138

97. Eysencx, S. B. G., anp Eysenck, H. J. ‘“‘On the DualNature of Extraversion.’’ Brit J Social & Clin Psychol 2:46—-55F ’63. * (PA 38:916)

98. Eysencx, Sysit B. G., anp Eysencx, H. J. ‘‘TheValidity of Questionnaire and Rating Assessments of Extraver-sion and Neuroticism, and Their Factorial Stability.” Brit JPsychol 54:51-62 F ’63. * (PA 37:8009)

99. Forrest, D. W. ‘‘Relationship Between Sharpening andExtraversion.”’ Psychol Rep 13:564 O ’63. * (PA 38:8488)

gga. Furneaux, W. D. “Neuroticism, Extraversion, andSuggestibility: A Comment.” Int J Clin & Exp HypnosisI1:201-2 Jl ’63. *

100, Furnraux, W. D., anp Linpant, L. E. H. “HowValid Are Questionnaire Validity Studies’ Abstract. B BritPsychol Soc 16:19A Ap ’63. *

tor. HESELTINE, G. F. “The Site of Onset of Eczema andPersonality Trait Differences: An Exploratory Study.” JPsychosom Res 7:241-6 D ’63. * (PA 38:8360) —

102, HiLtcarD, ERNEST R., AND Bentier, P. M. “PredictingHypnotizability From the Maudsley Personality Inventory.”Brit J Psychol 54:63-9 F ’63. * (PA 37:8091)

103. Howarts, E. “Some Laboratory Measures of Extra-version-Introversion.”” Percept & Motor Skills 17:55-60 Ag"63. * (PA 38:7242)

104. Kiss—EN, Davip M. ‘‘Personality Characteristics inMales Conducive to Lung Cancer.” Brit J Med Psychol36(1) :27-36 °63. * (PA 38:921)

105. LirTLe, ALan. “Professor Eysenck’s Theory of Crime:An Empirical Test on Adolescent Offenders.’’ Comments byH. J Fysenck. Brit J Criminol 4:152-63 O °63. * (PA 38:6477

106. Lovisonp, S. H. ‘‘Conceptual Thinking, Personality andConditioning.’ Brit J Social & Clin Psychol 2:100-11 Je ’63. *(PA 38:3654)

107. McGuire, R. J.; Mowsray, R. M.; ann VALLANCE,R. C. “The Maudsley Personality Inventory Used With Psy-chiatric Inpatients.”” Brit J Psychol 54:157-66 My ’63. *(PA 38:1032)

108, MANNE, SicmMunD H.; KanvEeL, ARTHUR; AND ROSEN-THAL, Davip. ‘‘The Relationship Between Performance MinusVerbal Scores and Extraversion in a Severely SociopathicPopulation.” J Clin Psychol 19:96—-7 Ja ’63. * (PA 39:1676)

109. Mrzry, A. G.; Couen, SamMueu I.; anp Knicut, E. J.‘Personality Assessment Under Varying Physiological andPsychological Conditions.”” J Psychosom Res 7:237-40 D ’63. *(PA 38:8527)

110. Ratu, R., AND Misra, S. K. “‘Change of Attitudes asa Function of Some Personality Factors.’’ J Social Psychol60:311-7 Ag ’63. * (PA 38:4182)

111. Ropinson, J. O. “A Study of Neuroticism and CasualArterial Blood Pressure.”’ Brit J Social & Clin Psychol 2:56-64 F ’63. * (PA 38:1347)

112, RUTTIGER, KATHERINE Forp. “Individual Differencesin Reaction to Meprobamate: A Study in Visual Perception.”’J Abn & Social Psychol 67:37-43 J] ’63. * (PA 38:419)

113. SINGER, JEROME L., anp AwntTrRosBus, Joun S. “AFactor-Analytic Study of Daydreaming and Conceptually-Related Cognitive and Personality Variables.” Percept &Motor Skills 17:187-209 Ag ’63. * (PA 38:7418)

114. SINGH, S. D. “Extraversion, Neuroticism, and Con-formity Behaviour.” J Psychol Res 7:66-71 My ’63. * (PA38:4253)

115. STANLEY, GorDON. ‘‘Personality and Attitude Charac-teristics of, Fundamentalist Theological Students.’ Austral JPsychol 15 “121-3 Ag 63. * (PA 38:6037)

116, Tauss, W. “A Note on Stability and Equivalence ofLong and Short Forms of the M.P.I.” Austral J Psychol15:118-20 Ag °63. *

117, WENABLES, ETHEL. ‘‘Personality Scores and Achieve-ment Among Technical College Students.” Abstract. B BritPsychol Soc 16:58-9 Jl ’63. *

ARTHUR R. JENSEN, Associate Professor ofEducational Psychology, and Associate Re-search Psychologist, Institute of HumanLearning, University of California, Berkeley,California.By all criteria of excellence in test develop-

ment the MPI is an impressive achievement.It has grown out of years of intensive re-search on the dimensional analysis of person-ality. A great amount of evidence (36) hasshown that two relatively independent super-factors, identified by Eysenck as neuroticismand extraversion-introversion, represent mostof the variance in the personality domain.

