ED 045 687

DOCUMENT RESUME

24 TE 49S 853

AUTHOF Loveless, Eugene J.TITLE Artistic Preferences, Conceptual Thinking and

Intellectual Attitudes. Final Report.INSTITUTION Notre Dame Univ., Ind.SP0N5 AGENCY Office of Education (DHEW), Washington, D.C. Bureau

cf Research.BUREAU NC ER-8 -E-027PUB DATE Feb 70GRANT CEG-0-8-080027-3723(010)NOTE 44p.

EDRS PRICEDESCRIPTORS

EDRS Price MF-$0.25 HC Not Available from EDRS.Ability Identification, *Art, Behavior RatingScales, Cognitive Ability, Cognitive Measurement,*College Students, Divergent Thinking, Forced ChoiceTechnique, *Individual Characteristics, Painting,Personality, Personality Studies, *PersonalityTests, Psychological Testing, *Student Reaction,Test Construction

ABSTRACTSlides of great modern art works by Picasso,

Mondrian, and Van Gogh were selected as stimulus materials for an arttest administered to a heterogeneous group of 395 college students.The slides were endorsed for desirability, and the results werefactor analyzed. Scales of items with similar levels of endorsementfrom different factor scales were placed in a paired-comparisonforced choice task and administered to college men. Attempts tovalidate the test across a wide variety of tasks, such as divergentthinking tests and personality tests, met with marginal success. Thedevelopment and utilization of a free choice form of the artpreference test with the same pictures as stimuli and with similarsamples of men taking some of the same tasks as were completed bygroups along with the fcrced choice form produced a greater numberand greater magnitudes or significant correlations. Each of the artscale subtests did not conclusively constitute the analogue for eachof the cognitive styles as hypothesized, but the instruments seem aviable entree into the study of the relationship of art preferenceand personality. [Not available in hard copy due to marginallegibility cf original document.] (Author/ME)

U.S. DEPARTMENT OF HEALTH, EDUCATIONWELFARE

OFFICE OF EDUCATIONTHIS DOCUMENT HAS BEEN REPRODUCEDEXACTLY AS RECEIVED PROM THE PERSON ORORGANIZATION ORIGINATING It POINTS OFVIEW OR OPINIONS STATED DO NOT NECES-SARILY REPRESENT OFFICIAL OFFICE OF EDU-CATION POSITION OR POLICY.

(41

Do

7)

Final. Report

Project No. #1.42460.4.3.6,8-6-- o 47

Grant No. OEO-0-8-080027-3723 010)

Artistic Preferences, Conceptual Thinking and

Intellectual Attitudes

Euggre J. Loveless Ph.D., Principal Investigator

University of Notre Dame

Notre Dame, Indiana 46556

February, 1970

U.S. DEPARTMENT OFHEALTH, EDUCATION, AND WELFARE

Office of EducationBureau of Research

The research reportedherein was performed pursuant to a

GRANT with the Office of Education, U.S. Department nf

Health, Education and Welfare. Contractors undertaking

such projects under Government sponsorship are encouragedto express freely their professional judgment in the

conduct of the project. Points of view or opinions stated

do not, therefore, necessarily represent official Office

of Education position or policy.

CONTENTS

Page

Summary

Introduction 1

Methods and Procedures ... 5

Results

Conclusions and Recommendations 11

Appendixes

A.--List of Pictures, Factor Loadings, AuthorsSection for Title Al

B.--The Development of a Forced Choice Measureof Preference for Modern Art Styles andValidation with Personal Needs B1

C.--Art, Politics and Sex; A Preference forModern Art Cl

D.--Dimensions of Preference for Modern Art;(A Preliminary Study) Dl

E.--List of Measures and Subscales Utilized inthe Study El

F.--Tables I - IK of Correlations of the ArtTest Subscales and External Criteria Fl

References 13

Bibliography 14

SUMMARY

Slides of great works of modern art were selected as stimulusmaterials for an art test. The works were primarily those of Picasso ineach of his periods with additional works of Mondrian and Van Gogh. Theitems were endorsed by a heterogeneous population of college students(N=395) for desirability (i.e. liked or disliked). The results werefactor analyzed. Scales of items with similar levels of endorsementfrom different factor scales were placed in a paired comparison type offorced choice task and administered to college men. Attempts to vali-date the test across a vide variety of tasks including divergent thinkingtests, cognitive style measures, personality tests (HIT, OPI, PRP,Harvey's Conceptual Systems Test) and ability measures met with m-rginal

success. The development and utilization of a free choice form of theart preference test with the same pictures as stimuli and with similarsamples of men taking some of the same tasks as were completed by groupsalong with the forced choice form produced a greater number of signifi-cant correlations and greater magnitudes for the correlations.

It could not be concluded that each of the art scale subtests con-stituted the analogue for each of the cognitive styles as hypothesized,but the instruments seem a viable entree into the study of the relation-ship of art preference and personality.

CHAPTER I

INIRODUCTION

It is a central theme of current thinid_ng that creativity, innovaetion, diver Bent thinking, or just openminiedness, represents the productof individual differences in cognitive style (Klein 1949, Gardner,Holzman, Klein, Linton, Spence 1959), perceptual style (Barron 1952), orintellectual organization (Gardner, Jackson, Messick 1960). Since thecreative process is dependent, at least in part, upon an,interest in newideas and a tolerance for new representations of familiar objects,may be that the personality organization of the individual operates toinfluence the preferences people have for art vrorks. These works mayvary in the directness with which they convey the ideas embodied within

- them. The persistence required of a subject to find the meaning embeddedin a complex or abstract representation of an idea may be indicative ofcharacteristics which distinguish the _analytic student froM the conforming students and the creative from the pedantic.

. Individual differences. in the likes and dislikes people havetowards the form or style in which information is conveyed to them inpaintings or other media may reasonably be expected to influence theiracceptance or rejection of an idea. It is likewise true that a ceztaintolerance for differences is necessary for creative -thought. This "re-gression in service of the ego", which may be marked in peoples' atti-tudes towards modern art forms, may more subtly manifest itself in thestyles people employ in more commonplace acts of thinking and perceiv-ing, such as are involved in the .measures of cognitive styles. The re-lationship between art preferences and cognitive styles may represent acommon product of training, personality and cognitive attributes nhichinfluence students' attitudes towards new ideas; towards differences inthe type of information they retain; the manner in which they retain theinformation; and the value they place upon new information, ideas, orunusual thoughts.

The art test is seen as a new tool for categorizing the differenceswhich exist in people as a result of habits of selective attention a ndconceptual organization. In this research the use of artworks is hypo-thesized to operate at what psychoanalysts would call the preconsciouslevel of stimulation. In dream studies (Fisher and Paul 1959) and sub-lind.nal perception studios (Spence 1962); and most recently, studies ofperemptory ideation (Klein 1967) there have been extensive suggestionsthat the personal perceptual or symbolic significance of the artisticform can be responded to by an onlooker with a compatible pattern of con-ceptual or cognitive style. 'Klein states, "By effectively resonating17ith the motivating fantasy of an artist, a formal product that gives tethe onlooker a freshened awareness of a reality hitherto obscured by hisprogmatically habituated schemata may come to life through transform.-tions of color, texture, depth, and pattern."

1

The use of both verbal and performance measures of personality pro-vides some variation in the content of the criterion measures to preventwhat Cattell has called instrumental variables, from indicating rela-tionships based on the use of only questionaire type, verbal instrumentsemployed rather than the hypothesized multi-modality constructs understudy.

These patterns of response (called cognitive control principles orcognitive styles by Klein) have been tied to the motivational charac-teristics of the person and particularly to the reality testing level ofthe personality. These styles are derived .from simple perceptual andpsychophysical judgments but have been sham to be stable and to haveuseful relationships to both the defensive and conflict-free spheres ofpersonality organization..

In theoretical terms the cognitive styles may be considered percep-tual attitudes which underlie an individvells interaction with any newintellectual tasks (Guilford 1967). As attitudes they to repre-sent the broadest possible set of measurable constructs available for theinvestigation of perceptual-cognitive tasks. Of special interest is therelationship between the cognitive control principle designated, toler-ance for unrealistic experiences and the preference for modern artworks.It may be hypothesized that this may be the "master control principle"underlying whether a person can accept and develop a liking for goodnon-representational, abstract or impressionistic styles and the othercontrol principles determine differences within the tolerant 'and in-tolerant groups of subjects.

The validation of the art, teat includes many measures of putativerelation to personal differences in cognitive style and artistic pre- .

ference was warranted. : .

.The psychological significance of a preference for modern art inall or any of its forms is that it may indicate an openness to look-ing at old or familiar ideas in a new may. Modern art may be construedas an attempt by the artist to communicate with the beholder in afashion -which distorts or accentuates some special characteristic of anobject or scene. It is by an accentuation of a comaionplace detail thatthe artist 3s creative in seeing something of special significance in astereotyped scene. This technique of. distorting the obvious or common-place is essential if the artist is to communicate some new insight to aperceiver accustomed to responding to familiar cues in a stereotypedfashion. It is a sequela of this postulation that some people, parti-cularly artists and other creative people, utilize perceptual and cogni-tive strategies which differ from the typical. Therefore, for the nerawareness of the artist to be communicated to the public, the receivingpublic must be willing or capable of decoding the message of the artistin his idiosyncratic format. The degree of distortion of the artnorkwill presumably interact with the tolerance level of the receiver toproduce a willingness to contemplate the artist's product or to rejectconsidering the product at all.

