Th e Picture Completion Test - Forgotten Books

mut ationa l flagrgnmgg fimmg aphs

'h im by Guy finntrnat mum'

No . 20

Th e P icture Com pletion Test

RUDOLF P INTNER AND MARGARET M . ANDERSON

Oh io S ta te Un iversi ty

W

CRWICK YORK , INC .

BALTIMORE , U . s. A .

1917

19 17

By Warwi ck York . In c .

EDITOR’S PREFACE

P rog ress in mental testing has been greatly reduced ,as all workers in that field are well aware

,because of

lack of standardization both with respect to the methods of administering tests and also with respect tothe methods of scoring them . Closely allied withthese lacks i s the fur ther lack of adequate norms ofperformance

,gathered from a sufficient number of

cases to indicate not merely the average performancebut also the entire range of performance as conditionedby age

, sex, school training and what other factorsmay be influential .The present monograph supplies these desired data

for a single test— the Healy Pictur e Completion Test .By apply ing it to over 1500 children the authors havebeen able to arrive at an empiri cally determi nedmethod of scoring

,to establish norms of percentile

distribution for each age from 6 to 14 and incidentallyto ascertain the connections between performance init and sex

,social env ironment and school standing.

Work of thi s sort is directly valuable to all personswho make use of mental tests with children and it isequally valuable to those who are interested in thetechnique of standardizing tests and in working outby empirical methods the best rules for administeringand scoring them . G . M . W .

CONTENTS

Ch apter I . INTRODUCTION

Ch apter II . THE TEST .

Chapter III . THE METHOD OF PROCEDURE AND THE SUBJECTSTESTED .

Th e Procedure

Ch apter IV .

Chapter V. THE DETERM INATION OF THE SCORESPercen tag e DistributionCorrel ation s .

Ch apter VI . NORMS .

Percen til esYea r Sca le Norm s

Ch apter VII . SEX, SOCIAL STATUS AND SCHOOL STANDING

Differen ces in Socia l StatusRa cia l Differen cesSchool Stan din g .

Chapter VIII . OTHER METHODS OF SCORINGHea ly

’s Meth od

Number of Right an d Wron g Moves

Ch apter IX.

Ch apter X.

CHAPTER I

INTRODUCTION

The growing demand on the part of clinical psycholog ists for a greater variety of tests to aid in mentalexaminations and for adequate standardizations oftests is leading to the introduction of new tests andto more intensive studi es '

of tests already in commonuse . The necessity on the part of the practical workerfor a number of tests and the relative easewith which anew test can be devised has resulted , in many instances,in the use of a test for practical work before anyreal analysis or standardi zation of the test has beenmade . To draw conclusions from any specific ‘perform an ce in a test before a real study of it has beenmade, is at best a very dangerous procedure, and thebasing of diagnoses of individual cases upon the resultsof tests that have not been carefully studied is one ofthe things which in some quarters has helped to bringclinical psychology into disrepute . If we are to use

a test for purposes of differential diagnosis, we mustbe certain that we know what is a normal performance .

We have no justification in saying that any specificperformance is a bad performance

,say for a six-year

old ch i ld , until we know what the average six-year-oldcan do . Wh at may seem to us as adults a very badperformance may be the common six-year-old responseto the Situation . In short

,the careful clinical worker

wil l not use for diagnostic purposes any test until ithas been adequately standardized .

The question of the adequate standardization ofmental tests has been until recently a somewhat negleeted phase in the psychology of individual differen ces. Previous workers did not attempt to set

standards, but were pI im ari ly interested in the resul ts

2 THE PICTURE COMPLETION TEST

of tests as applied to specific groups of individuals,

and the main emphasis was thr own upon the study ofthe test rather than upon the individual . It was thetest that was primarily under consideration . Furthermore, great interest was shown in the question as towhat the test was testing

,whether memory or imagin

ation , or other psych i cal process . This point of viewdominates the earlier work in the psychology of tests

,

such as the studies of Cattell and Farrand,1 Ki rk

patrick'Woodworth and Wells, 3 and many others .This point of view is

,of course, a necessary one and

naturally precedes the use of tests for practical diagnostic purposes . The results of the large amount of workwith individual tests of varying nature are best seen inthe compilation of Wh i pple, 4 and the arrangement ofthe tests into tests of sensory capacity, of attention, ofperception

,etc .

,is indicative of the point of view.

Paralleling th i s development of the psychology oftests

,and in recent times very largely overlapping , we

have the question of the relation of test to test , or ofthe relation of abil i ties as tested by any two or moretests . This is the question of the correlation of testsand the decided interest shown in th i s aspect of theproblem may be traced back to the work of Spearman, 5

1 Catte l l, J. MeK . , and Farran d, L. , Ph ysica l an d Men tal Measure

m en ts of th e Studen ts of Columbia Un iver sity . Psych . Ren ew, 3,1896 , 6 18

—648 .

2 Kirkpatrick, E . A . Individua l T ests of Sch ool Ch ildren . Psych .

Review, 7 : 1900 , 274—280 .

3Woodworth , R. S . an d Wel l s, F . L . Association T ests . Psych .

Review Monographs, 13 : 1911 , NO . 5 .

4Wh i ppl e, G. M . Man ua l of Men ta l and P h yswa l Tests . 2 Vo ls .

Warwick an d York,1914—15 .