THE SIXTH MENTAL MEASUREMENTS YEARBOOK [ 288

While it is possible to slice the variance inthis domain into many different ways of mak-ing up “scales” consisting of various combina-tions of many kinds of personality inventoryitems, these scales are almost always highlyintercorrelated, despite their widely differinglabels. When they are factor analyzed, eitherat the scale level or at the item level, the firsttwo or three independent factors almost in-variably account for all the appreciable com-mon factor variance in the lot. The MPI hasbeen developed to measure two of the mostcomprehensive factors, Neuroticism (N) andExtraversion (E). Neuroticism refers to gen-eral emotional instability, emotional overre-sponsiveness, and predisposition to neuroticbreakdown under stress. Extraversion refersto outgoing, uninhibited, impulsive, and so-ciable inclinations. The method of developingthe inventory was factor analytic and is ade-quately described in both the British andAmerican editions of the manual.The MPI consists of 48 items, of which 24

are keyed to N and 24 to E. Unlike some per-sonality inventories (e.g., the MMPI), noneof the items could be construed as sociallyobjectionable; thus the inventory can be usedwith adolescents or adults in almost any set-ting. Though the MPI takes only about 10 orI5 minutes, there is also a short form—de-scribed in the British manual and by Jensen(6)—consisting of six items from each scale.The short form has satisfactory reliability andhigh correlations with the total scales and canbe useful when time is very limited.The MPI derives much of its importance

from its theoretical underpinnings. Probablyno other psychological test—certainly no otherpersonality inventory—trivals it in psychologi-cal rationale. This is particularly true of theE dimension, which has been the subject ofintensive experimental research in Eysenck’slaboratory for more than a decade. A reviewof this research is, of course, impossible here.The manual prepared by Robert Knapp forthe American edition has a bibliography of112 items of the most relevant literature, and

the manual itself summarizes much of the pub-lished findings. Factor-analytically sophisti-cated readers are also referred to Carrigan’s(32) critical appraisal of E as a dimension ofpersonality.

NoRMS. A great deal of normative data arepresented, both for English and American sub-

289 |

jects. The American manual presents Ameri-

can college norms (percentiles and stanines

based on 1,064 university undergraduates).

Means and standard deviations are presented

for 32 different groups, including various psy-

chiatric, prison, and industrial populations,

totaling over 7,000 subjects (including the

American norms group of 1,064 and the Eng-lish norms group of 1,800). Bartholomew (83)

has published some Australian norms, which

differ little from the English, except that theAustralians seem to be slightly more extra-

verted, as are the Americans.

There are slight sex and social class differ-

ences on both the N and E scales; these are

fully discussed in the manual. The scales arenot correlated with intelligence.

RELIABILITY AND VALIDITY. Split-half andKuder-Richardson estimates of item intercor-relations for each scale are between .75 and

.gO in various samples. N consistently has

slightly higher internal consistency than E.

Test-retest reliabilities range from .70 to .go.

In short, the reliability of the MPI 1s among

the highest to be found for personality inven-tories. The MPI has also been studied for

effects of various types of “response set.”

These seem to be negligible.Assessment of the validity of the MPI is a

complex matter. There can be little questionof its factorial validity. That is to say, theN and E scales invariably have high loadingson factors that are also heavily represented inother measures considered to be indicative ofneuroticism or extraversion, and there is little

factorial overlap between the scaies. Thoughthey were intended to be completely indepen-dent measures, it has been found that they arecorrelated about —.15—slightly more or lessdepending upon the population sampled. Thenegative correlation is somewhat higher (usu-ally about —.30) in psychiatric and collegepopulations. Data on correlations with otherpersonality inventories are presented in themanuals. Note, for example, that the N scale

correlates almost as highly (.76) with theTaylor Manifest Anxiety Scale as reliabilitywould allow. There is, however, a_ slightlygreater negative correlation between the Tay-lor scale and the E scale than between the N

and E scales.

