personality attributes associated with two measures of cognitive style

12
Aeta Psychologica 31 (1969) 353-364; © /¢orth-Holland Publishing Co., Amsterdam NOt to be reproduced in any form without written permission from the publisher PERSONALITY ATTRIBUTES ASSOCIATED W~TH TWO MEASURES OF COGNITIVE STYLE GEORGE SHOUKSM1TH Department of Psychology, The Queen's University of Belfa.~t, Ireland ABSTRACT A new ol3en-ended,abstract task which has n, true solution is t~sed to produce two measures of cognitive style. The task consists of a number of drawings, each of which shows two sticks which are partially buried in the ground. From a consideration of the choice pattern of subjects who are asked to suggest which stick has the longer total length, two style measures may be derived. Meast~re 1 assesses individual non- conformi~.ywith the group approach to the problem. Measure 2 is a positionv.l score, reflecting the effect on judgment of the relative positioning of the ~wo sticks, qhe nature of these measures, in relation to certain group and pcrsonality variables, is investigated more fully. It ~s cor~cluded that people adopting a particular style vary in their personality characteristic:;by sex and from group to group. These differences are discussed and some general conclusions presented, relating the 'abstracting', consistent style to certain personality attributes commonly associated with t~e imaginative, creative thinker. A ,~umber of recent studies reflect a growing interest in non-cognitive factors which affect cognitive functioraing. The development of research into the nature of creativity is, in one sense, an example of this trend. In essence this field of research has to do with the typical way in which an individual tackles the cognitive problems with which he is faced. These are not new concerns, but uatil recently they have perhaps been less the concern of the psycho!ogist than of the literary critic. B~net, however, was clearly interested in both the problem of 'style' and of 'bias'. REEV~ (1965) commenting on Binet's treatment of thinking, points out that Binet's analysis of the performances of the two 'high speed human calculators', Inaudi and Diamondi, led him to talk about individual differences in imagery, which we may equate with style, and the evidence of direction imposed by the prob'tent itself, or bias. In a symposium on 'Style m language" (SEBeOK, i960), Jenkins, speaking of "Psychologica~ approaches to the problem of style', concludes that 'what we hope to find is some characteristic of the individual some enduring quality of the man which would pop up in his poetry, in his novels'. A similar i:,_~erest in variability of approach to a cognitive 353

Upload: george-shouksmith

Post on 26-Aug-2016

216 views

Category:

Documents


4 download

TRANSCRIPT

Aeta Psychologica 31 (1969) 353-364; © /¢orth-Holland Publishing Co., Amsterdam NOt to be reproduced in any form without written permission from the publisher

PERSONALITY ATTRIBUTES ASSOCIATED W~TH TWO

MEASURES OF COGNITIVE STYLE

GEORGE SHOUKSM1TH

Department of Psychology, The Queen's University of Belfa.~t, Ireland

ABSTRACT

A new ol3en-ended, abstract task which has n, true solution is t~sed to produce two measures of cognitive style. The task consists of a number of drawings, each of which shows two sticks which are partially buried in the ground. From a consideration of the choice pattern of subjects who are asked to suggest which stick has the longer total length, two style measures may be derived. Meast~re 1 assesses individual non- conformi~.y with the group approach to the problem. Measure 2 is a positionv.l score, reflecting the effect on judgment of the relative positioning of the ~wo sticks, qhe nature of these measures, in relation to certain group and pcrsonality variables, is investigated more fully. It ~s cor~cluded that people adopting a particular style vary in their personality characteristic:; by sex and from group to group. These differences are discussed and some general conclusions presented, relating the 'abstracting', consistent style to certain personality attributes commonly associated with t~e imaginative, creative thinker.

A ,~umber of recent studies reflect a growing interest in non-cognitive factors which affect cognitive functioraing. The development of research into the nature of creativity is, in one sense, an example of this trend.