“ Spea rman, C .,

Gen era l In tel l ig en ce, Ob'ectively Determm ed and

Mea sured , Amer . J . of Psych ,15 : 1904 , 2 0 1

—293 .

Hart , 3 , an d Spea i m an, C .

,Gen era l Abil ity , i ts Existen ce a nd

Nature, Brit. J . of Psych , 5 : 1912 , 51—79 .

Krueg er, F an d Spear-m an, C . ,

D i e Korrel ation Zwisch en versch i e

(

I

iian en g eistig en Leistun gsfah igkei ten ,

Zei tsch r . f .Psych ,

44 : 1907, 50

4 .

INTRODUCTION 3

although some work had been done before his time ,notably by Wissler

,6 and although the question had

been raised at the very beginning of the interest inmental measurements by Galton'h imself . The studiesbearing upon the correlations of tests have been numerous . Among such studies may be mentioned thework of Burt

,8 Simpson

,9Wh i tley

,1 0 Brown , 1 1 and others .

Although the results of this work are often con tradi c

tory, yet the point of view emphasized has thrown agreat deal of light upon our knowledge of tests . It isa line of approach that raises very definitely thequestions what the tests are testing and what is themeaning of intelli gence itself .Neither of these two streams in the development of

the psychology of tests leads us directly to the problemof standardi zation with wh i ch we are here more nearlyconcerned . Both of them are fundamentally concerned with the test itself or with the psychi cal process it is supposed to be testing . A slight Shift inpoint of view from the test or the psych i cal process tothe individual tested leads to another line of development in the history of mental tests . Thi s appearedas soon as the question was raised what light perform an ce in a test throws upon the psychologicalmake-up of the individual tested . This slight changein point of view was a natural outgrowth of the in

6Wissl er, C . , Th e Correl ation of Men ta l an d Physica l T ests, Psych .

Revi ew Monographs, 3 : 190 1 , NO . 6 .

7Ga l ton , F .,Inqu iri es in to Huma n Facu lty a n d I ts Developmen t, 1883 .

8 Burt , C . , Experimen tal T ests of Gen era l In tel l ig en ce, Brit. J. ofPsych , 3, 1909 , 94

—178 .

9 Simpson , B . R ., Correl ation s of Men tal Abil ities,Tea chers Col l .

Con tr . to Educ . , No . 53, 1912 .

1 ° Wh itl ey , M . T ., A n Empiri ca l S tudy of Certa i n Tests for Indi vi dua l

Difi'

eren ces,New York, 191 1 .

l g

l

l

l

l

Brown , W., Essen tia ls of Men ta l Mea suremen t, Cambridg e, En g ,


creasing interest in individual psychology . Binet ’swork1 2 on the different psychological characteristicsof individuals or groups of individuals led him to applytests for differential purposes . We have from h i m

and h i s co-workers,Henri and Simon

,a long series of

studies in which single tests or groups of tests are usedfor the analysis of individuals . This work culminatedin the construction of a scale of tests for practicaldiagnostic purposes . He originated the . idea of systems of tests with age-gradations

,which in turn has

led to the idea of norms of performance in practicalclinical work . Norms are demanded for comparativepurposes . On e normal standard is not enough forchildren , because every age must have its separatestandard . The only means of measuring the mentalretardation of a ten-year-old child is to compare hisintelligence with that of normal children of differentages or of many ten-year-old children .

Vari ous attempts have been made to establish sucha working basis with different sets of tests

,but in most

cases the investigators seem to have lost sight of theirend in the desire to put their material to practical use .

Thi s very fact has resulted in many false conclusionsand wrong judgments . No more striki ng examplecould be found than that of the Binet-Simon scale,itself. In the first place

,the separate tests were not

“ tried out with large enough groups of ch i ldren, and ,as a result

,many were misplaced . In the second place,

lack of a standard method of procedure has led to difficulty in the comparison of results of different workers .This lack of adequate standardi zation was felt immediately after the scale was put into use and led to

12 Bin et, A . A tten tion et Adaptation ,L

’année psychologique, 6 . 1899,

248—404 . Bin et, A . et Hen ri,V . La Psych olog ie In dividuel l e, L

’annec

psych ol , 2 : 1895-6 , 411-465 .

INTRODUCTION 5

the re-standardization of the scale by Goddard .

13

This was the starting point for a discussion of standa rdi zati on in general

,and has given us the work of

Kuh hn an,14 Bobertag ,

1 5 Stern,1 6 Terman

,1 7 Otis, 18 and

others . All this work was done mainly from the pointof view of the scale of mental measurement

,and the

discussion centers around the adequate placing of atest at a specific age . The appearance of other scales ,such as those of Knox

,1 9 de Sanctis

,2 0 Yerkes-Bridges , 2 1

and Terman 2 2 served still further to emphasize theproblems of standardization . At the same time

,the

use of special tests devised by different workers andused by them in practical clinical work led to the n ecessi ty of norms in order to evaluate the performance ofany individual .With this growing demand for age-norms appearedthe elaborate work of Sylvester2 3 with the Seguin

13 Goddard, H . H ., Two Thousan d Ch i ldren Measured by th e Bin et

Mea surin g Scal e of In tel l ig en ce, P ed . S em . , : 1911 , 2 32—2 59 .