Descriptive validity of the MPI has beenadequately established by the method of nomi-nated groups. Judges rated people on the basis

TESTS & REVIEWS: CHARACTER—NONPROJECTIVE 138

of observable characteristics in terms of neu-roticism and extraversion. These ratings show

highly significant correlations with the relevantdimensions measured by the MPI.Most important, but also the most difficult

to evaluate, is the “construct validity” of theMPI, that is, the elaborate network of theory,

predictions, and experimental findings concern-ing the N and E dimensions. Adequate dis-cussion of this topic must presuppose thereader’s knowledge of Eysenck’s theory ofpersonality which relates neuroticism to auto-nomic lability and extraversion to cortical in-hibition.t Since an exposition of the theoryand the related research is beyond the scopeof this review, the reviewer can only give hisoverall impression of this vast body of workas it relates to the MPI. First, there is no

doubt that both N and E scales have shownsignificant and replicable correlations with ex-perimental phenomena in the fields of percep-tion, motor learning, verbal learning, paintolerance, and attitudes. Some of these rela-

tionships are predictable from Eysenck’s andothers’ theories. All of the research, of course,

has not unequivocally supported Eysenck’s

theoretical deductions and there is a largefringe of ambiguity on the growing edge ofthe theory which is perhaps somewhat under-

emphasized in the MPI manuals. It is this areaof far reaching, but as yet inadequately sub-stantiated, implications of the theory that hasprovided Eysenck’s critics with an easy targetfor their often premature unfavorable evalua-tions. But if one reviews the research of theMaudsley group over the years, it is clear thatthe theory of personality associated with theMPIis sensitive to experimental findings andis constantly undergoing careful modificationand development. It seems to be Eysenck’s per-sonal style, more than the facts of the matter,which stimulates criticism and a counsel of

caution, since Eysenck tends to stride each stepof the way with a rather bold assurance. Allin all, it seems safe to say that no other per-sonality test is based upon a body of psycho-logical theory so far reaching and sodiligentlyand ably researched as is the MPI. The chiefreason for this is that the MPI is one of thefew personality measures that has grown outof a theory concerned with basic psychological

1 Eysencx, H. J. The Dynamics of Anxiety and Hysteria:An Experimental Application of Modern Learning Theory _toPsychiatry. London: Routledge & Kegan Paul Ltd., 1957. Pp.XIV, 311.

138

processes rather than out of purely empiricalattempts to predict certain currently practicalcriteria.

USES OF THE MPI. The MPI has beenlittleused in clinical diagnosis. It is not listed inSundberg’s? survey of the 62 most widelyused tests in clinical practice in the UnitedStates. The reasons are not hard to find: Clini-cians generally want more detailed informationthan is provided by a subject’s scores on twobroad dimensions of personality; the MPI di-mensions do not correspond at all well to thepresently used diagnostic categories (nor arethey intended to), and the psychological theoryassociated with N and E hasnot been generallyincorporated in diagnostic or therapeutic prac-tice. Those who wish to see how the theoryunderlying the MPI is related to psychiatricdiagnosis and therapy are referred to a dis-cussion by Eysenck.? As yet this reviewer hasnot seen evidence of the practical use of theMPIin clinical settings. Certainly it is not ofany value for conventional psychiatric diag-nosis. McGuire and others (707) gave theMPI toan unselected group of psychiatric pa-tients and found that the N scale differentiatedall diagnostic groups from the nonpsychiatriccontrols, but neither the N nor the E scaledifferentiated significantly among the diagnos-tic groups. Other studies have shown signifi-cant differences among various diagnosticcategories, but these differences have not beensufficiently reliable to support the use of theMPI for individual diagnosis. Since in theMcGuire study all psychiatric groups averagedIO-15 points higher on the N scale than thenormal controls, it is suggested that the MPImight be valuable as a psychiatric screeningdevice.

Also, for screening and group prediction ineducational and industrial settings, the MPIshows promise based on research. College ex-amination failure rate and academic achieve-ment, for example, have been shown to berelated to N and E in ways predictable fromEysenck’s theory. Persistence in menial andmonotonous tasks also is related to the MPIdimensions.The present reviewer has had most experi-

2 SuNpDBERG, Norman D. “The Practice of PsychologicalTesting in Clinical “Services in the United States.’’ AmPsychol 16:79-83 F ’61. *

3 Eyvsenck, H. J. Chap. 3, “‘“A Rational System of Diagnosisand. Therapy in Mental Illness.’?’ In Progress in Clinical Psy-chology, Vol. 4. Edited by Lawrence E. Abt and Bernard F.Riess. New York: Grune & Stratton, Inc., 1960. Pp. ix, 181. *