In essence this field of research has to do with the typical way in which an individual tackles the cognitive problems with which he is faced. These are not new concerns, but uatil recently they have perhaps been less the concern of the psycho!ogist than of the literary critic. B~net, however, was clearly interested in both the problem of 'style' and of 'bias'. REEV~ (1965) commenting on Binet's treatment of thinking,

points out that Binet's analysis of the performances of the two 'high speed human calculators', Inaudi and Diamondi, led him to talk about individual differences in imagery, which we may equate with style, and

the evidence of direction imposed by the prob'tent itself, or bias. In a symposium on 'Style m language" (SEBeOK, i960), Jenkins, speaking of "Psychologica~ approaches to the problem of style', concludes that

'what we hope to find is some characteristic of the individual some enduring quality of the man which would pop up in his poetry, in his novels'. A similar i:,_~erest in variability of approach to a cognitive

353

354 ~ SHOUKSMITH

task can be seen in a number of fields, however. Jenkins is concerned to analyse the way that a particular artist 'characteristically modulates, modifies, or changes standard art form'. Harway (TRAPP and HI~U~L- sr~IN, 1962) is concerned with the variability of responses of orthopedic- ally handicapped children in a level of aspiration task. The results of Harway's studies reveal seven distinct patterns of approach and also that handicapped children differ from the normal in having a higher variability of responsiveness to success and failure.

Tlae term cognitive style may be used to refer to the adoption by an indi~,idual c,f a certain strategy or group of strategies in his approach to problem situations. In the sense in which it is used here, the term cognitive style includes all those phenomena which have previously been referred to as 'set~' or 'thinking modes' or the like. A variety of cognitive styles may be distinguished, some of which may be measured experimentally (e.g. WITKIN et al., 1962). In the present research, two measures of cognitive style were derived from an open-ended, abstract task which has no true solution. An insoluble problem series was chosen, since it seemed likely that if no real, logical solution was possible, the responses to the problem would reflect the individual's natural approach to thinking, the factors operating to determine the man,~er and direction of his thought processes. The prc~blem series used cor~- sisted of a number of drawings, each of which showed two sticks whch had been partially buried in the grc~und, so that only their top parts were; visible. Subjects were asked to say wh]ch of the two sticks tt~ey thought would be the longer, if they vcere taken out of the ground to reveal their full lengths. It was stressed that one could not work out the answer mathematically, as there were no right or wrong answers. Interest was centered on the patterning of responses to these pseudo- problems, which it wa.s anticipated would produce scores reflecting aspects of the individual subject's style in ~pproaching a cognitive task. In particular, two specitic style scores were isolated, as follows:

Measure 1. Preliminary studies with a series of sticks drawings revealed that a regular pattern of choices was produced by subjects within a. given sample. A measure of the degree to which a certain sub- jec~ conforms to th~s pattern may be obtained by counting the number of occa~,ions on which the subject's choice differs from the group choice. when the group choice itself differs significantly from chance. This may be regarded a~ a measure of individual non-conformity with the group approach to the problem.

SOME PERSONALITY ATTRIBUTES OF COGNITIVE STYLE 355

Measure 2. If each pair of sticks is re-presented with its left-right positions in the drr~.,ving reversed, a further style measure may be ob- tained. By counting the number of times a subject shifts his choice on the re-presentation a second score may be derived, representing the degree og~ inconsistency the subject shows in his stimulus choices. Thi,; may be a positional score, reflecting the effect on judgement of the re- ative positioning of the two sticks.

To investigate the nature of these style measures we compared szore:; obtained from three different samples, of approximately the sam~:~ agc: range, but from different populations. These scores were used both to identify group and sex differences which occur in approaches to think- ing and in an exploratory study to attempt to isolate personality character- istics associated with the two style measures.

METHOD

Subjects Three samples were used, one drawn from first year University ;tudents,

/he other two from first year College of Education students. The first sample, sample A, consisted of 216 first-year students of p3ychotogy at The Queen's University of Belfast. The sample was drawn from both Arts and Sciet.ce faculties in the ratio of about 2:1. There were I04 males and 112 females in the sample. The second group, sample B, was the first drawn from a College of Education. This consisted of 103 first year students, 29 males and 74 females, all of whom weze pu-~suing a three-year course of teacher training. This sample was drawn f~'em the first year group at Stranmillis College of Education, Belfast. A random ~ample was selected by choosing every third name on the class lists, males and females being treated separately, The first two sa~r~ple.q, were, therefore, both drawn from Northern Ireland and represent two different student populations. In our third sample, we tried to ct~oo~e a student group which would equate with one of the Northern ~Ir~;land samples, but which would represent an English population. The most suitable population offered for comparison was "hat from Edge Hill College of Education, Lanczshire. This College, like Stranmillis, draws its students from urban and rural backgrounds on a predomir~antly regional basis. One third of the fi"st year group were randoml, y selected by the staff of the College, producing a sample of 44 studt~nts. This formed our third sample, s,~.mple C. Unfortunately, because of the divi- sion of sexes in the College as a whole, this meant that only 11 of the

3 5 6 ~3. SHOUKSMITIt

44 were male students. Administrative difficulties made it impossible for tile College to provide a larger sample.