14Kuh lm an , F .,T h e Resul ts of Gradin g T

8

h i rteep Hun dred Feebl em in ded Ch ildren with th e Bin et-Sim on T ests, J. of Educ . Psych 4 :1913 , 2 61

—2 68 .

1 5 Boberta g , O . , Ueber In tell i g en zpruefun g en ,Zei tsch r . f . an ge

wandte Psych ologie, 5 : 1911 , 105 ; an d 6 . 1912 , 495 .

1 “ Stern , W , The Psych ologi ca l Meth ods of Testin g In tel l i gen ce, T rs.

by Whippl e, Ed . Psych . Mon ograp ,h s No . 13 .

1 7 T erm an , L . M ., an d Ch il ds

,

p.H G. A T en tative Revision an d

Exten sion of th e Bin et-Sim on Measurin g Sca l e of In tell i g en ce, J . ofEduc . Psych , 3

° 1912 .

13 Otis, A . Som e Log ical Aspects of th e Bin et Scal e, Psych . Rev., 2 3 :1916 , 129

—152 , 165—179 .

19 Kn ox H . A . A Scal e, Based on th e Work at El l i s Isl an d, for

Efgim atin gg Men tal Defect, J. of Amer . Medi ca l A ssoc ., 62 ° 1914

, 741

2 0 de San ctis , S . Men ta l Developm en t an d th e Measure of th eLevel of In tel l ig en ce, J . of Ed . P sych .

, 2 . 191 1 , 498- 507.

21 Yerkes, R . M .,Bridg es, J. W. ,

syan d Hardwick

, R . S ., A P ai n t S ca le

for Mea surin g Men ta l A bi lity, Warwick an d York, 1915 .

22 T erm an , L . M . The Mea suremen t of In tel l i gen ce, Hough ton ,

fl in Comppany , 1916 .

15

23

1

Sgll

z

i

g

rester , R . H ., Th e Form-Board T est, Psych . Revi ewMonograph s,


Form-Board,who with some fifteen hundred children

worked out very reliable norms . P in tn er’s2 4 revision

of the Knox Cube Test and Wallin ’s2 5 norms for theSeguin Form-Board are further indications that theneed for more reliable standards in intelligence testingis being met . Other indications of the need for normsappears in the work of Schmitt

,2 6 who arrived at tenta

tive norms for many of Healy’s tests

,and in the work

of the Bureau of Investigation of the New York StateBoard of Charities . 2 7 In the latter study the PictureCompletion Test is included and reference to the resul tswill be made in a later chapter .The task of standardization probably falls under

two main heads , (1) the determination of a method ofpresentation and method of scoring

, (2 ) the establishment of norms . Under the first heading is included aspecific mode of procedure that will be common to allinvestigators and likewise a specific method of recording and scoring resul ts

,so\th at the results from differ

ent Workers may be comparable . A change in methodof procedure in giving a test may alter entirely theSignificance of the test . As Wh i pple2 8 has well said,“ No detail in the setting of a test is too trivial to beneglected . It is noteworthy that the lack ofaccordance between the results obtained by differentinvestigators in the use of what is ostensibly the Sametest almost invariably turns out to be due to seemi ngly

24 P in tn er,R .,

T h e Stan dardization Of Kn ox’s Cube T est . Psych .

Review,22 : 1915, 377

-40 1 .

2 5Wa l l in , J . E . W . , A g e Norm s Of Psych o-motor Capaci ty , J . of Educ .Psych , 7 : 1916 , 17

—24.

2 “ Schm itt , C . , Stan dardization Of T ests for Defective Ch ildren ,Psych . Review Mon ograph s, 19 : 1915, No . 3 .

2 7 New York State Board of Ch arities , El even Men ta l T ests Standar i zed , Eugen ics and S ocia l Welfa re Bu l letin , NO . V,

1915 .

28Whippl e, G . M . A Man ua l of Men ta l and Ph ysi ca l Tests, Vol . I ,p . .5

INTRODUCTION 7

trivial variations in the method of admi nistering thetest .” The second heading , the establishment ofnorms

,includes two possibil ities

,either the establish

ing of average or median performances at each age,i . e.

, the so-called age-norms , or the determination ofperformances for any number of percentile groups ateach age . The former is the method adopted by Sylvester

,Wallin

,and P i n tn er in the articles mentioned

above,wh i le Woolley

,2 9 in her work at the Bureau of

Vocational Guidance at Cincinnati, has preferred thepercentile method

,according to which the ch i ld

‘

iscompared with children of the same age . The agenorm assigns a mental age to a ch i ld ; he is either at ,above or below h i s chronological age . His performanceis compared to the average performance of anotherage-group . For example

,a ten-year-old child may be

spoken of as equalling in a test the average performanceof Six-year-old children . Th i s is the method madefamiliar by the Binet system of mental ages . Thegrowing criticism of th i s method is based upon thetheory underlying the growth of intell igence in general , whereby we know that growth of intelligence doesnot advance by equal stages corresponding to the equalstages of chr onological growth . No harm is done so

long as we keep strictly to mental ages,but con fusion

immediately arises when results are expressed inamounts of retardation as determined by the differencebetween the chronological a n d mental ages . Tryingto avoid this has led to the Intelligence-'uotientmethod of Stern , the Coefficient of Intelligence of

29 Wool l ey , H :T . , an d Fish er, C . R .

, Men tal an d Ph ysica l Measurem en ts of Workin g Ch i l dren , Psych . Review Monograph s, 18 : pp . 247.