The literatare dealing with art in psychology goes back to suchtough-ndnded theorists as Pavlov (1950), who believed differences inartistic and literary types was a function of the CNN. However, contem-porary use of art corks as psychological tools has usually focused itsattention on creativity, and aesthetic preferences. Of interest in theliterature on creativity, summarized in part by Golann (1953), is thefrequent finding that differences in motivational, and particularlystylistic differences, rather than intellectual traits, distinguish thecreative person from his contemporaries. Barron drew upon the earlierwork of Burt (1933) and Eysenck (1940, 1941), which had employed art-works of multitudiroue authors and had evaluated the free preferencesof their subjects. This left uncontrolled, the type and quality ofpainting as well as the style of the artist. This procedure alsofailed to control for a general tendency to express like or dislike forpaintings, and was probably plagued by for reliabilities of measurement,using scores based upon single items.

Barron (1953) has shown that artists differ from non-artists intheir preference for complex compositions and for asymetrical figures.This resulted in the artists as a group expressing greater interest inand tolerance for innovating sensual and non-representational art forms.Eysenck concluded that there was a general factor of estheticism whichwas randomly distributed through the population. When the influence ofthe t factor of esthetic preference was removed, the remaining influencewas described as a bi-polar personality factor precuraing Barron'ssimple-complex dimension. The most recent venture in this area isChild's (1965) studies of esthetic competeace which compares expert andlay judgments both within and across cultures. Child found evidence fora general esthetic factor and for the influence of artistic background.and training and for the influence of cognitive styles. McVihinnie

(1958) summarized the research on esthetic preferences and concluded thatdespite the general variance of Eysenck's t factor, perceptual trainingwaq effective and suggested the need to explore the effects of variouscognitive styles on esthetic preference.

3

The research cited so fax has alluded to selected and circum-scribed cognitive styles such as complexity-simplicity and synetry:-asymetry. The study of cognitive styles by Klein, Gardner and Witldn,has greatly increased the sophistication necessary to isolate andmeasure cognitive styles. Klein 1 in particular has extrapolated frompsychophysical measures of cognitive styles to theoretical correlates inpaintings. The lee of a wide variety of paintfngs to represent abroader spectrum of cognitive styles should extend the work on estheticsbeyond research presently available.

. .

This :study, therefore, attempted-to (1) start from the past -litera-ture, (2) select a relevant theoretical framework (lie. cognitivestyles) for the selection of materials for creating a "new and comprehen-sive art test, (.3) validate the test againsJ. a mute modality, broadspectrum, base of concurrently appropriate measures, and 04) integratethe -findings and evaluate the end product of .the research to develop amulti-dimension test of artistic preferences.

1. Personal latter to Riley -W, Gardner and made available by the latterauthor for use in conceptualizing this study.

1

1

1

1

CHAITER II

METHOD FOR TIE CONSMUCTION AND VALIDATION OF THE ART REFERENCE TEST

-- The rationale and .steps appropriate to building the art test pro-ceed from tk following concise descriptinn by Jackson (1967) of testvalidation procedures for the Fersonality Research Form..

Substantive Definition of Variables,

ughe central guiding principle in the development of the FRF vasthat the validational process should be integrated with the entire pro-gram of test construetion, rather than being something tacked on whenthe test is in final form. This point of view has been ably expressedin Loevinger's (1957) classic monograph. Loevinger suggests that thevaliclational procedure can be divided into substantive, structural, andexternal components. Each of these componeni-s g inTest development, but no one is sufficient in itself. Ordinarily, thereshould be an orderly progression from one stage to the next, accordingto Loevinger. The substantive component of validity refers to the de-gree to which the items comprising a test reflect on theoretical groundsan appropriate univeree of content. Such a formulation bears importantsimilarities to certain conceptions of other theorists; particularly tothe hypothetical latent continuum described by Green (1954), to Lennon's(1956) forraulation of content validity, and to the theory of generaliz-ability of Cronbach, Rajaratnam, and. Glaser (1963) . But nowhere is therea more explicit statement on the importance of substantive considera-tions in test construction than in Loevinger's monograph. The Struc-tural component requires that the investigator make explicits prior -to

data gathering; a structural model to which he expects item responses toconform. For example, he might expect his test to conform to a homogeneity model, as exemplified by high estimates of internal consistencyreliability. The external component of validity refers to the degree towhich test variables correspond to non-test manifestations of the trait.One important characteristic of this correspondence is that ideally it .

should be highly selective. Thus, an affiliation scale should corre-spond substantially more closely to traits believed to reflect affilia-tion than those which do not (cf. Campbell and Fiske, 1959) . If this

condition is met, the test may be said to possess both convergent anddiscriminant validity."

(r) )

Thus the first step in selecting the material for the art test wasfinding a theoretical basis on which to select or reject a slide of agiven painting. This presumed that enough pictures in high qualitycolor slidee would be available. The theoretical frame work was that o2Klein and Gardner which posited a series of relatively discrete cogni-tive styles, which operated across sensory.modalities and representedthe autonomous functioning of the ego. As adaptive rather than conflict

5

laden. structures, the cognitive styles framaaork was appropriate to thecreative and awareness expanding purposes underlying the artworks to beselective.

To achieve the purpose of this study it was necessary that onlyworks of art which embody an artist's personal and conceptual idiosyn-cracies within their expressive style, be utilized. It was natural toturn to Picasso, who is such a prolific and innovative artist and whohas developed a great many styles which could ba utilized as analoguesof cognitive styles. In addition, it was hoped to extend the styles toinclude the works of Van Gogh and Mondrian. In practice, it becamenecessary to select a few works .of other artists (ref. Appendix A). Thelimitation of a style to one palnter enabled the selection of picturesto be consistent within the theoretical framework underlying the opera-tion of cognitive styles.

This could not be observed in all cases, but where works by otherpainters were siullar and produced high factor loadings on the samescale, they were included in the scale. Slides were selected fromcommercial slide libraries, but where there was not adequate numbers ofslides to permit at least fifteen slides for each initial scale (hopingfor at least ten items in the final scale), other sources were sought,The later included copying slides from printed plates in books.

The initial sample of slides was presented to a large number(N=395) of college students, primarily in general psychology and educa-tion classes, bat other classes where cooperation was offered, wereaccepted in the early stage of establishing the levels of preference foreach picture. Subjects were drawn from public and private colleges anduniversities. in Indiana and New York. Private colleges were bothchurch-related and independent and the sample included men. and women.The heterogeniety of the sample proved disadvantageous in that it pro-duced adverse effects on both the reliability of the final scales andthe stabilitSr of the factors across validation groups. The internal .

validity for each subscale was based on a factor analysis of the initialsample of endorSements.

The validation studied were completed on the man from the NotreDame (Indiana) Campus. The greatest number of students were freshmenbut approximately one-third were upperclassmen participating for creditas part of the requirements of general peeichology.,

The architectural students were freshmen who were requested toparticipate at the conclusion of a freshmen orientation lecture.

.Note: An extended description of the methods utilized are reported in

Appendixes B, C, and D.

6

CHAPTER

RESULTS

The results of the study can be reported in two sections. Thefirst section will deal with the internal validity of the test and thesecond section to deal with the concurrent or external validity.

Despite highly prorising results, based on the factor analysis ofthe early forms of the art test (seeAppendix B), the final formshowed high intercorrelations between many of the art teet subscales.

. To accomodate the capacity of the computer programs.. the .art test wasrun as if it were really two tests of five subscales each. The inter-correlations are shown in 'Eable VII (Appendix F) . In both sets of sub-

scales only one scale seemed to show a relative -freedom from interde-

pendence with the other art scales (i.e. 1111, ID) The effect of the high

intercorreletions is to attenuate any hope of differential assessment bysubscales. In part, the problem of high intercorrelation is a dileanaproduced by the design of the study. In selecting art works for, the

study, one criterion was the clarity with which an item conveyed itselfas an instrumental:Ity of a given cognitive style. This procedure does

not lend itself to<the usual psychos:uric desire for a gradient of moremodest differential capacity. The result appears to have produced acontrast effect which degenerates into the large general factor found in

earlier research which bears upon the degree of clarity, familiarity andfreedom from bizarreness of ,each picture.

A further difficulty associated with the use of the forced choice

Option is the problem of ipsatization. This problem has been discussed

comprehensively by Clemens (1966). The basic problem of ipsative

measurement is that the. results reflect intra-indiridual evaluations,

rather than inter-individual evaluations. Much of the research reportedhere and in the data tables in the appendix are the results of work com-

plated with the forced choice form of the art preference test.. It was

possible however, to develop a free choice form of the same test and tovalidate the results against ,soma similar variables used to validate the

forced choice form of the same scale. The striking results reported in

Table IX offer a comparison with the results of the forced choice format

reported in Table I, especially for the Barron-Welsh Figure Preference

Test .(FliFFT) in which only the revised art scale was used. The few com-

parisons available strongly suggest that future validation studies willprobably be most productive if the free choice form is used despite theobvious limitations and extraneous influence due to testing artifacts.

The reliabilities of each form of the art test seem acceptable yet

range from .56 to .83 with a median reliaeility of .66. In part, the

relatively for reliabilities may be due to the brevity. of the scales.

The expectation is that longer scales would yield even higher reli-

abilities. .

The external validation of the art preference scales began with aseries of hypotheses and the support or non-support Of the hypotheses.The hypotheses and general conclusions about their sustenance are foundin Table 10 on the next page.