THE SIXTH MENTAL MEASUREMENTS YEARBOOK [ 290

ence with the MPI as an adjunct to laboratoryresearch in the field of human learning. TheMPI can be used by experimentalists who be-lieve personality factors may play a part inthe psychological phenomena under investiga-tion and who wish to account for more of the“between subjects” variance as a means of in-creasing the precision of experiments. Therelevance of anxiety in learning and condi-tioning experiments, for example, has beenamply demonstrated with research using Tay-lor’s Manifest Anxiety Scale. The N scale ofthe MPI can serve the same purpose as theMAS, with the added advantage that it isshorter, more reliable, and has a greater bodyof psychological research behind it. It has beenfound that the importance of the neuroticismfactor increases as task complexity becomesgreater (74). We have also found in our ownwork that subjects with high N scoresare lessapt to stand up well throughout an arduouslaboratory experiment and are less able to fol-low complex directions in an experiment, eventhough they may have high intelligence. Therelevance of E to experimental variables,though called for by Eysenck’s theory, is notso clearly established at present and mustawait further investigation. But it is in therealm of experimental psychology, as a co-variate in studies of perception, conditioning,learning, persistence, attention, concept attain-ment, and the like, that this reviewer sees themost immediate potential usefulness of theMPI. The American manual also discusses theuses of the MPI in market research and invocational selection and counseling.A word about the British and American edi-

tions of the manuals. Both cover the essentialsexpected of any test manual, but the Americanedition is more up-to-date and therefore morecomplete in its coverage of relevant research.Indeed, it is an exemplary model of what atest manual should be.

In summary, the MPI is a brief and highlyreliable measure of tworelatively independentbroad factors of personality—neuroticism andextraversion-introversion. Much sophisticatedresearch has gone into its construction, and thelarge body of normative data, plus the psycho-logical theory and experimentation associatedwith the MPI, make it one of the most impor-tant of all personality inventories, and cer-tainly the preferred measures of neuroticism(or anxiety) and extraversion.

291 |

The American edition of a new version of

the MPI, called the Eysenck Personality In-ventory (EPI), has been published by the

American publisher of the MPI. The EPIis described in a preliminary edition of themanual (August, 1963) as an attempt to make

the MPI scales more useful for certain pur-poses. The EPI measures the same two factors

as the MPI, but the slight correlation that

exists between N and E in the MPI scaleshas been removed entirely, by adding, sub-

tracting, and rewriting items and subjectingthem to repeated factor analyses. Also, many

of the items have been reworded in such away as to increase their reliability when used

with subjects of low intelligence or little edu-

cation. There are two equivalent forms of the

EPI. The EPI also contains a “lie” scale(borrowed from the MMPI), a worthwhile

addition if the inventory is to be used for

screening or selection purposes where subjectsmight be inclined to “fake good.” For experi-mental work the “‘lie’ scale is usually super-

fluous, however. The reliability of the EPI

scales is slightly higher than for the MPI and

the normative data for the English population

are quite adequate. American users will haveto develop their own norms until such databecome available. For experimental use withcollege subjects the EPI does not seem tooffer many substantial advantages over theMPI (unless one insists on eliminating theslight correlation between N and E or wishes

to do a retest on an equivalent form) and ithas the slight disadvantage of being more timeconsuming, since it contains 9 more items thanthe MPI. Further research should make pos-sible more valid and detailed comparisons be-tween the MPI and EPI. Potential usersshould, of course, examine specimen sets of

both the MPI and the EPI to decide whichinventory might best suit their purposes interms of the available norms,etc.

JAMES C. LincorEs, Associate Professor ofPsychology, The University of Michigan, AnnArbor, Michigan.The Maudsley Personality Inventory (MPI)

is a theoretically based instrument designedto measure the two rather pervasive and rela-tively independent personality dimensions ofextraversion-introversion (E) and neuroticism-stability (N) found by Eysenck and others in