The "stic/cs" problems

Ti~:; stiel,:s problems may be described as a series of insoluble non- verbal problem situations. Each problem i~ presented in the form of a drawing representing two sticks shown embedded in the ground. The drawings are schematic, consisting of a horizontal line in green to represent the ground and line drawings of those parts of the sticks which are above ground, in black, and with cet~t~'es 10 cm apart. The sticks represented in these drawings are of t~vo lengths above ground, two thicknesses, and are drawn in one of two positions in relation to the ground line, Long sticks are 10 cm along their own centre line and shot! sticks are half this length, i.e. 5 cmo Thic~.~ ~ticks are 2.5 cm wide end narrow sticks only 1 cm wide. Sticks of tile various ~;~7~ are drawn in one of two positions, either upright or sloping to the ief~ at an angle c~f 45 degrees. The above series of eight ,~tick,; are presented in all com- binations of two,, forming a series of 56 drawings. The 'sticks' used were chosen on the basis of a pilot study which showed that they formed a meaningful, scalable series and that they produced sufficient variability in choice patterns to permit scoring of the two style measures described earlier.

The A/lport ascendance-submission scale.

Thi~,; tc~t measures the tendency to seek adju:~tment through ascenda.ace or dominance in social situations.

7t2e Maudsley personality inventory

Tt:eeretical c:onsiderations ,,;uggested that those subjects responding in a restrictive way, that is in a conformist and positionally consistent manner, might well be anxious; groups. Thus ~he two low groups might

e~pected to score more highly on the Maudsiey scale for neuroticism. The use of the MPI also allowed us to relate the two cognitive styles to extroversion.

The G,gilford-Zimmerman temperament sc.~edule

The Guilford-Zimmerman inventory is a factored test measuring the follo ring: general activity (G), restraint (R), ascendance (A), socia- bility (S), emotional stabili~:~¢ (E), objectivity (O), friendliness (F),

SOME PERSONALITY ATTRI[~U'IES OF COGN,~TIVE STYLE 3~7

thoughtfulness (T), personal relations (P), masculinity (M). It provides

good general assessments o f var ious aspects of personali ty which accords

with the explora tory na~are o f the present study. Specifically, it provides

addi t ional measures o~' bo th ascendance and extroversion. In the case

o f the latter, Gui l ford and Z i m m e r m a n adopt an American approach

which, unlike that o f the M P I separates the various c o ~ p o n e n t s of

extroversion and measures them independently.

RESULTS

The first results compare performances o f the different g roups The

mean score for measure ~, or non-conformity , for each sample is shown

in table 1 together with the standard deviations, sample sizes and an

addi t ional item, the number of drawing~ for each sample on which the

style score is based. Similar statistics for style m . - s u r e 2, or incon-

sistenc~, are given in table 2.

TA)~ LE l

Number of significant drawings, means and standard deviations of mea,~,ure 1 scores for each sample.

Samples A B C

Males Females Ma'~es Females Males Females

Non of sig. drawings 33 40 20 33 6 6

Mean score 10.13 15.27 5.07 11.31 _ 1 1.79 SD 7.18 8.50 3.38 6,02 __ t 1.51 n 104 112 29 74 11 33

1 Sample ~ze and number of significant drawings too small to produce meaningful statistics.

TABLE 2

Meaas and standard deviations for measure 2 scores in the three samples.

Samples A B C Maies Females Males Females Males Females

Mean Score 3.13 5.84 8.76 8,09 - - 9.97 SD 2.22 2.29 4.26 3.75 - - 3.75 n 10,7 112 29 74 11 33

358 ~. SttOUKS/VIiTrl

The relationship between the two style measures was also calculated in the three samples. For male University students the correlation be-

tween measures 1 and 2 was r = 0.574 and for females, r = 0.534. Comparable figures for tt~e two College of Education samples were, sample B, males r = 0.516: females, r = 0.211 and for sample C~ females,

r-----0,181. The small n for th,: group of male,,; in sample C prevented

~Jur calculation ,of that eorreJation. The MPI and the a scendance-suh.nis~ion scale were administeled

• Lo sample A only. Groups h;gh and low on the two style measures, non-confbrmity and inconsistency, were formed for both males and females, by dividing the total sample at the median on those measures,

The me,Jian scores of the measure 1 groups on the A-S scale and on

the MPI neuror:icism and e×troversion factors were compared, the median test being used to establish the ,~¢ignificance of differences found between the:~e group scores for each dependent variable. The same proce- dure we, s followed for the high and low groups, distinguished in terms

of measure 2. The results of this analysis are summarized in table 3.