Most of the validation studies were conducted with the forcedchoice format. The results appear in Tables I to V. In Table Inegligible relationships are found in several of Guilford's tests ofdivergent thinking; the cognitive style of conceptual differentiationas measured by the object sorting test; Barron's ego strength scale; andthe revised art scale of the Barron-Welsh figure preference test.Several low correlations were found with items selected from the Oberlintest of art. Bierits adaptation of Kelly's Rep test showed little corre-lation between.fle art test and complexity, while of the omnibus person-ality subscales, the esthetic and thinking inteovereion scales showedappreciable correlations. This selective validity indica-tied sore supportfor the premise underlying the use .of pictures for the development of atest. Both of the Scholastic Achievement Test Scores (C.E.E.B.) verbaland quantatative, produced low, and interestingly enoughs negativecorrelations with many of the art test subscales; a curious but signifi-cant finding. The dogmatism and opinionation scales of Rekeac4 andBudnerts scale of the tolerance or intolerance of ambiguity cognitivestyle shored negligible relationships to the art test.

In Table II Guilford's Possible Jobs Test showed several low corre-lations as did the art test with the Concealed Figures Test, a measureof the cognitive style of field-articulation. The use of the color wordas a measure of the cognitive style of constructed or flexible controlrevealed somewhat more correlations With the reading colors section ofthe measure than with the interference section which is the criterionmeasure of the cognitive control principle. Interestingly enough, thesize estimation tests a criterion measure of the cognitive style calledFocussingScanning, revealed several differentiating correlations as didthe error and lag scores of the cognitive style called Leveling andSharpening.

In Table I two aspects should be noted. The first is that despitethe high intercorrelations of the subscales, there is a great amount ofdifferentiation in the correlation of the art scales with each of thecriterion or validation measures. Secondly, while it was plausible andjustified to hypothesize the expected direction of each correlation, aprocedure which clearly justified the use of one tailed tests the morestringent level of significance was utilized. This latter point mayhelp to clarify the state of the many correlations which barely miss be-ing significant at least at the five percent level.

Table III reports the 'correlation of the lorced choice art prefer -.ence test with the revised art scale and Holland's test of preconsciousthought. The students were freshmen and the premise of the later find-ings utilizing Holland's scale of evidence in this table...

Table ID

Hypotheses

. 1. Subjects high on dogmatism will show greaterpreference for veridical painting.

Results

Not supported

2. Subjects high on dogmatism and creativity Not supported(divergent thinking) will shoo greater pre-ference for abstract art with symmetricalgeometric configurations, particuinrly-whenthey are "balanced".

3. Subjects high on openmindedness and creativity Not supportedwill show the highest preference for art formswith least veridical content or geometricalformats (e.g. surrealie41; impreosioniem, etc.) .

4. When the artistic desirability of each painting Supported foris held constant (by using different styles of esthetic andthe same artist, e.g. Picasso, equated for en- thinking in-dorsement as liked or disliked), the choices of trovereionthe subjects will be correlated with their scales of thechoices on an adjective preference scale for OPI

such qualities as simplicity - complexity, etc.

5. There will be no differences in art prefer- Not supportedences attributable to intelligence test scores. See Table

6. The art preference test subscalee will consti-tute analogues of Klein & Gardner's cognitivestyles measures. This will be indicated bycorrelation coefficients between the intercor-relation tables of cognitive styles measuresand art preference test subscales.

modest support(See Text)

7. The more abstract and conceptually oriented the Supported byperson is, the greater will be his tolerance H.T.T. Scoresfor varied styles of artistic expression. but not C.S.T.

Scores

8. That those subjects who show an increase intolerance for more representational paintingsupon retesting will have the more flexible,conceptual and openminded traits.

9

Not supported

Table IV reports the correlations of the art testwith the com-puter scored results of the Holtzman Inkblot Test, The H,I.T. vari-ables showing the greatest number of correlations in descending orderare colors shading, abstraction, popularity, hostility and anxiety --a truly mixed assortment. The correlations of colors shading,abstraction and popularity seem consistent with the nature of artpreference and the pictures of the art to et; however the role ofanxiety and hostility is not clear.

Table V finds more support for the relationship of art preferenceto cognitive styles. In. this larger sample there are a greater numberof significant correlations with Gough's rigidity scale and a repeatof Budner's tolerance or intolerance for ambiguity. Conceptual dif-ferentiation as measured by the object sorting test reveals a modestnumber of correlations once again. The Personality Research Form didnot produce a great number of correlations, yet appropriately enough,the sentience scale showed some correlations of significant magnitudebut the scale of cognitive structure produced only one marginal corre-lation to note.

The forced choice form of the art preference scale, despite itshigh intercorrelation did not produce consistent correlations underconditions where possible equivalence should have produted such re-suns, It is difficult to decide if the lack of consistent correla-tion across related scales is due to the ipsative nature of the scales,the heterogenietyof the base population from which the factoranalysis Vara' drawn or to a truly differentiating character of thescales. This problem will be discussed in the next chapter.

The free choice form of the art preference test revealed few re-lationships with any of the subscales of Harvey's Objective Test ofConceptual Styles (Table VI). It eras hoped that the abstractnessscale would be informative but the correlations were not forthcoming.

Tables VIII and IC are highly informative. On those tables,correlations between the free choice form of the art preference test arecorrelated with measures which were earlier correlated with the forcedchoice format with students from other samples. The results are notonly striking with regard to the magnitude of thoir relationship withthe concealed figures tests and color-aword test (measures of the cogni-tive styles of field articulation and constructed flexible control) butaleo because of the larger number of subjects utilized in the freechoice samples. For each of the previously reported measures and theA.C.E. test of Critical Thinking both the magnitude, frequency andselectivity (i.e. not uniform across all subscales of the art test) ofthe relationships are striking.

CHkIJTER

CONCI.USIONS AND RECOMSENDATIONS

The technical difficulties associated with developing a test utiliz-ing art won } -s on slides as stimulus itemo were several. Items which weseidentified by factor analysis to represent independent categories basedon the statistical analysis provided several highly correlated scales inthe final forms of the art tests. There mey be two explanations for thisturn of events. The effect of the general factor of esthetic preference,as an overriding influence on the art discriminations, may have been tosubvert the influence of the relatively subtle cognitive styles inherentin the pictures. Another possibility is that by employing pictures,which clearly represent a style of modern art, it was not possible to de-

. velop a sufficient gradient of difficulty to gain the desired compar-ability of art works. It is of value to note that despite the highintercorrelations of some of the art scales, the occurrence of significant correlationswas not uniform but was more selective. It would bedesirable from some points of view to have greater independence ofscales. It does not seem possible within any design which seeks tolimit the number of artists used as a source of the pictures. Furtherresearch may be able to diminish some of the overlap by .using a statistical procedure to partial out any peesible general factor which may beoperating.

. The internal consistency of the scales seems acceptable for mostuses. Reliability coefficients were computed for the fcrced choice formonly, but it scene likely that the free choice form would be at least as

It would be desirable to establish temporal reliability by atest-retest procedure but that was not possible in this study.

The study was originally based on the premise that a forced choice'format utilizing items paired for rated endorsement of desirability (i.eliked or disliked) would be the most effective vehicle for an instrumentof this typo. Much of the cumbersome work in the early stages of thestudy was devoted to the technical problems of selecting, rating, pair-ing and trying out pictures to be utilized in the study. Following the

test development stager the art test in forced choice form eras given tocollege students in conjunction with the measures of cognitive style andpersonality measures among several others. Since many cognitive styleprinciples must be derived from timo consuming individual test sessions,there was some lag in finding the hypothesized relationships and furtherthat few correlations were forthcoming between the art scales and othermeasures. It seem likely that the heterogeniety of the originalstandardization process, particularly the mixture of men and women's re-sults, led to unstable relations in the sLbscales insofar as the purityof the cognitive style was concerned.

Of greater moment was the utilization of the forced choice formatas the vehicle for the art test. While avoiding response bias, desir-

.11

ability and other artifacts, it became apparent that the ipsative pit-fall could not be overcome in the correlation table. This conjecturewas verified when a free choice format was constructed and new groupssimilar to that used to validate the forced choice format were giventhe new form and several related tests. The increase in both the fre-quency of significant correlations an.:, the magnitude of the correlations,despite larger nalmber of subjects, seemed to redeem some of the promisefor the art test lost in the validation of the forced Choice test.Phil o it was not possible within this study to replicate all themeasures used earlier, there does seem to be the promise in the freechoice format that was originally held for the forced choice test.

In conclusion, a relatively useful design for the study of art pre-ferences has evol7ved from this research. There is the need for creatingscales with increased independence. The use of free choice items whichpermit the use of item analysis techniques should provide scales whichare more independent. It seems useful to attempt to identify the influ-ence of general factors or moderator variables which contribute to theoverlap of scales and to eliminate these statistically. The validationof the final form against concurrent measures, criterion groups selectedfor viable characteristics and the comparison of the scales employed inthis test with homogeneous scales of other arts, would complete thiscycle of studies.

The hypothesized relationship of art preference to many socialattitudes could not be sustained. This finding may question some of thegenerality of attributes accorded the authoritarian personality, dogma-tiom and other attitudes. There does appear support for a broad generaltrait found previously in the literature by several authors that relatesto a preference for representational or non-representational art. Theart test subscales do not seem to constitute the analogue to cognitivestyles as was hypothesized. In part, this seems due to the scale over-lap. However, it does not appear likely that either form of the testwill produce the factor analyzed loadings on the same factor as themeasures of the cognitive style measures.