See a lso Wool l ey , H . T . A New Sca l e ofMen tal an d Ph ysica l Measurem en ts for Adol escen ts an d Som e of Its U ses, J. of Educ. Psych , 6 :1915, 52 1

—550 .


Yerkes, and the Percentile Method as suggested byWoolley. The last two methods are based upon adirect comparison of any performance with the perform an ces of chil dren of the same age .

The percentile method allows the comparison to bemade with groups of ch i ldren of the same age

,

‘

sincethe percentile scores give us the distribution of theability of each a g e

-group . It is thi s latter methodthat has been used in the computation of the resultsof this study . At the same time

,the fifty percentile

gives us the norm,or average performance

,for each

age, so that age-norms are obtained . It is the belief

of the authors that ul timately the percentile methodwill prevail in standardization . The drawbacks ofthe method at present are that we require a large number of cases at each age i n order, with any degree ofcertainty

,to mark off the percentile limits

,and further

more,we do not yet know what Signi ficance is to

attached to a ten-percentile or twenty-percentile orany other percentile performance . It will not takelong

,however

,for us to learn to think in terms of

percentile abilities .

CHAPTER II

T HE TEST

Thi s test was devised by Healy1 through the desireto secur e one which would involve the principle of theEbbinghaus Combi n a ti on smeth ode and at the sametime elim inate the language factor. The EbbinghausCompletion Method is now largely used for languagetests and it is proving to be one of our best methodsfor measuring language ability.

2 It seems also to beh i ghly correlated with well-known tests of generalintelligence . It seems reasonable to suppose thatmuch the same sort of ability is required to complete apicture as to complete a sentence . In both cases theessential element is the noticing of something lackingin the general situation and the supplying of a mi ssingpart to complete the general scheme . In the PictureCompletion Test

,however

,the choice of a mi ssing

part is limi ted to the blocks supplied to the subj ect,whereas in the language-completion tests in commonuse the subj ect has the whole range of his vocabularyfrom which to supply the missing word . A directanalogy to the Picture-Completion Test would be alanguage-completion test in wh i ch the subj ect wassupplied with a limited number of words from whichhe must select the word best sui ted to complete thesense of the sentence before him .

The material consists of a picture,brightly colored

,

measuring 10 by 14 inches (see Figure It repre

1 Hea ly , W., A Pictoria l Compl etion T est , Psych . Revi ew,2 1 : 189

20 3,19 14 ; an d Heal y W . an d

pFern a ld, G . M .

, T ests for Pra ctica lMen ta l C lassification ,

yP sych . Review Mon ographs, 13 : 1911 , NO . 54 .

2 T rabue, M . R ., Compl etion T est Lan guag e Sca l es . T ea ch ersCol l ege, 1916 .


FIGURE 1

https://www.forgottenbooks.com/join


for a choice reaction . There are fifty blocks fromwhich to choose and the unfini shed situation can becompleted in a great many

,

different ways . Thi s givesus the possibility of measuring different deg rees ofadequacy in responding to the situation .

Already a fair amount of work has been done withth i s test . Healy himself gives the results of some fivehundred cases

,including the following groups : (1)

1 10 children from a private school, (2 ) five unusually

bright young children, (3) 248 juvenile del inquents,

(4) 15 feeble-minded cases , (5) 95 Wellesley Collegestudents, (6) 33 psychopathic individuals , and (7) agroup of unselected intell igent adults . These results

,

whi le valuable in studying various possible types ofreaction

,are not sufficiently represen tative for the

establishment of norms .Healy says that the group of exceptionally bright

youn g'

ch i ldren did the test quite well . Three eightyear-olds were able to do the test in less than fourminutes with a perfect s core . This

,he says

,offers

“ most convincing proof of the validi ty of the test fornaive minds . ” Healy presents a table of his 1 10normal cases arranged accordi ng to age and schoolgrade

,Showing the range of total errors

,the range of

illogical errors and the range of time,as well as the

medi an total errors and the median illogical errors .Because he has di vided each age—group into sub-groupsaccording to school grade

,there are never more than

14 individuals in any one group . The medians, therefore

,are of little value . We may say here that Healy

divides his errors into two types,logical and illogical .

He has selected somewhat arbitra rily t en moves whi chhe calls logical errors and on this basis his evaluationof the performance i s based . A totally di fferent

THE TEST 13

method has been used in th i s study, and therefore theresults are not comparable with his .The performances of his delinquents correlated well

with their apparent mentality. In speaking of theresults obtained with the group of feeble-minded , thebrightest members of the school at Vineland, Healymakes the following statement :

' “ The feeble-mi ndedgroup

,as seen under two conditions

,tally well in their

extremes only. Nearly all make bad failures . It isobvious that rare individuals among them have devel oped the ability to apperceive such relationsh i ps asare demanded by this test

,even though on other levels

their mental equipment is demonstrably poor . This isas we

,who observe the special abilities and disabilities

of these mental defectives as well as normal persons,

would expect . It is not to be anticipated that anysingle test can be evolved which will discriminate thefeeble-minded .

