weak, test-retest interval in this study. mately 50 papers, many of them unpub- In summary, published studies on re- lished, in the application of the IdBTI in liability of the MBTI are relatively few clinical arnd counseling settings were in number, However, the studies that summarized by Carskadon (1979b). The are available show stitisfactory internal following review focuses on published consist$ncy of each of the four scales studies relating the MBTI 1.0 other as- and, w~tbthe possible exception of the sessment devices and to a variety of be- TFscale, satisfactory stability of scores haviors in various clinical, counseling across several months. The latter gener- and research contexts, alization is particularlly true of nominal type. In one study, test+retestreliability Intertest correlations was u~zaffectedby verbally induced mood Wakefield, Sasek, Brubaker, and Fried- change^. At this time, there is a need for man (1976) correlated the MBTI with more reliability assessment, especially Eysenck's Personality Questionnaire across longer periiads of time and a wid- (EPQ). The latter contains scales for er variety of test conditions. In addition, psychoticism, neuroticism, extraversion, since mast studies have sampled only and lying. The only significant correla- Downloaded by [KSU Kent State University] at 04:54 15 June 2016 college students, there is a need for ss- tion obtained was between the EI scale sessment with a wider range of subject on the MBTI and the extraversion scale populations. on the EPQ. The authors also reported nonsignificant but moderately positive Construct Validity rs between the TFscale and neuroticism By far, the largest number of recent (a result paralleling that of McCary & reports on use of the MBTI have been Grant, 1976, using another anxiety test). with respect to its construct validity, Steele and Kelley (1976) speculate that is, the extent to which the test mea- that considerable overlap exists between 360 Recent Assessment of the Myers-Briggs Type Indicator the personality dimensions assessed by those who, when combining scores on the Eysenck extraversion scale and the the conflict-mode tests, were integra- EI index of the MBTI. In this studv, as tive. assertive. and coooerative. Neither in that by Wakefield et al. (1976), the El the SN or JP scales ;ere significantly scale was found to correlate significant- correlated with the conflict tests in this ly with Eysenckysextraversion factor (r analysis. The authors conclude that the = .74). In addition, SteeIe and Kelley Jungian dimensions are useful "in doc- argue that their data provide an assess- umenting and explaining psychological ment of the discriminant validity of the bases of interpersonal behavior" in con- other scales on the MBTI. That is, the flict situations (Killman& Thomas, l97sS, remaining scales failed to correlate sig- p. 979). nificantly with one another or with any Carskadon and Knudson (1978) re- of the scales of the EPQ. ported correlations between the MBTI A result that may not be expected on and 0. J. Harvey's "This I Believe" test. intuitive grounds was reported by Eliot The latter is used to classify individuals and Hardy (1977) in a study of the rela- primarily with respect to emphases up- tionship between EI preference on the on concreteness and autonomy in be- MBTI and "internality" on Rotter's havior. At one end of a four-step conti- (1966) locus of control scale. An internal nuum, System I individuals have "low- orientation on the locus of control test is er" conceptual systems, tend to agree said to be characteristic of individuals with external standards, and tend to be who attribute control of significant out- relatively resistant to expressions of per- comes of behavior to their own behav- sonal autonomy. At the other end of the ior. An external orientation typifies an continuum, System IV individuals use individual who attributes control of "higher"conceprual systems are open to reinforcers and punishers to outside for- expressions of autonomy andlor multi- ces or people (Carlson, 1982). Eliot and ple alternatives for action (Carskadon & Hardy argue that internals should tend Knudson, 1978). The researchers specu- to be more extraverted since, theoreti- lated that the SN scale also assesses the cally, they have experienced more suc- tendency not to go beyond the objective cess in attaining reinforcers as a func- facts (sensing types) or to trust largely tion of their behavior in social contexts. indirect and personal sources of infor- The results of the study support this mation (intuitive types). The results fa- analysis since extraverts on the MBTI vored this analysis since sensing types ob:ained significantly more internal were overrepresented in the System I scores on Rotter's test. category and intuitive types were over- Apparently, very few attempts to cor- represented in the Systeni 1V category. relate the other three scales on the Kelley's Role Construct Repertory MBTI with other tests have been made. Test and the MBTl were administered In one exception, an application of the by Carlson (1980) to adult subjects in M BTI to modes of conflict handling was one of three studies (also discussed in proposed by Killman and Thomas (1 975). the next section). The former test is de- Collegestudents were administered three signed to assess "indi\iduals' construc- tests designed to assess an individual's tion of their personal worlds." Carlson