TESTS & REVIEWS: CHARACTER—NONPROJECTIVE

138

a large number of factor analytic studies. The24 items for each trait were selected on the

basis of both item and factor analyses as beingthe purest questionnaire measures to date ofEysenck’s factors.The MPI is an easily administered, quick,

reliable, and fairly simply scored test. Withthe following exceptions, the manual to theUnited States edition is commendably success-ful in meeting the various criteria of technicalexcellence stipulated in the APA TechnicalRecommendations. Some minor criticisms ofthe manual are: (a) that the professionalqualifications necessary to administer and in-terpret the test are omitted, and (0) that false-positive and false-negative rates are missingin the discussion of validity by nominatedgroups. Significant mean differences are in-sufficient to assess properly the value of suchstudies. More serious deficiencies in the manualare: (a) the omission of tables of item inter-correlations and factor loadings as well asother item statistics, essential ingredients forfactor based scales; and (b) the very inade-quate delineation of the N factor as a descrip-tive or clinical concept to aid the user interpre-tatively in the individual case, one of the im-portant recommended uses of this test. Theuser should satisfy himself on the above pointsby referring to the relevant literature listedin the 112-item bibliography of the manual.Of more crucial concern to the prospective

user are the following observations regardingthe test itself and its relationships with othertests purporting to measure the sametraits or

factors.First, the MPI is not a general personality

test, even though the traits it assesses accountfor most of the variance in personality inven-tories. One should not confuse statistical sig-nificance with clinical importance, as Eysenckhimself would acknowledge. Consequently, ifone is looking for a more complete personalityprofile on a subject, other tests would be morepertinent, e.g., the Guilford-Zimmerman Tem-perament Survey or Cattell’s i16PF, beinglogical choices among factor based tests, or theMMPI, being the best among clinical personal-ity instruments. As a clinical tool the MPIwould serve best in an ancillary role, supple-menting data from other tests. A two-dimen-sional approach to personality is insufficient toencompass all the functions and purposes

138 THE SIXTH MENTAL MEASUREMENTS YEARBOOK [ 292

typical of the average clinical setting, no mat-ter how pervasive or important the factorsmaybe.

Second, it has not been established that evenfor the traits that the MPI measures, it meas-

ures them better than comparable instruments.It should be noted that E is only one of severalkinds of extraversion, 1.e., social, and a num-

ber of studies reported in the manual indicatethat this trait is measured at least as well by

other tests as it is by the MPI if one takes

the internal consistency data as a yardstick.Thus, E correlates to the extent of .81 with

social extraversion from the Minnesota T-S-EInventory and —.80 with Heron’s introversionscale. These validity coefficients lie within therange of Kuder-Richardson and split-half re-

liability coefficients for E, i.e., between .75 and

85 with the majority above .80, and they are

certainly higher than the equivalent forms

coefficients (.75) reported for the Eysenck

Personality Inventory. Furthermore, based

upon the original sample of 400 cases used in

the item analysis of the MPI, E correlated .79

with the rhathymia scale of the Inventory of

Factors STDCR and N correlated .g2 with

the cycloid disposition scale, the latter coefh-

cient being higher than the split-half or K-R

reliability coefficients reported for N.

Admittedly, other factors must be considered

in a consumer’s decision to use one test as

opposed to others, e.g., cost, ease of admin-

istration, ease of distortion or deception, pro-

fessional time spent in interpretation, the pur-

poses to be served by testing, readability of the

items, appeal to the examinees, etc., and while

some of the foregoing would favor the MPI,

comprehensiveness may well determine the

choice in the final analysis given equally good

data on reliability and validity. |

In conclusion, the present evidence on the

MPI would suggest that there was little reason

to omit in the American manual the caution

expressed in the original British manual, Le.,

“Tn all its applications, the M.P.1. should pri-

marily be regarded as a research instrument.”

Within Eysenck’s theoretical system, the MPI

and its revision, the Eysenck Personality In-ventory, may well indeed be the tests of choice,

but more evidence is needed on superiorrelia-bility and validity to warrant their supplantingother comparable and better established tests.

WILLIAM STEPHENSON, Professor of Psychol-

ogy, Umversity of Missouri, Columbia, Alis-SOUT.The MPI is excellently produced. The

American manual is especially informative andcomprehensive, listing 112 references up to theend of 1962. The American norms are for1,064 university and college students. Valida-tion is with respect to mean differences forgroups of subjects (sample sizes range fromas few as 8 to as many as 1,800 andtotalsome 7,200) variously described as Australianprisoners, psychopaths, industrial apprentices,psychosomatics, hysterics, English normals, re-cidivist prisoners, neurotics, dysthymics, etc.,mostly in Britain.