An interesting factor is presented in the inferiorityof the performance of the Wellesley students whencompared with the school children and the higherclasses of delinquents . Here we find college studentswi th a median of two total errors

,wh i le the median

for both groups of chil dren is one . The percentage ofillogical errors is 64 in the Wellesley group

,which is

noticeable in comparison with 50 per cent . at the privateschool and approximately 36 per cent . with the del in quen ts. These college records

,which were secured

by Eleanor Gamble, of Wellesley, seem to show thatthis sort of test may not be reliable when applied toadults whose more varied experience leads to an un

limi ted number of complex responses wh i ch could n ot

be evaluated by the method of scoring used by Healy .

The psychopathic cases tested displayed the ex

pected variability, and Healy suggests the idea that


the test might become of possible use in diagnosingthe various types of insani ty by observing the methodof procedure and pecul iar reactions displayed by suchindividuals .In concluding his article Healy says : We evidently

have in our completion picture a test for ability primarily adapted to the child type of mind . Everydetail of the meaning has proved to be understandableeven by morons . The performance of naive indiv i dua l s Of ordin arily good intelli gence above ten yearsof age should be better than in five mi nutes , and notmore than one ‘illogical’ and two total errors shouldbe made . A worse record than th i s Should arousesuspicion of defect in mental ability It is onl y inth i s way

,therefore

,that Healy has standardized the

test . A presentation of our resul ts worked out a o

cording to Healy’s method is given in Chapter VIII .

Work with the test has also been carried out by theNew York State Board of Charities3 with differentgroups of subj ects . In all

,659 cases were examined,

but th i s included only 180 public-school ch i ldren ; theothers were from various corrective and protectiveinstitutions . Such results are interesting for comparison ,

but not essential for purposes of standardization .

A summ ary, nevertheless, may be given here . P rac

tically the same method of procedure was employedin th i s investigation as in Healy’s

,that is, the kind

and number of errors were recorded and notes weremade upon the method of work and the kind of reasoning used by the subj ects . Perhaps the most Si gn ificant table for our

‘ purpose is that showing the resultsof the public-school children . The following Shows

3New York State Board of Ch arities, El even Men tal T ests Standardi zed, Eugen i cs and Socia l Welfare Bu l letin , NO . V,

1915 .

THE TEST 15

th e average number of errors for thirty children ateach ag e :

A geAv . NO . errors

A comparison of these results with ours is given inChapter VIII . Here it is indi cated that Healy’snorm s are too high

,Since h i s standard al lows only

two errors at age 10 . Here,however

,we find t h e

average number of errors at 10 years to be andnot even at 12 years does the average fall as low astwo .

These results and the resul ts of our own cases indicate clearly the necessity for a more complete standa rdi za tion and a more critical study as to the bestmethod of evaluating the performance .

CHAPTER III

THE METHOD OF PROCEDURE AND THE SUBJECTSTESTED

Th e P rocedure

As Healy says, this is a test easy of presentation .

It appeals to the child’s interest,wh i ch goes far toward

eliminating such disturbing factors as fear and selfconsciousness . The child likes to do it

,and the whole

performance is not so long that the task loses its charmand fascination . While the subj ect is at work theexaminer has ample time to study his general attitudeand to observe any lack of coordination or seriouswant of poise that may be evident .The picture should be placed directly in front ofthe ch i ld with the fifty blocks arranged haphazardlyabove it as shown in Figure 1 . It was found that afew

’

words of explanation Should precede the performance . In this investigation the same instructions weregiven to every subj ect from the five-year through theadult group . Each was told to look at the picturecarefully and see what was happening , or what

”

thepeople were doing, or what activities were going on .

Fur ther, he was informed that any of the blocks wasjust the right size to fit in any of the spaces

,but that ,

Since there were more blocks than spaces , he mustselect the ones which seemed to h i m to be the verybest with which to fini sh the picture . In order tomake certain that the directions were ful ly comprehended

,questions were asked about one group until

the Situation was perfectly understood . Like Hea ly,we have taken for an example the wagon group asbeing one of the Simplest of the ten and we have keptrather closely to his words The examiner says to the

METHOD OF PROCEDURE 17

child : Look at this picture and see what is happening

,look what the people are doing . You are to fill

in these empty spaces so as to make the picture lookright

, so as to make the best sense . Any of theseblocks up here wi ll fit into any of these spaces . Choosethose wh i ch seem to you the best , those which willmake the best picture . Then

,pointing to the wagon

group,What is the man looking for ? Wh at is gone? ”

If the answer is “ the Wheel ,” then he says to the

child : “ That ’s fine . Find the wheel among the otherblocks and put it in . If the correct response is notgiven

,he prompts the ch i ld ; if that is not successful ,

he explains to him more fully the situation and tellsh i m to find the wheel and put it in . The correctanswer is

,however

,generally forthcomi ng and the

child is told : “Do the others in the same way as care

fully and as quickly as possible .

” The factor of speedis not emphasized

,but it is always mentioned . Wh en

every space is fil led and the subj ect looks up,or other

wise indicates that h e has fini shed , the experimentersays : “ Now look it over carefully . See if every blocksuits you . See if it is exactly as you want it and thentell me when you have finished .