Downloaded by [KSU Kent State University] at 04:54 15 June 2016 methods for handling conflict (the di- hypothesized that the SX scale on the mensions being competing, collaborat- MBTI would differentiate subjects in ing, compromising, avoiding, or accom- terms of responses on Kelly's test be- modating). When the MBTI was admin- cause intuitive types are more likely istered, a preference for feeling on the than sensing types 10 develop personal TF scale was generally significantly re- constructs that are "inferential" and go lated to accommodation in a ccjnflict beyond the basic concrete facts. The re- situation as measured by the conflict- sults strongly favored this analysis. "Six- mode tests. In addition, extraverted in- teen of the 20 intuitivzs, compared with dividuals on the EI index tended to be only 3 of the 14 sensing types, offered a JOHN G. CARLSON 361 preponderance of inferential constructs" methodological problems. In the pres- on Kelly's tests, a highly significant ef- ent paper, only recent published studies fect (Carlson, 1980, p. 808). Other MBTI are cited and, in the main, these are scales were not predictive in this study. more positive in outcome as well as meth- Finally, Padgett, Cook, Nunley, and odologically sound. Carskadon (1982) generated a number The MBTI in treatment settings. of clear predictions concerning the rela- Among studies of the MBTX in clinical tionship between androgyny, as mea- settings, Dewinne and Johnson (1976) sured by the Bem Sex Role Inventory, suggested that in the Jungian-defined and type preferences on the MBTI. extravert there is an attempt to suppress Three of their predictions were con- infantile and primitive demands due to firmed in a group of male and female cultural pressures. In addition, the ex- _ college students. It was found that an- travert is said to be impulsive and to be inattentive to the possible effects of be- droevnous- w* males tended to arefer feeling, whereas androgynous fe'males pre- havior on the body. Therefore, there is ferred thinking on the TF dimension. some theoretical basis for the likelihood The El scale predicted significantly for that the extravert may turn to drug women only; mare androgynous women abuse for the relief of cultural pressures. were extraverts than introverts. Also, in Dewinne and Johnson assessed the type men both SF and NF preferences were characteristics of drug addicts and did more cloaely associated with androgyny, find a significantly larger percentage of whereas ST and NT preferences were extraverts than introverts in their small associated with sex-typed responses. The sample. authars conclude that the METI may Tlais outcome might lead to a number add information to sex typing that will of hypotheses concerning various other aid clinicians in identifying the flexible indices of psychopathology and type on and adaptive behaviors that charaoter- the MBTI. For example, in Dewinne ize both andrageny and mental health. and Johnson's interpretation of Jung, Clearly, however, considerably more introverts are said to be shy, distrustful, data is needed before the value of the defensive:. su'saicious. and more likelv MBT1 in making such predictions is es- than extraver& to be subjecn to neuroses tablished. characterized by fear and anxiety, ob- sessive-c~ampulsivereactions, feelings of Criterion-Related Studies inadequacy, and the like. 1r1 one related The remaining studies in this review study, however, a negative outcome was have attempted to relate types on the obtained in an attempt to predict neuro- MBTI to a variety of behaviors in sis and psychosis fro111 the MBTI. No treatment and resear~hsettings. It may significant correlation between intro- be noted that following Carlyn's (1977) version and extraversion on the Elindex genleral review of a number of earlier and clinically diagnosed neuroticism or published studies it was concluded that psyohoticism was found by Hughes and "The Indicator appears to be a reason- Johllson (1975). Note that this result al- ably valid instrument which is potential- so conflicts with the Wakefield et al. ly useful for a variety of purposes" (p. (1976) report of a correlation between 471). A more negative conclusion was the EIscale and test neurosis, cited in an