.12

Appendix A

Sid (3.0 Factorliura)er Facter. 1_,-,:-_n z ;II n tr Artist.....:.1= .4--2

1 - 313 22 Picasso .Harlequin and By 3.9052. ..- 113 19 Mondrian Breactway Beo.e Woogie 1923/1.-43., 1k 146 Picasso The Gourmet . 1901it. 1k 56 3,..):Icasso The Bathers

* 5. lA 60 Picasso Old Guitarist 19036, 3A 35 Picasso 'Pink Torso 1906 ,7.: : 1k 145 :- Picasso 'Acrobats with a Dog 1905 :.,8. 20. 31 :. Picasso Woman in Chemise ....,i9. 2B 21 Picasso Woman's Head 1903.

10. 1k 61 : - P; CaLirit) 'Harlequin's Family w. 3.905Monkey

*11. 2D '30 : Ri.casse Boy Leading A Horse 190532. Van Gosh Church at Auvers :189013.

.o4.233 132

2)3 69 icesz:o Woman in Blue 1902*15. aE 4o. ,:. Picasso16... 1317. 3E 29 Picasso Ma Jolie18. 2B 23 Unknown

*19.. 2E :78 Picasso Still Life with Antique 1925Head.

:190. 2E 73 'Picasso The Schoolgirl21. 113

. and Grapns22, 213 3it .:,.Picasso Violin .. 19i3

*2234.,.::..

3.D 31 .. Picasso Harlequin .. 19152D .36 ... , Picasso

.,,or.1 2D 58 :- Picasso Boy With a Cock

1.11999:3494....

,:t..,0

26. 20 36 Picasso The Table ..

27.*28.

.2D 32. Picasso Still Life :with Cake2D 69 . Picasso Man Smoking

*29. 2D .69 . Picasso Three Musicians 1921:*30. 2D ..r..; .55 .. Picasso Green Still Life*31 10 49 Picasso The Lovers 192332. 2D : . at , Picasso Women at Fountain 192033. 2B 43 Picasso Harle.,quin Seated .. 1923

*34.: 2D 35 Picasso Pierrotp Artist's Son 1925....- Paul at 4 . . ..

35. 1A 53. Picasso Harlequin) Artist's Son 192)3. -. .. 3atPaul. , : ....

36. 113 36. Picasso Maternity37. ...it 56 Picasso Sleeping Peasants 1919 .:. .,_: (Gouache)38. 1k ...; 514 ,. .Mondrian The Dunes39. 2A 44 Picasso

* Utilized in Forced Choice Test

I1

1

Al

Slidei FactorlIttribe.r Factto::' J^1dirr Arti t------- ------. Title Year

W.. 20 49 Fic;2.aio

hi -. 2B . h6 y?i CW'JSO Hrirloquin Seated 1923 ::*h2..:- 2A. =

31 - Pice.sso : 'Woman 1-1 White 1923133. 3A 1j8 -P.i.r.suso ,.....e170 NII.,Ias r

.

1920,Odl. IC 139:'; Van 'Gogh '- Sunf1c,we2.^ :":

45.46. JO 41- Monlrian -Tableau II 1921-47. ik : 56 Mondrian - Tableau I 1921h8. 2B .141 .- Mon2rian Composition With Red 1936-

and Black ..:49. ID . :34- Monerim Church Tower at 1909

.,.. Domburg

*50.- ].A 59 :Mondrian Still Life with 1912 --s.: .:..:-.:-.. - Gingerpot II .::

51. IA. 49 Mondrian .. Composition in Gray & 1932.---:Blue

. . .

52. 3A 55 Mondrian C Crilposi. tion with Trees 191253.: 3A. 56 Mondrian Lighthouse at West 1910.

-Icary.-Ale54. 61 Mondrian ' Drms 1910,55. :),.t;

:. , -58. :, .: Mondrian Oval Cor9osition,Troc.--ts 1913.56 1.13 34 Mondrian Ccaiipositi.021 1,3, Trees 3912

"c7 , 31: 66 Van Gcen 01 ive Grove*58. 113 30 . Mer,-.1rian J.,?..1.1-1.scaps with Farm-

house59 141 51 Mondrian :Hcfrizontal Tree6o. 2)3 42 Van Gogh : The Sea61. 111: 57 Mondrian ,Composition in Gray 193.9.62 2B :`:' 54 Van Gogh : Bridge at Arles

i:63. . 21) 35 -Van Gogh The Harvester*613. 2B 62 .. Van Gogh Self-Portrait with Pipeit-65. 213 59.. Van Gogh The Sower'65.. IA '1,50 Van Gogh -LIArlesienne: St. Remy 1890-67.. 2A 145 Van Gogh Landscape with Olive

Trees*68. , 2A 511 Van Gogh - Road with. Cypresses690. 311, 53 : Van Gogh The Orchard

*700 2)3 53 Van Gogh Orchard Springim71, 2A 52 Van Gogh The Sower

1888

72. .2C ': : 43 , ?ieasso '-

*73. IA 53 Van Gogh Olive Orchard74. 213 43 '. Van Gogh Road Menders 1889.75. lE 28 Van 'Gogh Good Satoritan (aft.- 1890

Delacroix)*76. 11) ho - Van Gogh. .. Stairway at Auvers 1890

A.2

-3-

Slide FactorEUmb:.%r r LFactor Artist

....,_.,. .. S.A.,. P ww!,.' .Titln Year

*77... . lE Co Knndrian Composition /,!-14 191478, 10 31 Mondr:i.an Composition 00, plus 1915

and Minus

79.i 1E 51 Picasso . Sunbathers ,

*80. 1D 54 Picasso*81. :ID 50 Ensor Carnival on the Beach 1867

1E 57 Ensor . .ITIT:,.Shells.. 189583. 20 142 Ensor Discarded Boats 1900

*84. 11) 37. lacasso . Owl on a Chair and Sea 1946. Urchins 0

. .

45. 20 53 Picasso Pastoral*86. 2A .. 55 Picasso The Kitchen*87., 113 . 1t6 Picasso Portrait of Claules

Artist's son with:,

. . , Hobbyhorse*88. 2A , 50 Picasso Mediterranean Landscape 1952'

*89. 20 50 Picasso Smoke of Vallauris 195190. lE )0VP Picasso Womm Seatd 194191. lA 59 Picasso Daughter of Concierge 3.9147

92. 2C Ill Picasso Figure 194593. 10: . 58 .

Picasso Still Life 1945213 66 Mondrian New York City 19)42

95. 213 66 Nondrian Composl.tion pith Reap 1921, Yell= and Blue

96. IA 20 Mondrian Red Tree 1909

97. 31) 42 Mondrian Weeds Ncar.Oale 190798.. 2B

. 54 Mondrian Mill in Sunlight99. 10 116 Mondrian Composition

1911.1922

100. lA 40 Mondrian Composition with Red, 3.930

.. Yellc and Blue .'

101. 2B - 56 Mondrian Oval Composition 1913*162. 10 62 Mondri.an Horizontal Tree . .

*103. 10 53 .. Monlrian Evening Landscape 1904-gaoh.. lc 59 Picasso Two 'Female Nudes 190605. 213. 70:: ._atcasso Harlequin Leaning 1909'.106. 2B ri?-. Picasso Flowers en a Table 1907107. 2B 62 ,. -Picasso Fruit and Wineglass 1908108. 211 47 Picasso Harlequin and his 1908 .

_ . . Family . . .

il309. lB .:51 Picasso _WorAn Seated 1909*110. 2A 60 :.Picasso Girl with Mandolin 1910*111. lB 57 Picasso Clarinet TaAyor 1911-12

*112. 13 59 . Picasso Landscape at Hortrt de. 1909...Ebro . .

,

*113. 2C 63 Picasso Nude With a Towel 1.907

*114. 2A . 61 Picasso 'Figure 1910

A

SliJe Faet,L.r

Nanbcv Factf-Ir 1.. ndinr Artistilr...0...*. Sm,..7,

*1150 13 49. ; Picasso :.. , - r...

*116, lh 64 7 Van GDgh117. 2A 37- -Picasso118. 2A 36!

119. 20 41 -Picasso*120. 2A 38 -PicAsso121. 2A 44 ....Picasso

Title Year,,

Les DohDiselles 1906-07diAvigncaStarry NightWoman Dancing 1907Throe Ssatad Figures 1907.

Self -Portrait

Still Life With Tube 3.909-:.... of Paint

122. 20 39123. 2B 37

ila24. 2E 65 Picasso*125. 2E 59 Picasso

*126. 10 55 ions,0127,. '20 64 51=-,so128. 1C 64 Picasso129.

1300 2B131. Ficasso

132. 10

1.33, 2B

*134.

136.

'B2C1B

*1370 1B138. 20

*139. 1B

*140.. 20

141. 20

142. 20

143. 10

144. 2A1450 IB

*146. 2C

247. 2C

148. lA

*149. IB150, 113

151..*152. 2C

*153. 2D*154. 2Eicfe 2B

49

646968

Picasso

PicassoYEicasso.'-inOVX30 I

.68 Ficasso72 ...1Cd000

PicassoPicasso

70 Ticasso

59 Picasso32

45 norteso67 Picasso52 Picasso72 Picasso

*5567

475651

62

59

35

PicassoPicassoPicasso

PicassoMondrianPicassoPicasso

A4

Still Life with Glass 1942 -

Still Life with 1937.GuitarTable Ssrvice 1941Woman at Mirror 1937-Bullfight 1934.

Still Rife with 1938Bull's Head, Beek,Palette and CandlestickCat and Di-r'. 3939.Worm Rc.oliniag on 1939DivanWomln on Iron SeatDull an1 HeroStill LIfe by Can:De-lightWcepingWom-InYoung Girl at MirrorO the Beach, DinardNuJeBather, Seated'Woman with a Flower

On the Boachs DinardThe StudioNude in Rod ArmchairCircus Family (Water-color)

Two NudesThe Farm House

Still LifeThe GraL2

1937-381934

1937

1932,1928.1932.1929.