It would not be difficult, in the present re-viewer's judgment, to find other compilationsof personality statements which, when sub-jected to such gross validation procedures,would fare no better or no worse than the setof 48 put together for the MPI. It is possiblethat they may be useful in experimental studiesusing samples of the order 100 to 1,000 per-sons. What is not so certain is the credibilityof the data the test provides. The public iswarned in this respect. But there is an issuewhich, it seems to the reviewer, requires con-

sideration as psychology grows professionally.After a very careful review of Eysenck’smajor work,? Storms and Sigal (9) have toconclude that the attributes of extraverts andintroverts listed by Eysenck in certain studieshave not, in fact, been unequivocally demon-strated. Doubt was raised about the validationof the E continuum.The reviewer would raise again the improba-

bility that a scale based on R-methodologicalgrounds can ever really indicate dynamic con-ditions such as Eysenck has persistently pro-posed to examine. Davis,? for example, re-minded us, and Eysenck in particular, thatfollowing a traumatic situation immediate reac-tions to the situation were apt to be ones ofoveractivity, or of psychological withdrawal;

subsequently, recovery from the shock wasattended by preoccupation and fixation ofmemories, with the establishment of defenses,

with the abandonment of defenses, and with

a phase of working through the memories.

1 Eysencx, H. J. The Dynamics of Anxiety and Hysteria:An Experimental Application of Modern Learning Theory toPsychiatry. London: Routledge & Kegan Paul Ltd., 1957. Pp.X1V, 311.

2 Davis, D. RussEti. ‘Clinical Problems and ExperimentalResearches.” Brit J Med Psychol 31:74-82 pt 2 58. *

293 |

These, it seems to the reviewer, are typicalof human reactions: to imagine for a momentthat either E or N in general have anythingto contribute to such a flow of phenomenaseems to the reviewer to be clutching at feath-ers in the wind. Moreover, there is a simpleway to show that dynamic factors can in factbe brought to light, using Eysenck’s 48 state-ments but in the form of a Q-sample so thateach person can use the statements relative toone another to display fixations, defenses, etc.

The E and N scales never do anything of thekind because they are by definition measure-ments of behavior im a general context. Theproof of the matter, that at least one factorcommon to Q couldn’t possibly appear in dataderived from R (and vice versa), is there forEysenck to note. It is astonishing that so dili-gent a worker has not looked to see what thatone factor, at least, could mean for his studies.

Puitip E. Vernon, Professor of EducationalPsychology, Institute of Education, Universityof London, London, England.

Despite the enormous number of availablepersonality inventories, Eysenck’s test couldwell meet a need for a short, simple instrumentfor use in mental hospitals, in student counsel-ing, and in a variety of experimental researcheswhereit is desired to control major personalitydifferences among the subjects. Only 48 itemsare included, selected on a factorial basis togive highly saturated measures of extraversion-introversion and neuroticism-stability or anx-iety. Reasonable Kuder-Richardson and repeatreliability coefficients ranging from .75 to .85for Extraversion and from .85 to .go forNeuroticism are obtained in I0-15 minutes’testing time, and the two scores are virtuallyuncorrelated except in certain selected groups.The scoring of the American edition can bedone by punchedcard or, in less than a minuteper blank, by stencils; the British edition isscored by transparent stencil.

Since his first book in 1947, Eysenck hasstressed the pervasiveness of these two per-sonality factors, and in The Structure of Hu-man Personality (36) he makes a strong casefor reducing most of the manifold factors thathave been claimed in questionnaire data, rat-ings, and objective personality tests, to thesesame dimensions. Much as Spearman, Burt,and the present writer prefer to cover as muchvariance as possible in abilities by means of g

TESTS & REVIEWS: CHARACTER—NONPROJECTIVE 138

and major group factors, and to regard Thur-stone’s, Guilford’s, and other multiple factorsas minor subdivisions—so Eysenck considerspersonality as hierarchically organized, withthese two factors as the most inclusive. More-over, during the ensuing 16 years, he has linkedthese with a nomological network based onHullian learning theory, and collected a con-siderable amount of experimental evidence tosupport his theoretical deductions covering ex-traversion and neuroticism, albeit many ofthese theories and experimental results areopen to dispute. He can reasonably claim,therefore, that scores on this inventory possessa good deal of construct validity derived frompositive experimental findings in the field ofconditioning and the effects of drugs, fromfactor loadings, from differentiation betweensuch pathological groups as psychopaths anddysthymic neurotics on the one hand and neu-rotics and normals on the other, and fromcorrelations with other well known tests ofrelated constructs.Some commentis called for on the extraver-

sion measure. In many of his writings Eysenckhas criticized the American conception of ex-traversion as consisting largely of sociability,and the consequent tendency for extraversiontests to give rather high negative correlationswith neuroticism or emotional instability. Hefavours, rather, Guilford’s notion of rhathymia,or uninhibited carefreeness, as being orthogonalto neuroticism and closer to Jung’s originaldescription. However, the definition in themanuals of the present test, together with manyof the test items, clearly involves the socialaspect of extraversion; the highest correlationof the extraversion scale with another test is81 with the social introversion—extraversionscale of the Minnesota T-S-E Inventory.Indeed the fairly good reliability for so shorta scale may be largely due to the reiteration ofquestions about social mixing. The content ofthe neuroticism items is, however, more varied.