Downloaded by [KSU Kent State University] at 04:54 15 June 2016 more recently reachled in Carskadon's earlier section of this paper. (1979b) review of the MBTI as applied In a very large study comparing psy- in clinical and couriseling settings in chiatric patients with earlier studies on which problems in statistics, interpreta- "normal" populations, Bisbee, Mullaly, tion, magnitude af differences, and fail- and Osmond (1882) found many signifi- ures to replicate are cited. However, it cant typle preferetlces. The Iresearchers should be noted that this review in- compared patients in four groups: de- cluded a large numbier of unpublished presdon, schizOphwnia, substance abuse, theses and dissertations fraught with and bipolar-manic disorder. Across all 362 Recent Asses3 went of the Myers-Briggs Type Indicator four groups, ISFJ, ISFP, and ISTJ with intuitive types than when paired types were overrepresented relative to with sensing types. Similarly, on the JP the normal populations. As a group, scale, judging types reported fewer prob- substance abusers tended to be more ex- lems when paired with judging types and traverted than the other patient groups. significantly greater satisfaction in mar- In the schizophrenics, no types were sig- riage than all other combinations of nificantly overrepresented by compari- type. By contrast, perceiving types paired son with the patient groups, suggesting with judging individuals reported the to the authors that this disorder occurs most problems. These and a number of without reference to type and thus is other relationships found by Sherman biological in origin. Based upon these demonstrate the clear, although com- and a number of other results in this plex, utility of the MBTI in one aspect of study, the authors conclude that knowl- counseling work - identifying sources edge of type is essential to providing of relationship problems. appropriate treatment. However, it is The MBTIin research settings. In this clear that considerable research yet re- research context, five recent reports on mains in order to clarify the actual utili- the MBTI are notable. Among them, ty of the MBTI in the treatment of psy- Carlson (1980) outlines studies with col- chiatric disorders. The Bisbee et al. re- lege students typed on the MBTI and sults merely demonstrate potential dif- subjected to various tests of memory ferences among disorders in terms of and social perception. (See also the dis- type preferences. The differences have cussion of one of these studies in the no clear theoretical bases and, therefore, reliability section above.) In the first no particular usefulness in the estab- study of the series, subjects were asked lishment of a nomological net. More- to describe personal and critical life in- over, while the comparisons between cidents relevant to seven basic emotions. patient groups were sound, those be- Essentially, extraverts remembered more tween the groups and earlier "normal" incidents that occurred in social con- populations lack the credibility of more texts, whereas introverts remembered traditional control methods, that is, more emotional events occurring in in- control groups within the study itself. dividual contexts. Also, feeling types In a counseling setting, Sherman (198 1) reported significantly more intense emo- demonstrated several clear correlations tions ofjoy, excitement, and shame than between MBTI-assessed type and prob- thinking types. In another study, adult lems in intimate relationships in an im- subjects were asked to imagine that they pressive analysis of 167 couples. Not were writinga letter to an unknown per- surprisingly, the fewest relationship son in another country and attempting problems were reported by individuals to provide a self-description. Most of similar to one another on all four MBTl the intuitive subjects but hardly any of scales. However, the next lowest catego- the sensing subjects made reference to ry of couples' problems was among in- the imagined other person. Conversely, dividuals with completely opposite ty- a significantly higher proportion of sens- pology (ESTJ males paired with INFP ing types made reference to the physical females). Among other findings, all males appearance of the other person. Carlson (extraverted or introverted) reported concludes that the SNscale successfully

Downloaded by [KSU Kent State University] at 04:54 15 June 2016 fewer relationship problems with intro- differentiates subjects who will directly verted females. However, extraverted participate in imagined events from those females reported significantly fewer who do not, and it differentiates subjects problems with extraverted males than who use concrete self-description from with introverted males. For either sex, those who do nor. In passing, it may also sensing types reported significantly few- be noted that a high degree of discrimi- er couples problems when paired with nant validity was also established in sensing types. lntuitive types, however, these studies since the only predictive reported more problems when paired MBTI scale in each study was the one JOHN G. CARLSON 363 originally hypothesized to be relevant to the SN, TF reversal and the completely the manipulations. opposite type as "very true of me." In In one creative experiment, Carskad- short, subjects