1932.

1928.1928.1932

1920.-

APPENDIX B

The Development of a Forced Choice Measure of Preferences for ModernArt Styles and Validation with Personal Needs 1

In an effort to circumvent some of the limitations attached to theuse of verbrl materials to study personalty and to clarify some of thecharacteristics underlying preferences. for non-representative art,.a -

test was constructed utilizing different styles of modern. art as itscontent. The modern art scales were based on the research reportedearlier (Loveless, 1958) which had identified five factors from among160 slides of the paintings of Picasso, Mondrian, and Van Gogh. Twosets of 80 slide evaluations were factored independently and five fac-tors emerged from each analysis. The five factors were dubbed striae-turalrealism; primitivism-analytic cubism; synthetic cubism-surrealism;and expressionism after the characteristics of the paintings represent-ing the highest loading on each factor for each set. Paintings fromeach factor 'ere paired with paintings from each of the four remainingfactors according to their similarity of endorsement as being liked ordisliked. The paintings were paired within a range of 10 percentagepoints of endorsement from highest to lowest of each of the five factorsand were placed in a paired comparison paradigm. Each group of 5 slidesconstituted the basis of 10 choice pairs. Twelve sets of 10 paired cam,.parisons were composed and presented to a group of students (Nrz57).The pairs were alternated so that no similar pairs were presented con-tigously. Five scores for each of the two groups of six sets of pairedcomparisons were obtained. The students had previously completed. thePersonality Research Form (PR F) (Jackson, 196() an-.3. were eembers of apsychology class completing semester requirements for research partici-pation. The results were then intercorrelated and factor analyzed intotheir principal components and the factors rotated to a varimax crieterion, five scores from set 1 of the forced choice group, five scoresfromset 2, and 15 scores from the PRF. The results appear in Table 1.

RESULTS

The intercorrelation between the 10 art scales ranged from .02 to.) 9 with a mean of 20.5 and a standard deviation of 13.4 indicating ageneral independence of each of the ID scales. Ten factors were ex-tracted from the variwex program on the basis of the relative indepen-dence of the ID art scales.

The first factor to emerge from the analysis is a complex factorwherein cubism (synthetic) and abstractionism represent the art scales

1 Prepared for submission to division 10, American Psychological Assn.

B1

while five FRI? scales load highly on the same factor.. The scales ofachievement, affiliation, aggression, autonomy and dorrinance all loadnegatively on this factor. One apparent suggestion might be that thefactor represents a form of int,ealeetualism or tolerance for distortion and nonmeaningfulness. Factor V loads moderately (414.) alongwith needs for order (c64) and social recognition (.85).. The realistic expressionism of the art, works suggest a high need. for conformityis associated with this factor. Factor VII loads on the expressionismfrom set 1 and ti e abstract works of set 2 and only a weak relation(.33) is found with a need for harm avoidance.

The results in general indicate only mild congruence between needsand art preferences since many factors load with only art scaleb orwith personality scales.. A question remains whether the forced choiceformat which results in ipsative measures may not have attenuated the 'results.

:82

Table I

Factor Analysis Results; Forced Choice Art Scales And Personality Research

Form Scales With Factor Loadings

Art Scales

Art IA-expressionistic'e.g. old guitarist-PicassoArt IB-cubisme.g. clarinet player-PicassoArt IC-classicale.g. the lovers-PicassoArt ID-expressionisme.g. stairway Auver-Van GoghArt IE-abstracte.g. landscape-PicassoArt IIA-classicale.g. woman in white-PicassoArt 218- cubism

e.g. woman seated-PicassoArt TIC - surrealism

e.g. smoke at vallaruris-PicassoArt IID-realistic/expressioniste.g. Pierrot-PicassoArt TIE - abstract

e.g. woman in blue- Picasso

PRF Scales

AchievementAffiliationAggressionAutonomyDominanceEnduranceExhibitionHarmavoidanceImpulsivityNurturanceOrderPlaySocial RecognitionUnderstandingInfrequency

'I II III IV V VI VII VIII IX X

07 08 08 -15 -03 00 41 16 11. 67

-94' -09 -09 -04 -05 -06 01 04 -00 -08

23 05 06 -18. 05 -25 08 29 -20 55

24 -14 06 -68 01 23 -03 11 -02 04

-96 -10 -08 -07 01 -06 -05 -03 -00 01

-22 -11 -07 -55 23 02 19 41 -16 18

28 -11 -14 -18 -00 -10 -10 -13 -06 74

-05 -02 26 -78 -19 -10 -10 -04 14 16

36 32 04 04 -44 15 04 26 12 07

08 -00 07 00 16 -00 80 07 -01 11

-91 -03 -10 00 04 19 -03 -04 -01 -01-79 02 15 01 36 15 -11 -17 -06 00-92 -16 06 03 09 -10 -06 06 -01 -00-92 -13 -08 03 -14 01 06 00 01 -09-91 -07 01 -01 13 15 -06 07 -02 -0911 12 -28 -04 -06 78 07 -02 08 -14

02 -05 30 -12 26 66 07 -16 02 -0831 -11 -45 05 09 -31 -32 -28 -02 0314 00 73 -07 -27 -22 16 -05 -02 -0027 28 21 02 27 43 -21 16 -15 0022 -18 -24 -01 63 28 29 -08 18 0330 03 68 -16 23 09 -12 -05 09 -04-11 -01 03 10 84 -03 05 07 -02 -0021 35 -23 04 -22 59 -08 25 -15 01-00 -06 -06 07 -02 -04 01 -18 -83 06

133

APPENDIX C

15

Art, Politics, and Sex

The study

The results of a recent study at the University of Notre Dame indicate thatartistic perception may be a more complex phenomenon than has been previouslysuspected. Current studies of artwork as psychological tools are continuing. Mostof these current studies still employ expert judgments and/or accepted rankingprocedures to categorize art on a given dimension. These procedures result in aunitary concept being employed to account for all the differences between theworks of art i.e., it is either aesthetically pleasing or it is not pleasing. The NotreDame study analyzed the preferences expressed by college students for a wide varietyof modern art. These preferences were analyzed into a series of independent sub-scales; that is, the standard "like" or "do not like" groups were broken into definedsubgroups using a standardized computer program procedure. Thus the availabilityof the computer and computer technology makes possible a more complex andperhaps more realistic description.

It would appear from the Notre Dame study that the identified dimensions donot offer clear divisions between paintings as logically categorized by style, or byartist. Apparently, however, there is an interaction of style, content, color, andclarity to define each dimension. Previous studies have been able to identify only asingle bipolar category or dichotomous reaction to the paintings. This study,through computer analysis, has been able to identify five groups that are, in essence,breakdowns of the previous dichotomous reactions. The five identifiable groups arelabeled, for convenience, as follows:

1. Structured realism: highly representational paintings and photographic-typeart.

2. Primitive-analytic cubism: paintings that severely tax a person's ability toidentify the original theme of the painting.

3. Synthetic cubism: the use, of angular marks to create an appearance of aperson or an object.

1Dr. Eugene J. Loveless, Queensborough Community College, New York. This study is takenfrom: "The Dimensions of Preference for Modern Art," Proceedings, 76th Meeting of the AmericanPsychological Association, 1968.

Chapter 15, pp. 171-75 inINFORMATION PROCESSING:APPLICATIONS IN THE SOCIALAND BEHAVIORAL SCIENCES, byWilliam I. Davisson, Pub. byAppleton-Century-Crofts.Copyright 1970, Meredith Corp.

171

"PERMISSION TO REPRODUCE THIS COPYRIGHTEDMcyATER IAL B M CROFICHE ONLY HAS BEEN GRAN D( hamTO ER C AND ORGANIZATIONS OPERBTING UNDEAGREEMENTS WITH THE U. S. OFFICE OF EDUCATION.

FURTHER REPRODUCTION OUTSIDE THE ER IC SYSTEMREQUIRES PERMISSION OF THE COPYR IGHT OWNER."

PREFERENCEFOR MODERN ART'

172 / Computer applications in the social and behavioral sciences

4. Surrealism: an amorphous fluid-type of distortion which permits specificdetails to become exaggerated.

5. Expressionism: the accentuation of one particular color, modality, form, orother facet, to obtain a desired effect.

Thus, the implication clearly appears to be in support of a more complexconcept of artistic perception than is currently available even when a relativelysmall number of works of art (160) and artists (3) are studied.

The identification of the components of artistic preferences is made clear bythe specification of the underlying characteristics that contribute to these judgments.These characteristics are complexly interwoven into each picture so that classifyingart on a single dimension makes it difficult to deduce psychologically meaningfulimplications from these works. The computer analysis has helped to isolate eachdimension so that pure forms may be identified and applied in psychological re-search.

The approach

Studies which seek to employ works of art as measures of individual variationrequire a few statistical control procedures. These are relatively easy to master andto administer. The computer programs necessary to successfully complete such workare also availabl,...2 The first question that comes to mind concerns the method ofresponse. The University of Notre Dame study illustrated here used the "like" or"do not like" dimension. This simple dichotomy of responses is the broadest andmost comprehensive category that may be employed in studies which use artisticmaterials as data. On the Hollerith card the data would appear as 1 for like, and an0 for dislike. Conceptually each card column would represent a picture, with theindividual's response to each picture coded into the appropriate column as a 1 or an0. The usual method for this is to have the answer written or inserted into a pre-printed form, with the data subsequently keypunched onto the Hollerith card.