In the American manual, percentile normsfor American college students (one collegeonly) and tables of group means and standarddeviations are given. The latter reveal interest-ing differences. Thus on Extraversion, psycho-paths average 31 (out of a possible 48),American women students 29, English stu-dents and normal adults around 25, hysterics24, and dysthymics and neurotics 19. OnNeuroticism, the means of psychopaths and

138

neurotics mostly fall in the 32-38 range, Eng-lish students 23-27, American students 20-21,and English normal population 18-20. Negligi-ble relations are reported with sex, social class,and intelligence, except that men are slightlymore extraverted than women and women areslightly more neurotic than men. Among Eng-lish college students both introversion andneuroticism correlate appreciably with academicachievement. One would have thought that thiscould be more simply explained in terms of theweaker gregarious interests and greater intro-spectiveness of the serious student than bymeans of learning theory constructs.

Responses indicative of extraversion may beeither Yes or No, but all neurotic responsesare Yes. The author of the American manualdrawsattention to the possibility of acquiescentresponse set affecting the latter, but dismissesit. No mention is made of the effects of socialdesirability. The American manual is well de-signed, with due attention to the APA Com-mittee’s suggestions. But the manual to theoriginal British edition, published in 1959, 1s

much more brief and should be brought up todate.

In general the test should be of some use in

educational guidance and personality counsel-

ing as a quickly obtained index of two impor-tant personality trends. It could be given in

mental hospitals by nurses as a preliminary

aid in psychological assessment, or included ina battery of tests for surveying a population,for example, in market research or, as already

indicated, in experimental researches with nor-

mal adult subjects.

Brit J Psychol 51:185-6 My ’60. A. Bursill.[Review of the British manual.] This Manualreports up-to-date information available on thisYes, No, ?-type questionnaire (MPI) com-prising two scales of 24 items each, one pur-porting to measure neuroticism (N), the otherintroversion-extraversion (/-E). * The scales

can be conveniently adapted to form an even

shorter questionnaire (SMPI) comprising sixitems each for N and J-E, simply by utilizingthe first page of the printed form only. * Thetwo scales N and /-E intended to be orthogonal,have a low correlation (—o.15 for the MPI,

and —o.05 for the SMPI) for normal samples—the correlation increases to the rather unsat-isfactory dimension of —o0.3 to —0.4 in neu-

rotic groups. Eysenck assigns these anomalies

THE SIXTH MENTAL MEASUREMENTS YEARBOOK | 294

to the non-linearity of the regression lines. Theargument is supplemented by a graph of re-gression lines for 1,200 normal subjects whichdoes not show a serious state of affairs exceptf-scores below about 10. But the explanationas it stands is hardly sufficient to give rise tosuch large negative correlations between Nand /-E in neurotic groups, whose mean scoreslie well within the distributions given for thewhole range of the population. This needs in-vestigating in more detail. Meanwhile, whenattempting to assess the effects of varyingdegrees of extraversion in any experiment,Iysenck suggests matching criterion groups off and & for N. A table containing the size ofsamples, mean scores and S.D.’s for the dif-ferent standardization groups is given—but thereader is left to work out the significance ofthe differences. Unfortunately, the originalsources of muchof the data are omitted, orare still not available, so that the procedureswhereby subjects were selected cannot be as-certained in detail. Two methods of validationare presented: (1) comparison of the stand-ardization groups on N and J-E; (ii) constructvalidity—i.e. a set of interlocking predictionsforming a theory confirmed by experiment. Ina strict sense, neither method can yet be saidto have reached satisfactory standards in em-pirical confirmation. * unusual answers to sometwo to three items are sufficient to place anindividual amongst the most extreme group ofdysthymics on J, whereas some ten unusualanswers are required to place an individualamongst dysthymics on N. At the very leastthere is some reason for attempting—in subse-quent versions of the test—to stretch the /-Fdimension somewhat. But the reviewer is notcertain whether the data cannot be taken asundermining one of Eysenck’s basic tenets,which is not merely that hysterics are moreextraverted than dysthymics but that they arealso more extraverted than normals. In viewof Hildebrand’s similar findings with objectivetests (Brit. J. Psychol. 1958, 49, I-11), thereis an increasing likelihood that the position ofthese various abnormal groupsis a true featureof this J-E dimension, and not some distortionin its scale units—particularly at the E end.In fact, Eysenck hints at this situation in theManual and elsewhere, without explicitly rec-ognizing that it contravenes his and Jung’stheoretical position. A factor to be taken intoconsideration, however, is that presumably this