TABLE 3

Median sco,res of the four cognitive style groups on the MPt variables and the A-S scale.

Me~e~ure 1 Non-conformity Measure 2 Inconsistency ~Tales Femate~ Males Females

Variable High Low l-iigh Low High Low High Low

A-S Score ---0.9 --1 ] .4 t ---3.5 --5.5 --0.5 --9.3 t --0.5 -4).75 MPI.-N 12.2 t5.71 14.25 15.5 12.2 14,4 14.0 15.0 MPI--E 15.2 11.5 14,0 14.7 14.5 14.3 14.3 14.5

1 Indicatez a pair of scores which differ significantly at the 0.05 level at least.

The Guilford-Zimmerman scale was administered to sample C and a median-split technique again adopted to produce groups high and

low on each of the style measures. Scores obtained by subjects in the

i~igh group on each personality factor were compared with scores for the low group in that style measure. The M~nn-Whitney lJ test was used to test the significance of differences found on scores for each variable for t1~c two groups. S i n ~ this was an exploratory study, we adopted WALLACE: and KOOAN'S (1965) procedure of ac:epth~.g a differ- once o f p < 0.10 for extracting significant factors. Tables 4 and 5 show

SOI~:IE PER~ONALITY ATTR;iBUTES OF COGNITIVE ;qTYLE 359

those facto;s , scores for the g roups concerned , significar~ce levels and

the r aw scores giving -he 50 ~ tile on the A m e r i c a n n o r m s for the test.

TABLE 4

Guilford-Zinamerman factors producing significantly different scores in high and lov groups on the t,vo styie measure~ ~ for females in sample C.

Measure 1 Non-conformity Measure 2 lncont~istency Personality f'~ctors R S T R A T

Mean ~ore for High group 13.6 14.8 18.7 Low group 16.9 17.2 21.4

p of diff', being zero 0.079 0.082 0.079 Test norms f,31 50% tile 16 20 18

14.4 15.4 19.7 17.9 12.2 21.4

0.049 0.028 0.076 16 14 18

TABLE 5

G,ailford-Zi.nmerman factors producing significantly differet~t scores in high and low groups on the two style measures for males in sample C.

Personality factors

Mean score for High group Low group

Measure 1 Non-conformity Measure 2 Inconsistency F T A

7.2 19.6 10o8 t4.6

11.2 15.0

p of diff. being zero 0.089 0.063 0.063 Test norms for 50% tile 14 18 16

(No i n f o r m a t i o n is p resen ted here regard ing the re la t ion o f cognit ive

style scores to persona l i ty var iables in sample B. Results f rom sam.+-"

B are inc luded in the pre,.~ent s tudy so tha t more extensive group com-

par i sons may be ~nade. In a la ter repor t , it is hoped to invest igate cer ta in

additiona'~ persona l i ty cor re la tes o f style in this group. )

DISCUSSION

The n o n - c o n f o r m i t y score measures the exten~ to which a subject

devia tes f rom the g roup n o r m ; the g roup i tself m a y be highly con fo rm-

mist , or to a vary ing degree, n o n - c o n f o r m i s t i~1 i ts th inking. Thus, the

m e a n n o n - c o n f c r m i t y scores seen in table 1 ~re be t te r regarded as

measures o f the ¢ariabil i ty of the g roup on ~.he, style measure in quest.-

360 G. SHOUKSMITH

ion, or measures of the extent to which the group permits divergence from its own norms. If any measure of the overall conformity of think- ing style cf the group can be obtained, it must be found in the first row of table 3, which indicates the number of drawings which produced conforming choices in each group. A general consideration of the ~rend of these figures would suggest that this is tr.ot all they measure, however. S~nce the number of significant drawings emerging is roughly correlated with the size of the sample one may argue that this latter factor is atso involved. How this factor of sample size operates, or how it interacts with conformity, it is not possible to say. One might tentatively conclude that Northern Ireland students are more conformist in thinking than English s~:udents, a conclusion whzch would not contradict general impressions of the Province! In addition, one might further tentatively suggest tl~Jat female groups are more conformist in their thinking than male groups.