The reason for recording each response is twofold. In the first place, it is im-portant, for the psychologist's work, to know how frequently the general public likesor does not like a given work of art. This constitutes the "level of endorsement" fora work. After a given picture or group of pictures have been seen by a large numberof subjects, the endorsement level will be measured by the percentage of the totalthat do like the picture. This endorsement level is usually obtained by a frequency-count computer program, normally available as a standardized program procedure.3

The second reason for recording and analyzing the data is that it becomesimportant to sort the artwork into different categories, based on style, content,author, or whatever characteristics are implied from a clear-cut objective andempirical basis. These objective categories are revealed by the application of factoranalysis to the responses of pe 'pie who like and dislike the pictures.

While it is not necessary to explain in detail this second phase or to explain howa factor analysis program operates, the following sentences may be useful in indicat-ing something of the nature of factor analysis. It is sficient to add that without the

2The most general program group would be the BMD group, cited in Chapter 9. See also:Hallworth and Brebner, A System of Computer Programs (British Psychological Society, London,England, 1967).

3See Chapter 9 for a discussion of the standardized program procedure for single-column fre-quency count.

C

computer this kind of analysis (even on the small number of pictures and artists),would not be practicable.

Factor analysis is a statistical technique for breaking down an intercorrelationmatrix (table) among the major abilities, dimensions, or components which contrib-ute to each correlation coefficient. In the case of the art, the like or dislike responsesshowed systematic relations among the five groups of pictures (realistic, surrealistic,etc.). The relative clarity or purity of each style was represented as a coefficient be-tween each picture and the factor representing that style. These coefficients arereferred to as factor loadings and are considered to be meaningful when they aregreater than 0.30 (as a rule of thumb). Each factor then reflects a linear descriptionof the most representative pictures (based on the empirical analysis), in the orderof the saturation of each picture on that factor.

In order to understand the idea of factor analysis, one must first considercorrelation. Basically, the correlation between two variables is an index of the degreeto which they are associated. For e:, ample, the degree of correlation between twopictures from the study is an index of the extent to which people who like onepicture also like the other. If all the people who like one picture also like the other,the correlation between the pictures is perfect and the value is 1.00. If none of thepeople who like the one picture like the other, the correlation is zero.

The basic material for factor analysis is a correlation matrix as shown in Figure15-1, below. In the correlation matrix the correlation (association) between eachpicture (or variable) and every other picture (or every other variable) is represented.The matrix illustrated below is a hypothetical matrix for illustration.

The diagonal (Cells 1,1; 2,2; 3,3; 4,4; 5,5) contains the correlation coefficientsbetween each picture and itself; thus, all the diagonal values equal 1.00. Since thematrix is a square, all the information above the diagonal is repeated below thediagonal. A correlation matrix contains a series of values, and each value relates toonly two of the variables (in this study, pictures), involved. If the matrix is a large

2

3

4

5

PICTURES

2 3 4 5

1.00 .90 .15 .09 .95

.90 1.00 .25 .10 .85

.15 .25 1.00 .80 .17

.09 .10 .80 1.00 .21

.95 .85 .17 .21 1.00

FIGURE 15-1. I ntercorrelation matrix.

.3

Art, politics, and sex / 173

174 / C-omputer applications in the social and behavioral sciences

one e.g., 160 by 160 it may be very difficult to interpret the matrix by inspec-tion.

An early attempt at a sort of factor analysis consisted of analyzing a correlationmatrix into a set of clusters. A cluster is defined, for our purposes, as a set of vari-ables (pictures) which are highly correlated with each other, and which tend not tobe correlated with other variables (with variables in other clusters). If it is possible toanalyze a matrix containing, for example, 100 variables into a set of 10 clusters, it isreasonable to assume that the relationships expressed in the matrix are due to nomore than 10 basic and relatively independent influences or factors. The nature ofeach factor can be determined, roughly, by examining the variables that comprisethe cluster.

The matrix shown in Figure 15-1 contains two clusters. Note that the correlationbetween Pictures 1 and 2 is 0.90, the correlation between 1 and 5 is 0.95, and thecorrelation between 2 and 5 is 0.85. Since these pictures are correlated with eachother and not highly correlated with other pictures, they comprise a cluster asdefined above. The other cluster consists of Pictures 3 and 4, since they are highlycorrelated with each other and not with Pictures 1, 2, and 5.

Factor analysis is much more sophisticated than cluster analysis, but the resultsof a factor analysis can be interpreted in a similar manner. The results of a factoranalysis are typically presented as a factor matrix in which the rows represent vari-ables, and the columns represent factors. The entries in the cells of the factor matrixare called loadings, and a given loading can be seen as representing the correlationbetween a factor and a variable. Thus the nature of a factor can be determined byexamining the variables that have high loadings on that factor.

Table 15-1 is illustrative of the two tests used in evaluating the reactions to thepictures. The table shows 25 of the 160 pictures as an illustration. As suggested in thetable, the number of artists represented was small in order to minimize the effect ofpersonal stylistic variables. Picasso was the basic artist because of his range of styles.Mondrian pictures were included in the structured realism group. Van Gogh pictureswere included in the expressionist group. Each artist was included on an a priori basisbecause they appeared to represent psychologically meaningful dimensions of art.In consultation with a resident artist the pictures were grouped into sixteen groups.These sixteen groups included the five shown above but were more detailed. Thefactor analysis program reflected only the five factors or groupings, perhaps re-flecting the differences when professionals and laymen regard artworks.

Table 15-1 shows the title and the painter for each of the pictures. The FactorNo. (I, II, III, IV, or V) indicates the factor into which this picture was classified bythe factor analysis program. Loading is defined as the amount of commonalitybetween/among items in one factor with some trait in common and identifiablefrom other factors or groupings in the same matrix. In looking at the column loading,any value over 30 represents a commonly accepted standard for inclusion within afactor and uniqueness of that factor as compared with other factors. The finalcolumn simply shows the percent of total viewers that liked each picture i.e., thelevel of endorsement.

The data set

The pictures were projected from color slides before two separate groups andwere presented to 30 to 35 viewers at a time. The first subgroup viewing the picturesrepresented 600 students at several colleges throughout the United States. These

TABLE 15.1. Paintings, factor loadings, and preference levels for artworks

Permission to reproduce this table forERIC could not be obtained from theAmerican Psychological Association.

Eugene J. Loveless, "Dimensions of Preference for Modem Art," Proceedings of the 76thAnnual Convention of the American Psychological Association, 3 (1968), pp. 445-446, reproduced bypermission.

students were primarily from psychology and education departments. The studentsresponded on optical scan sheets for each picture. The scan sheets were convertedto Hollerith cards for analysis. Use of a single-column frequency-count programprovided the information for Column 5 of Table 15-1.

A second subgroup represented students from a private nonsectarian collegein a major metropolitan area who were taking their first psychology course. Thisgroup constituted 193 viewers. These persons responded as did the first group, andthe resulting cards were subjected to the BMD factor analysis program (Chapter 9).The results of the analysis are shown in columns 3 and 4 of Table 15-1.

Conclusions

Whatever method of external validation is to be used, the materials originallyselected a priori have now been established to have constituent subgroups or sub-scales. These represent empirically established dimensions of artistic preference. Thepruning and selection of the most potent and independent representatives of eachsubscale provides rapid and dependable materials for further application and valida-tion procedures on any equivalent population.

App ma x D

Reprinted from the Procudiv. 76th Annual Convontion:APA, 1963

DIMENSIONS OF PREFERENCE FOR MODERN ART

Art, politics, and sex / 175

CS"

EUGENE LOVELESSUniversity of Notre Dante

Permission to reproduceAppendix D for ERIC couldnot be obtained from theAmerican Psychological Assn.

Appendix E

List of Measures and Subsea les Utilized in this Study .

Guilford .Tests of Divergent Thinking.

1. Uses for thingsa) total uses b) clever uses

2. EXpressional Fluency Scores - Total Score

3. Plot Titlesa) Total titles b) clever titles

4. Possible Jobs - Total Number of jobs

Scholastic Aptitude lest (SAT)a) Verbal scoreb) Quantitative - Mathematical Score

Oberlin Art Test - Sum of relevant art questions .

from Al B, Co and D Forms of the test

A. C. E. Test of CritleK1 Thinking

ThursbnIs Concealed Figures Test (Spatial Ability)a) Number of correct responsesb) Percentage correctc) Last item attempted

Color Word Test - (Stroop)a) Individual administration1. Words alone2. Colors alone

3. Interference score corrected for initial readingspeed by a regression method.

b) Group Administration1. Colors alone2. Interference procedure

Schematizing Testa) lag; score

b) Error score

El

Complexity Scores (Bieri "Rep" Test)a) Vortical Complexity - across categoriesb) Horizontal Complexity - within categories

Size Estimation Test - Error Score

Budner Scale for Tolerance /Intolerance for Ambiguity

Gough Rigidity Scale

Holland Preconscious Activity Scale

Rosenberg's Measure of Self-Esteem

Rotter's Internal/External Scale of Social Reinforcement

Barron's Ego Strength Scale (M 1LP I)

Revised Art Scale - Barron-Welsh Figure Breference Test

Rokeach Social Attitudes Scalesa) Dogmatismb) Opinionation

Holtzman Inkblot Test - Computer Scored

Location2. Rejection3. Form Definiteness1. Color5. Shading6. Vxmement

7. Integration8. Human

9. Animal]O. Anatomy11. Sex12. Abstract13. Anxiety14. Hostility.15. Barrier16. Penetration17. Popular

E2

Omnibus Personality Test --

1. Anxiety Level2. Altruism

3. Autonomy4. Complexi.'y5. Estheticism6. Impulse Expression7. Masculinity - Femininity

(Heist etal.)