” The time wa s takenwhen the child indicated he had finished and also afterhe had been allowed to make corrections .An exact record was kept of every move made by

the examinee . Although the recording would seem atfirst somewhat complex

,it is really quite simple . At

the top of the record sheet were written the names ofthe nine Spaces ; the tenth was disregarded since it isused as an il lustration . As each block is put in

,its

name is recorded under the proper heading unless ithappens to be the right block, wh i ch is then denotedby a check mark . If changes are made

,they

,too

,are

,

of course, recorded . If a change is made after the


exami ner has told the ch i ld to look over the picturecareful l y, i . e.

, during the last phase of the test, thenth i s is noted on the record blank by a horizontal lineingthe space where the change is made . Under thishorizontal line the change is recorded . A sample recordappears thus :

N am e

A g e and g rade .

This sample record is to be read as follows : The Curtain-Window block was put in the Broken-Windowspace ; the Dog block, i . e .

,the right block

,was put in

the Dog space ; the D . Cat (departing cat) block w a s

put in the Cat space and th i s was later on changedfor the right block ; the Baseball block was put in theFootball space and th i s was later changed for theFootball block

,i . e.

,the right block ; the right block

was put in the Hat space ; the Bucket block was putin the Basket space and in the second phase of thetest

,when the ch i ld was given an opportunity to make

changes,the Cherries block was substituted for the

Bucket block in the Basket space ; the Log and Ch i ckenspaces were fill ed in correctly ; the S . Bird (standi ngbird) block was put in the Flying-Bird space and inthe second phase of the test changed for the rightblock . The two horizontal lines Show us that the ch i ldmade two changes in the second phase of the test .

2 0 THE PICTURE COMPLETION TEST

The ages of the subj ects ranged from five to fifteenyears , inclusive . The number tested at each age isshown in Table I . The second column shows thenumber of ch i ldren in the better-class schools ; the thirdcolumn the number in the middle-class school

,and

the fourth column in the poorest school . The nextcolumn gives a few additional cases that were not usedin the determination of the method of scoring

,because

they were collected after this part of the work wascompleted ; they are included, however, in the di stributi on of scores for each age and thus enter into thedetermination of the medi ans and percentiles . Thelast column Shows the total number of indi viduals ineach age-group . It will be noted that there are onlyeighteen cases between the ages of Sixteen and eighteen .

Since the number of these cases was so small, and sincea great many of them were extremely retarded pupilsin the grades

,it was decided that they were not repre

sen tative of any of these ages, and therefore theirrecords have been omi tted entirely in this work . Thisgives us a total of 1520 cases

‘

upon which the mediansand percentiles are based . The small number. ofcases at Age 5 makes the norms for this age somewhatless reliable . We may say here that the usual difficultyin securing 14 and 15-year-old chi l dren was en countered . It was felt that the subj ects at those agesfound in the grade schools would probably be considered retarded pupils . Consequently, permissionwas obtained to get an equal number at those agesfrom one of the junior h i gh schools in a good residential section of the city . These chi ldren, beingprobably slightly accelerated

,would tend to balance

any possible retardation of those in the grades . Anequal number of un selected children was tested atevery age from six up to and including th i rteen, wh i chincreases the reliability of comparisons between theseages.

METHOD OF PROCEDURE 2 1

The adults,as mentioned above, were mainly uni

versity students who had volunteered to act as ob

servers in a series of experiments,but the others were

unselected,presumably normal persons who coul d be

prevailed upon by the examiner to take part in theexperiment with a fair amount of seriousness . AsHea ly has pointed out

,th i s is not a test adapted to

the adul t mind . Not only is there evident thattendency toward lack of seriousness among adultswhen taking tests

,and especially this kind of per

form an ce test,but also the simple natur e of the picture

calls forth from mature individuals all sorts of oritici sm s wh i ch never seem to trouble the chi ld . Thenature of the drawing

,with its lack of perspective ,

and the somewhat exaggerated,though simple

,situa

tions present difficulties and possibilities to the adultwhich sometimes make his performance little superiorto the performance of the child .

The conditions prevailing while this test was beinggiven were practical ly ideal . The test was given tothe children individually in a room where the examineronly was present . If a thir d person chanced to comein and the performance of the child seemed to beinfluenced in any way

,that record was cast out .

Each child was made to feel at ease,and without ex

ception ,unless it was among the five—year-olds

,the

problem was attacked with interest and without selfconsciousness . There seemed to be great demand inevery grade to be allowed to be the next one to dothe puzzle .

”

With such a large number of cases an d such a shorttask , it would have been impractical to have limitedthe giving of the test to any particular hour of theday . The experiment wa s, therefore, performed during the regular morning and afternoon school session s .

CHAPTER IV

THE RESULTS

As h as been described,a complete record was taken

of every move made by the subj ects in completing thepicture . A portion of the record sheet has been shownin the previous chapter. Af ter working with thetest , one becomes impressed with the inadequacy ofthe method of scoring used by Healy and other previoususers of the test . Healy says : “ The errors are oh

vi ously of two kinds, rational and irrational . Tha tmay be perfectly true

,but after considerable exper i

mentation it was found that individuals differed decidedly as to what should be called logical and whatillogical . It was decided

,therefore

,to abandon any

attempt to determine a pr i ori what moves are logicaland what illogical and to base the method of scoringupon the frequency with which the moves were actuallymade . To this end

,the first tabulation of the resul ts

was made on large sheets that showed the number oftimes any one position occurred . These numberswere then converted into percentages and transferredto other sheets to Show the percentages for each possible position at ea ch age . A record of these percentages is given in Tables II to XLII .