were generally good at on (1979a) gave subjects who were dif- selectingthe type description that matched ferentiated as extraverts and introverts their MBTI preferences and at recogniz- on the MBTI an opportunity to present ing that the opposite type was unlike a five-minute talk in front of judges on them. the subject of political candidacy. As In a somewhat more complex study, hypothesized, extraverts stood closer to Cohen, Cohen, and Cross (1981) com- the judges, spent less time in silence dur- pared self-ratings and ratings of spouses ing the talk, and remembered more of on Likert scales corresponding to defini- the judges' names later than did intro- tion~of type (Myers, 1962) to actual verts. However, extraverts did not use MBTI types. By contrast with Carska- more gestures and expressive behavior don and Cook's ((1982) resusts, the au- in their presentations. thors found no relationship between self- In a recent repsrt, hlatthews, Miller, ratings of type and the MBTI preferen- and Carskadon (1981:) examined con- ces. Similarly, the subjects' ratings of formity in judgment of line length in their perception of their "ideal" selves group Gontexts as a function of type on showed nlo relationship to hdBXI type. the MBTP. It was hypothesized from However, except on the JP scale, the earlier studies (e.g., Cooper & Scalise, spouses' ratings of the subjects' types 1974) that extraverts would be more corresponded significantly with the likely to conform to group influences MBTI types. The authors argue that the than introverts. Also, prior evidence failure to obtain ti relationship between suggested that feeling types would be subjects'ratings of ideal selves and MBTI more conforming than thinking types. preferenc~eswas an indication of the Confederates of the experimenters par- "divergent validity" of the MBTI. That ticipated in line-length judgmentk (par- is, the MBTI does not measure ideal or alleling the classic Asch conformity perceived socially desirablr: self~percep- procedures) in an attempt to influence tions but rather actual type preferences. statements concerning comparison lines Further, the fact that spouses redog- relative to standard lioes. The results nized the subjects' type on most scales were: complex but esstntially supported demonstrates the construct validity of the hypatheses: IN types conformed the the MBTI: that is, its measures corres- least of all of the groups and EF and ES pond with the psychological reality of types together tended to conform more others' assessments of the individual. than IT and IN types. This experiment should weigh heavily Finally, in two studies investigators with those who are concerned with the have compared types as measured by the fundamental psychological usefulness MBTI with subjectshnd others' de- offthe Intlicator. scription of their type. Carskadon and Cook (1982) examined the ability of Conclusions their subjects to describe their own type As noted earlier, both internal and in terms of their selection of four one- test-retest reliability of both Forms F page type descriptians. Based upod the and G of the MBTI have proven satisfac-

Downloaded by [KSU Kent State University] at 04:54 15 June 2016 results of the MBTI administered two tory in recent assessments, with r values months before, the subjects were each af individual scales often exceeding 30. given packets of self-descriptions that The greatest need for reliability assess- were either their actual type, their type ment at this time appears to be studies with SM and TF scales reversed, or the with a greater diversity of test condi- complecly oppasite type. Fifty percent tions and populations since many of the of the subjects rated the actual type de- existing studies utilized testing envi- scription as "very true of me,"whereas ronments in university settings and stu- only 10% and 13%, respectively, ranked dents as subjects. Moreover, longer test- 364 Recent Assess 'merit of the Myers-Briggs Type Indicator retest intervals should be sampled in fu- research on the Indicator, such as has ture research. characterized, for instance, development Validity of the MBTI remains in great- of the MMPI. Perhaps future efforts to er question than reliability. Although validate the Indicator in treatment and relationships between the Indicator and research settings will incorporate more other tests have generally supported systematic approaches and thereby help hypotheses concerning underlying theo- tighten the nornological net, reti,cal overlap, five of the eight studies reviewed in the intartest correlations References section concentrated upon the EI scale of the instrument. However, it is notable that the three more recent studies cited have found evidence of validity of some of the remaining scales when correlations with other tests were performed. Future studies should give relatively greater attention to these other scales. Eight other reports of attempts to correlate the MET1 with a variety of behaviors in