8. Personal Integration.

9. Practicality10. Religious Orientation11. Respon-.2 Bias12. Social Extroversion13. Thinking.Introversion14. Theoretical Orientation

Harvey's Conceptual Systems Test (Objective Form)

1. Divine Fate Control2. Need for Simplicity / Consistency3. Moral Absolutism4. Need for Structure /. Order5. Need to Help People

Personality Research Form

A.1.

2.

3.

4.

B.

AbasementChangeCognitive StructureDependence

1. Achievement2. Affiliation

.AggressionAutonomyDominance

6. Endurance

3.

4.

- (Jackson)

6. Need for People7. Interpersonal Aggression8. Anomie

9. Abstractness

S.. Sentience6. Succorance

7. Desirability

7. Exhibition 12. Play8. Harm-avoidance 13. Social Recognition9. Impulsivity . .14. Understanding .

10. Nurturance 15. Infrequency .

il. Order

E3

p*I771,57,

Appendix F

Tables of Correlations

1. Correlation of Forced Choice Art Preferenc, Test with Personality,-Cognitive Styles and Divergent Thinking Tests.

2. Correlation of Forced Choice Art Preference Test with CognitiveStyles and a Divergen!, Thinking Test.

3. Correlation of Art Test Scores in Forced Choice Format with Barron-Welsh Art Scale and Holland Preconscious Thinking Scale, (Architec-ture students) .

Correlation of F C A-P T with Computer Scored Results of HoltzmanInk Blot Test (H.I.T.)..

5. Correlation ofFCAPTwith Cognitive Style and PersonalityWasures.

6. Correlation of Free Choice Version of the Art Preference Test withScales of 0. J. Harvey's Test of Conceptual Styles.

7. Intereorrelations and Scales on the .Forced Choice (F C A P T)Version of the Art Preference Test and the Free Choice Version ofthe Test Utilizing the Same Pictures.

8. Correlation of Free Choice Art Preferences with Two CognitiveStyle Measures. .

9. Correlation of Barron Welsh Figure Preference Test, Holland Pre-conscious Thinking, A.C.E. Test of Critical Thinking and theScales of the Art Preference Test (Free .Choice Format).

Fl

1

I.

I

I

I

I

I

I

I

I

I

C)0.

nI na to toW ti 0 to ',"J7,. 1.71 C)

0Ci'0P-P

I-, to CAA) kAa c+I c ig

Ch C.)

oio o o o

0

0 0 00 L-2tr 4-- 1 0 R F5

c3

t8 rT) 'et Pop:1a Fo. ,S

,. . .

|R/ 6 G 6 c) C5 1-6 0i1) OD 3 -.3 IV

lep-oHL o 0 00

j

k 0.CI 0

ktj1,1

1-- P 0H1,,?

to W/:,-.1 .0 (4 cio 1.4It.4

ti iee .. 0talO t...-

a0ttl G o

r-3v-0o L-1-

Fr C; !,1'

P. 1-4.icl- 01.-I 0 Cl, It Io S

t.]-2 c) 0 0O00 1.1 Co

1.-J Ki PI: 6- 00 , 0<4 0 ..

4 ry,co Is'

1,-, o0 I-4ti et 'I'd 0) 0 v-BP- . Ner 0 :'

WO 0 Clri syv 1;1O t.1 P. 11 tN 1\3 H11./ GO til. 'Y

N cp

0 0

13

..1 0,,0

r 24 0o LI0-

00tlj

k 1-D 0o

0 tr,ato riu I.J.

P At.:A.1 C,

14

0 IJ, O-

h C-t1 p. 4

,

.

'&16118000

8a0' 1.47.)

F2

ks)CYJ 1-3 cto

ft;

MK

T 1

(C

ent I

d.)

Cer

rela

tion

-.Fe

z-co

t:I.-

Cho

ice

Art

.1=

refe

renc

e :T

est w

ithPe

rson

ality

: Cog

nitiv

e SI

KT

los

and.

Div

erge

nt -

T'n

inki

ng .T

ests

tri.

kA,

Cr"

.tor-

in.

Obe

rlin

Art

Tes

t-:

: Mer

lR

ep.

Rep

.H...

.....

Neg

ativ

e

-._

Anx

.le

vel

02.

17 06 19'.0

3.0

2..:

03

03 -13 01

:7,0

5= 2

9

Cor

rela

tions

are

und

erlin

ed

Altr

uism

Aut

on.

Ccm

p le

::E

s.:1

-let

.

::::

.:

-..

Imp.

Exp

. :..

M. /

F..

Art

_..._

1P.

13 10 1D 13 2L 23 2C 2D 2E

28 25_

30 09 23 28 38 30 11 21

-. :04

13.

051L

0973

1-4

.30

19..

0403

-.-

09_

01_.

......

'01

.06

06, :

2:..

-08

03-'

' 03

06

.__.

.._13 13 10

3 21 .18 t

Oa

:11

est-

%4....

08 .22

-04

05 -.05 03 -1

0T

2.

-T. 2

-b-3

.03 02

21 20.:

17 .19

';28

.: .2

8. 2

4...

19-.

..15

..,27

.37

- 28

:- 3

31)

,

36 :5o So 29

.

_37

.16

.05

o8.1

),,:;

. 09

.11

08..1

921

.... 0

5;

02' .

09...

, 05

:02

...M

....

'''o6

:- o

4.-

.T6

-.:-

, T.i.

:. 12

:;:: 0

6

(Cen

t

..t

TABLE 1 (Cont id.)

Correlation of Forced Choice Art Freforenco Test withPersonalitys Cognitive Styles and. Divergent Thinking Tcats.

N244 00517 29 IP>,01:-2 38.

Negative: Correlations a-e unaer.lin.*.

CriteriaArt Pers.Tnt, FracoOr4ar3,A 03 . 091B Ziff 0810 o5 0331) 13 023,E, ..:. 03. -a2A , 'CZ 53213 '69 oo20 07 122D 03 "trg

2E D 25

lh1B10'ID1E2A2B2C2D2E

Re3ig. Or ysap. Dias

:

Soe.Ext Th.Int

04To1571

Oh7.2'"371.

:Ti .,11.; ...-.

'Cill 710

11-19

13, 1.6

::160515.113.9

18

01bfoo1.1Oh

. 01020506014

*,aar.2828241/31252938

.2923

Theor .Or CEEBV CEED::2

Ro.A.iaoll.0pin. ......._...

or;

.91.12.........0-2

o60302*5315812

Dog.

15ITOIigIiii.o2154oh05o826

010403li06

OU080202

3835'33317t222.2iliT5-3.,,,...1.

18

07o;r.

311

--1.'

22Via.

32Ti7?31'

-2i

Eh

...Buller

TI A

oh0802

Ki6-o'd

07oh150016

3A113

. 303D3.E

2A213

2021)2E

113

3D

3.E

213

202D2E

TABLE II: 2:.1

Corre1.tion of Forced Choice Art IP:eafevence. Tent, withCognitive Styles ana a Divergent Thinking Test

Nu7:.1-ydr of SubjectsVarabie 3. -

15: 21:'22 - 23

126

5468

Significonce.: 3eve.i3.s13>G03.7 i'>.01,22

P>.01:z3h?>.O23 P>.01=30

Net.; tive Correlations are underlined

-Poss1.133..-..

Jol..-7,

2226030615:11

. 2'2Ti22

-.

02.

Conc;i:alc;', Fi.gives TestRts Triecl 9/0 Rt.----07 01 0321 o6 1302 30 . 12

.313 97:, -6i:C6 it 0210 07 3.1n)e...) 22 : .... 1 5

Ili 11 2126 13 22

Colo. Word Tot,tColor:I; Co lrIrs (9,01-r5

03 047;46 22 .3223 i6 22

:16.:';f,1'0, .a.;

A .iti'd

3.3 27 3.3T6 cirt 7fiq

10 IC) 13.22 ,.. 06 16

Color Vord,Tcpt cam!Ire;;,-Pc. kEnt;.:JfNaz...1MI. ...m.o.. 4,0

0123

o30OD

173:

-g401

"i53

0 7. 0 2. 0 2

bO18 02 .

:.

28 22 2602

03 0913 02 1097 Oh 01

F5

Schemativ.ing Testrag Error

35 2112

3203 03

073207 Ni10 13>Zin5 *.fiTt

03

Correlation ofwith Barron-Welsh Art

1 . ;

TABLE III

Architecture Students

Art Test Scores in Forced Choice FormatScale and Holland Preconscious Thinking Scale.

N=33 Y).05.3 35 P>001=4

C. fiil'FAfi EPC.

lkan

1605

h1

1C

27 171P1 062A. 03 312B 21 1Di

2D2t

23..

2E 27

F 5

TABLE IV

Correlation of F C A P T 'with ComputerScored rosvits of Holtzman Ink Blot

test. (HIT)IM96 P>45:r..20 jP>.0Di: 26

31Blc3.D

la21i2D202D21

.

Rer(r)0703:I)!

02' '05

091505

.2h,Oh

Lecat.

01: ,7:: 26 -.,

"ol03160610021306

Reject

00.00 .

0000000000000000

Form def. Color Shaeo

-O6-f.3'

Oh'12232303052405

170522.01

:.060900031361

-13-5303

..:'?1,.

o3272608hl01

lh1B

nwt,

.0209

JILL:a

0700

: Moan

17oh

10 . Alt 20 1631.E :a i6 -0'113 :.02 ti2 002A .o6 06 06213 02 0)4 162C 11 0)4 182D T6 17 -66253 10 21. 09

Abf.ltr. Hestne

3.A og 12 0) r

1B 20 08 2!;30 16 9,7, 163D -07 06 073.13 r. 1.6 022A 01 03. 014

2B 07 08 oh2C 22 26 2721) 50 22 j2E 03 02 06

P7

Anirial

12lit713`i-§0207071611'a

.. .