1

1 In th ese tabl es an d i n th e text th e fol l owin g abbreviation s for th en am es of th e bl ocks an d Spa ces wil l be used . Broken Win dow=B .

Win d ; Curtain ed Win dow=C . Win d ; Footba l l =F . Ba l l ; Baseba l l =B .

Ba l l ; Fly in Bi rd=F . Bird ; Depart in g Cat=D . Ca t ; S l eepin g Ca t= S .

Cat ; Mil k ott1e=M . Bottl e ; Spr inkgl in g Can =S . Can .

2 2

23THE RESULTS

TABLE II .

52 0 60 0

TABLE III .

0000oo

170

070

7

m

0

6

0

7

0

7

0

2

0

4

0

5

70

73

3

73

7

0

0

2

1

3

0

3

0

0

0

6

0

0

770

2

m4

4

2

0

0

4

0

0

0

TABLE IV .

0


57 3 53 4 52 0 64 0 59 4

TABLE VI .

0 72 6 2 0

1 3 2 0 0 72 0 2 7 2 0

0 18 7 38 0 46 6 0 68 0 72 6 82 6 880 7

1 3 1 3

B . WindDog 1

Cat 4 0 7 0 7Footba l l . 5

Hat 2

Basket . 30 0 37 0 77 3 78 0 88 1 0 86 0 85 8

Log 5 0

Ch icken 5 0 5F . Bird . 15 0 7 3 1 3

25THE RESULTS

TABLE VIII .

2

3

3

6

0

0

0 96 0 95 6 76 88 0 92 0 93

4 0 1 1

4 6 82

70

3

6

770

3

0

4

1

4

0

0043

1

70

3

3

7

7

73

7

0

2

5

1

2

2

0

33

0

3

3

7

3

0

0

3

6

3

0

3

2

4

5

2

0

0

0

0

5

55

54.

1

1

6

1

1

37

0

0

0

0005

0

0

0

0

3

1

4

®

73

73

70

0

1

0

1

2

m

03

77

0

0

1

0

0

2

25

0

0

3

3

4

0

4.

2

5

4

4.

7

0

3

7

77

0

2

1

0

0003

3

0

3

773

3

1

2

1

2

2

1

92

76

3

0

3

0

3

7

0

4

3

2

1

2

100

0

0

5

0

15 0 1

THE PICTURE COMPLETION TEST2 6

Footba l l

TABLE XII .

50 49 3 41 4 42 654 0 58 7

1 3 0 7

7070 .0 .0.2

73

70

0

1

0

2F . Bird .

TABLE XIII .

4

7

0 7 0 7

5

0

73 0 7

2 7 2 0 3 3 0 7 1 3 2 6

0 7

4

0

O

1

,0 7 2 0

1 0

6 0 2

0 7 1 3 0 7 3

3 0 7 2

COMPLETION TEST

TABLE XVII .Ha tchet

TABLE XVIII .

4 0 2 6

TABLE XIX.

THE PICTURE

5 0 2 0 2 7 4 6 2 7

28

F . Bird . I

3

73

77

3

7

1

0

0

3

70

7

7

0

3

4

5

0

000

70

73

3

7

2

2

5

0

2

0

1

3

M

6

6

0

0

0

0

73

0

A

Zalu

m/u

n

iu

n

/u

n/u

opsu

mu

0

0

0

THE RESULTS

TABLE XX.

Footbal l

TABLE XXII .S prinkl i ng Can

B . Wind0 7 0 7 O 7

0 0 7

1

1 2 0 2 6 0 71 1 3

1 0 72 2 0

THE PICTURE COMPLETION TEST30

TABLE XXIII .Glove

7070707

7

0

0

7

77

0

0

0

77

7

0

0

0

73

7

7

77

0

1

0

0

0

0

777

7

7

0

0

0

0

0

TABLE XXIV.

Bottle

7 0 7 1 0

TABLE XXV .

1 3 2 6 5 3 7 3 5 0

3

3

6

7

1

1

4

0

70

73

4

3

0

77

7

0

1

9

1

2

0

0

0

31THE RESULTS

TABLE XXVI .Cup

B . Wind . 5

1 371 30

5 0

TABLE XXVII .S h oe

05

M

n

m”

m.

,

m

a.

m

B

h

B

MCF

TABLE XXVIII .Flowers

0170

7077

3

0

0

1

73

7

7

0

1

0

0

1

0

0

0

0

73

3

7

0

1

1

0

7

77

0

0

2

0

2

76

0

0

0

0

2

2

4

2

0S


TABLE XXIX.

Pumpkin

. 5 0

Footba l l

TABLE XXXI .

Fish

2 6 0 70 7 0 7 0 7 0 7

Footbal l 1 3

Hat 5 0 4 0 2 0

Basket 5 0 0 7

Log 2 0 1 3

Chicken 4 0 0 7 2 0 0 70 7 0 7 0 7

THE RESULTS

TABLE XXXII .

0 7 0

TABLE XXXIII .

TABLE XXXIV .

Ch erries

0 1 3 0 7 1 3

0 4 0 2 7 4 0 3 3

14 7 10 7 9 3 6 0

1 3

0 4 0

1 0 2 0 1 1

4 0 3

33


TABLE XXXV.