~It is i~Jteresting to note that the two University groups produce a greater m~mber of significant drawings than. the equivalent College of Education: groups. If we could 'partial out' the influence of sample size, however, this d~fference might not be significant. Indeed, we might find that ~he males in sample B, who produced 20 significant drawings in a ,;mall group ,.gf 29 were the most conformis~., thinkers. It can be seen from the foregoing, that measure 1 is a much tess simple measure than it at first appeared to be. It is best regarded as measming the extent to which a subject deviates from the norms of his or her own group. Since it is based only on those drawings for which the group as whole achieves significant consensus of choice, and no~ on all drawings present- ed, it is probably inappropriate to use it to compare the general level of conformity between groups. Measure 2, however, is derived always from the responses to the complete set of drawings. Hence it provides a more stable indicator of group differences. From table 2, it can be seen that University groups of both sexes are relatively lower on this measure. Possibly the abstract rnaterM involved in the "sticks" problem is more meaningful to the Universi'!.y student, particularly the males, than it is to other ~oups. Thus, they respond to individual sticks, irrespective of their positions, whereas other groups respond to the material in a more rela6onal manner~

The relatior~ship between the two m~:asures was also investigated. If any reliance ca:a be placed on tl~Le result for females in sample C, these results would suggest that the *w~ style measures do not form

SOME PERSONALITY ATTRIBUIES OF COGNITIVE STYI.E 36~

related thinking modes in female Coll,:ge of Education students. Amongst University students of both sexes and male College of Education stu- dent.,; in Northern Ireland, ho~'ever, there appears to be a reasonably stror~g common element in the two cognitive styles.

Turning to table 3, and the investigation of personality correlates of these nieasures of coglfitive style, we find that this difference in meaning between n:~ak~s and fem~.les ~,s confirmed. The only significant differences between high and low groups appear in the male groups. Male U~lider- sity students who are non-conformist in th0ir approach to thinking, as defined here, tend to be less svbmissive than their con~'ormist counter- parts. It is also noteworthy thal male subject,~ adopting an 'inconsist- ent' or relational style a~-e significantly less submissive than the conl'orm- ist male students. It seems better, in the case of both variables, to ~;peak of the high groups as being 'less submissive' rather tha~l 'more dorni. nant', in terms of the original scoring standards for the A-S scal:e the entire sample is somewhat more submissive than average. The results for the female groups show no significant differences for A-S scores in the various cognitiwe style gr~ulc, s. As the scores in table 3 ~how, however, the non-significant differences for the female group axe in the same direction. When we turn to th~ relationship of cogrfitive style ~o Neuroticism as measured by the MPI we find that although only one difference between pairs of scores is significant, all differ~:nces are in the predicted direction. The MPI may not be sufficiently sensitive to reflect differences on this factor. It was the anxiety comvonent of the factor, as a measure of drive, which it was hypott~esised would relate to the two cognitive styles. The overall trend of the figures may be taken to suggest that the conformist-consistent thinker has a higher level of drive, or even anxiety, than the non-conformist, relational or inconsistent thinker who scores highly on both style measure,~. Perhaps the former cares more about doing the 'right thing'.

Somewhat surprisirgly, no significant differences are found between the extroversion scores of the various cognitive style groups. ]'here is some suggestion that male subjects who score highly on measure 1 tend to be more extroverted than the low, 'non-conformist' group, but the difference is not significant (p = 0.15). A further finding is that there are major sex differences in personality attri0utes associated with cognitive style. I t appears that the two cognitive styles a~'e less closely re~ated to uederlying personality dimensions in the female group. It may be thz~t female students are more 'field dependent' !n

362 o. snoogsMtrn

I~ITKIN$~S (1962, op. ci0 ~ense, and respond variably in terms of the situation without adopting a particular style. Alternatively, the male- female differences found here alay be simple reflections of differences in patterning of responses to the sticks problem series, outlined pre- viously.

Furtk.er insight into the nat~rre of the two measures is given in the results of the later study with sample C, using the Guilford-Zimmerman inventory. Amcagst females ill this sample, thinkers whose cognitive style is, in our terms, predomi:aantly conformist tend to be more re- strained and se.rio/ls, more socially minded and more thoughtful and reflective 1than the 5~on-conformist". Those subjects with a high measure I score tend to De impulsive and to be. social isolates. Females scoring high and low on style Measure 2 also produce significantly different scores on three Guilford-Zimmerman factors. Two of these, restraint (R) and thoughtfulness (T) are the same and t te third is factor A, ~.czqid~mce. From table 4, it can be seen that low measure 2 thinkers, referred to as consistent thinke:rs, tend to be more restrained and serious, more submissive, and more thoughtful and reflective than are 'incon- sistent' thinkers. Conversely subjects with a high score oa measure 2, presuraably those affected by ttae relational positioning of the sticks in the drawings, tend to be more dominant and assertive, more impulsive and less reflective. One might h~ypothesize that they respond in manners similar to the 'field dependents' of WITKIN's (1962, op. eit.) analysis.