295 | TESTS & REVIEWS: CHARACTER—NONPROJECTIVE 138

scale measures sociability rather than the otherfacets of extraversion emphasized by Jung—

since J-E correlates highly with the “social”

scale and lowly with the “thinking” and “emo-

tional” scale of the Minnesota TSE. Anotherfeature worth noting is the limited range ofitems, many of which closely overlap in con-tent. There might here be a tendency to sacri-fice validity for reliability. Possibly a source

of distortion on the N-scale, on the other hand,

resides in the fact that in all items neuroticresponses are scored in the affirmative (yes).Space does not permit an appraisal of Eysenck’sattempt to demonstrate “construct validity” inhis Dynamics of Anxiety and Hysteria, as

claimed in the Manual. However worthy thisattempt was, there has been a growingtide of

criticism (e.g. by Storms and Sigal, VernonHamilton, D. E. Broadbent, R. L. Reid, Taylor

and Rechtsscaffer, Spivac and Levine) of theevidence presented for the various components

of the theory in this monograph. Eysenck’s

adroit defences do not altogether dispel these

criticisms. Consequently, the reader may haveto exercise caution in accepting the constructvalidity as indeed valid. Generally, Prof.Eysenck is to be congratulated on obtainingsuch an unusual amount of data on one per-sonality test. It is to be hoped that a moredetailed Manual will soon make its appearance;

some of the original work is published inrather inaccessible foreign journals. * There isthe danger that subjects selected on an N andI-E basis alone will unduly bias and filter the

human material and cause much of importanceto the clinician to be omitted. *

J Consult Psychol 23:563 D ’59. Edward S.Bordin. [Review of the British Edition.] This

brief questionnaire of 48 items and its evenbriefer short form (12 items) have played anintegral part in the author’s well known re-search on personality. Many instruments haveyeen launched for full scale use with much lessehind them. The manual represents the height

English diffidence. Only the briefest sum-Mary is given of a few of the salient results ofresearch and the reader is referred to therelevant publications. He is told that “theM.P.I. should be regarded as a research in-strument. Different firms, organizations, hos-pitals, universities, and other bodies have dif-ferent problems, deal with different samplesof thle population, and aim at different solu-tions bf their problems. Only applied research

can determine whether instruments such as the

M.P.I. can be successfully used by them, andjust what form such use can best take.’ No

high powered American merchandising here!

J Psychosom Res 5:66 S ’60. G. A. Foulds.[Review of the British Edition.] * The stand-ardization data call for some comment. Neu-rotics were diagnosed by experienced psychia-

trists, or else had their case-papers carefully

scrutinized by three experienced clinical psy-chologists, who arrived at a unanimous diag-nosis independently. In his reply (/. abnorm.soc. Psychol. 1958, 57, 2) to the paper bySigal, Starr and Franks (bid) Eysenck ratherdeplores the latter method. It is unfortunate

that their somewhat conflicting results werenot available for the Manual, since the claim

that “successive samples from different hos-pitals showed great stability in means and

variances” might have required some modifi-

cation. Eysenck believes that the results ob-tained on the M.P.I. “in a sense....serve asvalidation of the scales.” This, unfortunately,

can only be in an illogical sense. It 1s notpossible to validate the theory and the inven-tory at the same time. It would be palpably

absurd to claim—and certainly Jung did not—that Hysteria and Extraversion are one and thesame thing. What Jung said, in effect, was

most neurotic extraverts have the character-

istics of Hysteria; most neurotic introvertshave the characteristics of Dysthymia. A dem-

onstration of differences between Hysteria andDysthymia does not necessarily tell us anything

at all about extraversion:introversion. Thedifferences in the particular instance may be

due to quite other characteristics. The M.P.1.

has, of course, considerable face-validity for

at least some aspects, particularly social, of

that elusive concept extraversion. It is doubtfulwhether reliance on “construct validity” is ofany value whenthere is a large logical hole in

the nomological network. In respect of the

extraversion :introversion continuum, the posi-tion of the recidivist prisoners and the psycho-somatic cases is close to the hysterics, afinding which to Eysenck is not unexpected.The reviewer would have expected recidivistprisoners to be closer to hospital psychopaths

than to hysterics. With regard to the psycho-somatic group, at least one large sub-group

consists of people whose intense affective dis-turbance has resulted in physiological changessuch as are rare in hysteria. If Stanley Cobb’s