An 'Al,,

....

Se>:

21 1302 2013 050.4 6505 1806 1013 014

25 02*67 35,15 by

Taaig Pcnot. Pop.

09 23 3209 06 1615 3,8 2018 'a Td6fr. 09 2700 13 16'03 12 0007 2)4 32di 03 -6613 03 22

TABLE V .

Correlation of F C A -P T with Cognitive Style and Personality Measures

Nz75 P>.05:: 22. ..

:P>.01t1.29.

Rosenberg -Buiner ...- : . 7 Group Obj. Sort TestSelf- Gough Tol/Int Rotter OST OST OST

Fstr,on Rigidly Ambiguity Tis...._ Groups Sfinglcs. Tot.:04

1 162 243 20Ii. c65

1.26 11-7 L.C.8 u >

9 2110;: 05

2931.

11-15::;

0925Y40503

....0.. :

.

1322

17163.1

72:,i290566

::.

..

.::

0222....,..1.ue10151415i:9'0913

:

153836-6613003615073.1

04033.3.

003,6oolit06 J

0023

052311o319002003.'63,3

Porsormlity Rcs::.).rch Form -, Jackson

3.

2314

56789

10

Abast-I.

25o617o 28102220210

Clin.l.s,rie

18. 16 :

o6I59:

7/3130931

752. ,

12

Coff.Srurl.

0823

150003Vli1330608

°Dep:In3.

1518

16:07 .

33-0n.07-a

F8

:Sent. Succor. Des 1 rc.,..

30 03 0307 05 3.3.

05 17 0520 07 12

73-ii.

6. o613

o603.

33_ 03 214

Il 19 '12o9 ti6 23

UO cost .1"---us

010 0!!-.710 0Kly);C..."

o0IJC) {_r to 010-.3 I-1

00O:010 C.).!N) 0 64%.0 0 a--31 )

C) tJo

CI-,

!API R II

GL

cf)

0 0c+

ct,

..

6.) C) to 01011.-i0

1-4 14 0 1.4 ro 0C/N \0 N) 1.1

OS(.` 00 C)0*-0 c.4co N vt

10 0 0 I-I f'0C,,i Es 0 I.-'t) 0 Cr,. 1.4 IQ 0 1.1

0 0 0 CIO0 ks,1 0 0

ski clO C)'1.1 C)

01--1Ali

0 I4 0re.

O 0 0Cs}-1C)

to cm r-t1-,. rig c+

$1 C'.1

I;:.1 'XIt)II

1-1t...; 0 C)4-# s4.) Cr) 1)

pti 82 FrCo rt.)

F9

0 0o

0

121

U0 CO

*T3 t.1 1-h- 0CI) :3

VI 0 <I- 018 0 C;)

COro c

o

oo ct1:1

I0

111.

g*

0 gFJ

C)0

TABU VII

Intercorrelations and Scales on the Forced Choice (F C A iP T)Version of the Art Preference Test and the Free ChoiceVersion of the Test Utilizing the salrn Pictvres.

N=105 for forcc;a choice N=80 for free choice(above. diagonal) . , . (baluz, diagonal)

FRRE

C

H0I

cE

lk

IB

10

ID

:1E

2A2D202D

2E

]J

83513817

-076797555

44

133

045638

:V.

8456565952

37

1C

01586115-2.;:J..

37563619

.22

FORCED CHOICE

1D.. I 2A 212

00 02 01 0874 17 71 6572 25 83 5967 15 80 6619 71 21 18

TE 66 81 701-6 66 72 62

1'....1 72 82 73.2-6 55 57 450r9 147. 47 149

2C.

.06

71617623817365

4755

_ . .

2D

0562647621817883

62

41

2E

005o6960

2473

62

67

7o

68

* Ruder-Richardson formula 20 reliability coefficients for theforced choice itcli.s appear in the diagonal.

.F1.0

: .

TABIE VIII. . -,

. Corm :Intim of Free Choico Art Pd7eforenceswith Ti, 7e. Cognitive Style Measures.

Nv:304 P>0052:19 P>.01.72.5.. .

..-. : .

Concealed Figures Test Color V;ord TestInterference3 .. 2

OV..2.1Pryi..... ...ft..... .10.10N1

:06 0,-...65 k?

No. Rt. y.T.C:, No.Tried Colors -1 2 -... 3 ... I 2:.WV

1 14. 09 05 19 282 39 10 . :' 07 : ' 33 353 09 03. ' 03 3.0 084..)

...3.3.

*Fa

32151,3

:.

01;

15 .:Ito

72V

23,37

6 16 : 13 04 27' 317 12 ; 03 10 17 . 268 13 l0 03 37 1,2

9 3,5 14 01 26 18 .

10 24 07 20 24 26

F3.1

oh01

'oh

o63.3

U04 3.6

05 1912 3521 0032 oo

TABLE IX

Correlation of Barron -Welsh Figure Preference Test, HollandPreconscious Thir Ozing, A.C.E. Test of Critical Thinking andthe Scales of the Art Preference Test (Free Choice For :t)

Nz.189

B2.rron-VolehArt.Scai.

11111.tM ,,M%/0.4.1,........,1,7714.

P .05= 21

HollandP C S

P .012 28'

A.'..C. E.Cri t TM n1-1.1,,,......... EP.]: setAW.AA

1 23 26 022 72 .67 IT§3 63 5o ho4 76 61 495 6953 5o6 69o 6 457 57 53 .288 3.3 3h 099 .53 .3:1 -:??:

10 19. 05 07

P32

References

Barron;, F., & Welsh, G. S. Artistic perception as a possible factor inp.:,sonality style. Journal of Psychology, 1952, 33, 199-203.

Barron, F., Creativity and Psychological Health, Van Nostrand, 1963.Burt, C. The Psychology of Art. In C..Bnrt et al., How the ;rind. Works.

London: Allen & Unain, 1933. Pp. 267-310Child, Irving J. Personality Correlates of Esthetic Judgment in College

Students. Journal of Fersonality,.33, 47644.Clemens, W. V. An Analytical and Empirical Examination of Some

Properties of Ipsative Measures. Psychometric Monographs No. 14, 1966.Eysenck, H. J. The General Factor.in Aesthetic Judgments. British

Journal of Psychology, 1940, 31, 94-102.Eysenck, H. J. "Type" factors in aesthetic Judgments. British

Journal of Psychology, 1941, 31, 262-270.-

Fisher, C.; Paul, I. H., The Effect of. Subliminal Visual Stimulationon Images and Dreams. Journal of the American Psychoanalytic

Association, Vol. 7, 1959. ,

Gardner, R. W.; Holzman, P.S.; Klein, G.S.; Linton, H.; Spence, D.P.;Cognitive Control.Psychological Issues, Vol. 1, 1959. .

Gardner, R.W.; Jackson, D.N.; Messick, S.J.; Personality Organizationin Cognitive Controls and Intellectual Abilities, Vol. II, 1960.

Golann, S.E. Psychological Study of Creativity. Psychological Bulletin,1963, 6o, 548-565.

Guilford, J.P. The Nature of Human Intelligence. McGrawrHill N. Y.,1967. , .

Jackson, D.N. Personality-Research Form Manual, Research Psychologists. Press, Inc.,. Goshen, N. Y., 1967.

Klein, G.S. Adaptive Properties of Sensory Functioning, Bulletin of theMenninger Clinic, Vol. 13, 1949.

G.S.Klein, Peremptory.Ideation;,Structure and,Force in Motivated Ideas,. Psychological Issues, Vol. V, 1967. . .

Loveless, E.J. Dimensions of Preference for Modern Art, Proceedings,76th Annual Convention, American Psycholofacal Association, 1968.

Mahinnie, Harold J. Review of Research in Aesthetic Preference, ActaPsychologica XXVIII, 1968.

Pavlov, I. Psychopathology and :Psychiatry, Foreign Languages PublishingHouse, 1960.

Spence, D.P. Subliminal Activation of Conceptual Correlates, Journal ofPersonality, Vol. 30, 1962.

13.

Bibliography

Barron, F. Complex:Lty-Simplicity as a Personality Dimension. Journal of .

. Abnorm. Soc. Psychology, 19531.48, 163-172.

Budner, S. Intolerance of Ambiguity as a Personality Variable. Journalof Personality, 1962, 30, 29-50..

. .

. Buros, O.K. Sixth Mental Measurements Yearbook, Gryphon Press: 1960(This is the basic reference for all published tests utilized inthis study).

{.

French, J.V. Kit of Selected Tests for Reference Ap'ritude and Achieve-.ment. Factors. Princeton, N.J.: Educational Testing Service 1964.

Horst, P. Personality, Jossey-Bass, 1968.

Rokeach, H. The Open and Closed Mind, Basic Books, N. Y., 3960.3

Stroop, J.R. Studies in Interference in Serial Verbal Reaction., Journal of Experimental Psychology, 1935, 18, 6113-661.

Thurstone, L.L. A Factorial Study of Perception, PsychometricLlonogr.No. 4, 1944.

Witkin, H.A.; Dyk, R.13.; Faterson, H.F.; Goodenough, D.R.; -& Karp, S.A.Psychological. Differentiation. New York; John Wiley,

H.A.; Lewis, H.B.; Hertzman, M.; Machover, K.; Yeissner, P.B.;&Wapner, S. Personality through Perception. .New York; Harper 1954.