Amongs~ the males in sample C, 'two factors, friendliness (F) and thoughtfulness (T) produce differences in scores between those high and low on style measure !. Table 5 suggests that all males in the sample are distinctly below the American norm for friendliness. Non-conform- ing males, however, are signi:ficantly less friendly and agreeable and are more hostile and belligerent. An interesting sex-difference emerges in respect of factor T. This factor operates in opposite directions in the two sex groupz. Whereas males adopti~g a non-conforming style tend to be thoughtful and rette~tive, females adopting the same style do not. Diverging from the approach of the group, as assessed by meas- :u'e 1, appears to mean diflerent things ia the two sexes. A similar sex- di~%rence operates on the sob factor related dift~rentially in males to style measure 2, factor A. On average, males in this group are less domina~.t than the American norm, but high measure 2 stylists are more submissive than the consisten~t ones.

At this stage we naa'¢ also draw certain tentative conclusions about

SOME PERSONALITY ATTRIBUTES OF COGN!TIV2 STYLE 363

the nature of the tv, o style measures. High scorers on measure 1, for example, appear to be typical of the thinker who diverges from his group :in respor.se tendancies. He is in this respect non-conformist in his approach t~ problems. He may well act'fieve this non-conformity by simply avoiding the obvious in choice sitt~ations. Major inter-g'roup differences occur in the level of conformity of choice and this measure is affected by these differences. Measure 2 emerges as a potentially more important measure of style, contrasting two kinds of thinker,s. At one extreme, the high scorer appears to represent a subject concerned with the o"erall ioatterning of situations, or responding in terms of the position of a stimulus. The low scorers on this measure produce more consistent approaches to problems in that tI:iey choose a stimulus irres- pective of its relation to other stimuli in the total situation. The 'consi ~ter~t' thinker appears better able to abstract ~nd hold on to a relevant stimulus in the situation, to center or focus as WERXnEI~ZR 11945) puts it.

CONCLUSIONS

Style measure 1 assesses the extent to which a subject picks an obvious stimulus, or an unusual one, in choice situations Measure 2 corLtrasts the subject who is influenced in his thinking by the whole situation with the one who can abstract and hold ~n his mind certain specifio features of the total and who responds to the abstracted elemciit, rather than the total situation.

One of the major conclusions to be drawn from the present studies is that people adopting a particular cognitive style vary in their personality characteristics by sex and from group to group. Conformity of approach or a eog-itive style placing emphasis on relation3 inherent in ta'oblems, mea'~ different things in a male group and a female group. Samples coming from different secondary groups show variations in personality dimensions re:ated to cognitive style. In general, however, the results suggest that the adoption of a consistel~t style implies a careful, deliberate and persistent personality which is opposed to the unrestrained, sometimes impulsive approach of the 'inconsistent' thinker. To a lesser extent, this element of self-control is also present .;~I t,~?o,~mist thinkers. It is interesting to note that the nature of this low pole core ele~ment is remarkably similar to CATT_ELL'S (1963) pen picture of the personality attributes of the imaginative, creative thinker.

~6~ G. SHOUKSMITH

CAT~LL, R. B., i 963. The personality and motivation of the researcher from measure- men~ of contemporaries ~md from biography. In: C. W. Taylor and F. Barton (eds.), Scientific cr~'.ativity. New York: John Wiley.

R~vEs, J, W., 1965. Thinking about thinking. London: Suker & Warburg. SES~OK, T. A. (ed.), 1960. Style in language, Cambridge, Mass.: The M.I.T. Press. T~PP, E. P. and P. H~!LSTHN (e(~).), 1962. Readings on the exceptional child.

London: Methuen. WA~ L^*.a~ M. A. and N. KOGAN, 1~,65. Modes of thinking in young children. New

York,: Holt, Rinehart and Winston. W E ~ ' t - ~ R , M., 1945. Productive ti~inking. New York: Harper. Why.IN, H. A., R. B. DYe:, H. F. FA~'ERSON, D. Ro GOOI3ENOUGH and S. A. KAaP,

1962. Psychological differentiation. New York: John Wiley.