treatment and research settings were also generally positive with respect to the hypotheses tested, with some exceptions that were noted in the criterion-related studies section. It is to the credit of the MBTI that the instrument successfully predicted behaviors as far ranging as couples problems to story imagery and groupconformity. However as in the case of correlations with other tests, the emphasis in several of the studies cited in this section was upon the EI scale, leaving the question of validity of the bulk of the items on the MBTI still open to more thorough examination. Fortunately, there appears to be a trend in more recent studies to examine the validity of the other scales. Perhaps the present review will encourage more research of this type in treatment and research settings. Also a greater variety of validity research with less restrictive populations, that is, outside of the university, and by investigators with little vested interest in the instrument would be a great advantage. The current ten-

Downloaded by [KSU Kent State University] at 04:54 15 June 2016 dency for a relatively small number of researchers to conduct a large proportion of the research in this area leaves the generalizability of some of the findings yet to be established. Finally, while for construct validation a wide variety of research has clear utility, at the same time there has been a notable lack of systematic programs of JOHN G. CARLSON 365

interaction of Jungian typology and conformity. Myers-Briggs Type Inventory and of criteria for Journal of Personality and Social Psychology, clinical interpretive hypotheses involving them. 29, 566-57 1. Research in Psychological Type, 6, 24-36. Cronbach, L. J., & Meehl, P. E. (1955). Construct McCary, P. W., & Brant, W. H. (1976). The rela- validity in psychological tests. Psychological tionship bet ween trait anxiety and Nlyers-Briggs Bulletin, 52. 28 1-302. personality - a question of validity. Colorado Dewinne, R., &Johnson, R. W. (1976). Extraver- Journal of Educational Research, 15, 28-30. sion-introversion: The personality characteris- Myers, I. 0. (1962). Manual: The Myers-Briggs tics of drug abusers. Journal of Clinical Psy- Type Indicator. Palo Alto, CA: Consulting Psy- chology, 32, 744-746. chologists Press. Eliot, J., & Hardy, R. C. (1977). Internality and Padgett, V., Cook, D., Nunlsy, M., & Carskadon, extraversion-introversion. Acrceplualand Motor T. (1982). Psychologioal type, androgeny, and Skills, 45, 430. sex-typed roles. Research in PsychoJogiral Type, Eysenck, H. J. (1973). Eysenck on extraversion. 5, 69-88. New York: Wiley. Rotter, J. (1966). Generalized expectancies for Howes, R. J., & Carskadon, T. G. (1979). Test- internal versus external control of reinforce- retest reliabilities of the Myers-Briggs Type In- ment. PsychologicaiMonographs, 80(1) Whole dicator as aEunction of mood changes. Research No. 609. in Psychological Type. 2, 67-72. Sherman, R. G. (1981). Typology and problems Hughes, R. C., & Johnson, R. W. (1975). Intro- in intimate relationships. Research in Psycho- version-extraversion and psychiatric diagnosis: iogical Type, 4, 4-23. A test of Eysenck's hypothesis. Journal of Clin- Steele, R. S., L% Kelly, T. J. (1976). Eysen~kper- ical Psychology, 31, 426-427. sonality questionnaire and f ungian Myers-Briggs Jones, J. H., L Sherman, R. G. (1979). Clinical Type Indicator correlation of extraversiion-intro- uses of the Myers-Briggs Type Indicator. Re- version. Journal of Consulting and Clihical search itt Psychological Type, 2. 32-45. iDsychoiogy,44, 690-69 1. Jung, C. G.(1923). Psychological types. London: Stricker, L. J., & Ross, J. (1964). An assessment Routledge L Kegan Paul. of some structural properties of the Jungian Kaiser, K. M.11981). Use of the first 50 items as a personality typology. Journalof Abnormaland surrogate measure of the Myers-BriggsType In- Social Psychology, 68, 62-71. dicator. Research in Psychological Type, 4, Wakefield, J. A., Sasek, J., Brubaker, M. L., & 55-61. Friedman, A. F. (1976). Vdlidity of Eysenck's Killman, R. R.,& Thomas, K. W. (1975). Inter- Personality Questionnaire. Psyciiological Re- personal conflict-handling behavior as reflec- ports, 39, 115-120. tions of Jungian personality dimensions. Psy- choZogica1 Reports. 37. 971-980. Levy, N., & Padilla, A. (1982). A Spanish trans- latiom of the Myers-Briggs Type Indicator Form G. Psychological Reports, 51, 109-1 10. Matthews, 5. K., Miller, D. I., & Carskadon, T. G,(1981). Conformity behavior in an experi- John G. Carlson mental sitnation as a function of psychological Dept. of Psychology type. Research in Psychological Type, 4.62-70. Universitv of Hawaii 2430 ~an;~usRd. Mattooa, M. (1981). Jungian psychology in per- Honolulu, HI 96822 s~eckive.New York: The Free Press. ~i~arle~,N., & Carskadon, T. G. (1983). Test- Received: March 24, 1984 retest reliabilities of scales and subscales of the Revised: January 23, 1985 Downloaded by [KSU Kent State University] at 04:54 15 June 2016