S tool

F . Bird :

TABLE XXXVI .

0 77 4 0 5 3 2 0 2 0 2 0

F . Bird

TABLE XXXVII .

Books

B . Wind . 5 0 0 0 7Dog 10 0 2 0 7 1 3 1 3 2 0 0 7 1 3

1 1 3 0 7 0 7 0 7 0 7Footba l l . 1 3 3 1 3 2 7 0 7 0 7 1 3 1 0

5 0 2 2 0 3 3 0 7 0 7 0 7 2 6 0 7Basket . . 2 0 7 0 7 0 7 0 7 0 7

. 10 0 1


TABLE XLI .

0 7 0

0 7 0 7 0 7

TABLE XLII .

N0 Block

The tables sh ow the percentages for each one of the41 blocks for each age-group . The horizontal rowsgive the name of the spaces and the vertical columnsthe percentages at each different age . Table II forthe B . Wind . blo ck shows every position in which thisblock wa s placed and the percentage of cases in wh i chit was placed in such position at each age . The tableis to be read as follows : the B . Wind . block was placedin the B .Wind . space (i . c.

,the correct position) by

15 per cent . of th efive-year-olds, by per cent of

THE RESULTS 37

the six-year-olds, by per cent of the seven-yearolds

,and so on . Continuing to Line Two, we note

that the B . Wind . block was placed in the space forthe dog by per cent of the six-year-olds, byper cent of the seven-year—olds

,by per cent of the

eight-year-olds,and so on . The last vertical column

headed “ Total gives the results for the combinationof all the age-groups and includes 1479 cases . Thepercentages do not total 100 because only 9 of the 41blocks can be used . Some of the blocks , e . g .

,Scis

sors (Table XL) were used very seldom . From these41 tables every possible type of move and the numberof times such a move occurred can be seen .

Tables II to X inclusive dea l with the nine correctmoves for the spaces . We note a fair ly steady increase in the percentages from age to age . Th i smeans that as we proceed to the h i gher ages an i ncreasin g ly greater number of subj ects are puttin g thecorrect block in its correct space . There is obviouslya better and better performance at each age . Furthermore

,we n ote that the ni ne blocks in question are

never inserted in any other space a greater number oftimes than they are inserted in the correct space .

These ni ne blocks,therefore

,have been shown by

observers at all ages to be better adapted to the spacefor wh i ch they were originally designed than for anyother space . In this particular

,at least

,the actual

performance of chi ldren at all ages corroborates theoriginal design of the test . There also seems to be agreater amount of scattering in the lower than in theh i gher ages .Comi ng now to Tables XI

,XII

,XIII

,XV

,XVI

,

XVII , XVIII , XIX, and XXV, we n ote a series of

blocks that seem to be preferred for a specific spaceby a fairly large percentage at each age . For example ,


in Table XII the C . Wind . is put in the B . Wind .

space by a fairly large percentage of Observers at allages . The same holds true of the S . Bird and Cherriesin the F . Bird space, of the D . Cat in the Cat space

,of

the Baseball in the Football space,of the Baby in the Cat

or Hat space, of the Hatchet in the Log space, of theCage in the F . Bird space

,of the M . Bottle in the Cat

space,of the Purse in the Hat space

,of the Cherries

in the Basket space, of the Cat in the Ch i cken spaceand the Ch i cken in the Cat space . In some of thesecases the percentage is very h i gh

,notably the C . Wind

in the B . Wind space, indeed at some ages a largerpercentage of subj ects choose this block rather thanthe correct on e. The percentage of the total numberof subj ects is or a little below half . None of theother cases mentioned reaches such a h i gh percentage .

The reason for thi s group of cases is obvious froman inspection of the test . The C . Wind is put in theB . Wind . space

,because the house is obviously lack

ing a window and the subj ect has failed to note theactivity of the boy wh o h as thr own a stone or thebroken pieces of glass or the en raged man, or if hehas noticed these th i ngs

,he has failed to combine

them into a story . The D . Cat is put in the Cat spacebecause the child is obviously Offering mi l k to a catbut the subj ect has not noticed that there is a moreappropriate cat than the on e that is walking away .

The substitution of the Baseball for the Football is againObvious and n ot entirely illogical . The Baby in theCat space is explicable by the common association Ofbaby and milk

,and in the Hat space by the idea that

the girl is very much concerned about the crying baby .

The Hatchet in the Log space is due to the associationbetween hatchets and logs . The Cage in the F .

Bird space seems merely due to the association be

THE RESULTS 39

tween on e cage and another. The girl may be transferring the bird from one cage to the other. TheM . Bottle for the Cat seems again to rest upon a mereassociation between m ilk and milk bottle . The S .

bird in the F . Bird space is due to overlooki ng the factthat there is a more logical block (the F . Bird) forthat space

,and the Cherr ies in thi s same space is sup

posed to mean that the frui t is dropping from thetree

,while the subj ect fails to notice the activity of

the girl with the cage . The Purse in the Hat spaceis interpreted as meani ng that the wind has blownthe girl ’s pur se away. The Cherries in the Basketspace means that the subj ect fails to see that the treeis an apple tree . The Cat in the Ch i cken space istaken for a ch i cken frightened at a cat . The Ch i cken inthe Cat space is interpreted as the gir l feedi ng a chicken .

These cases,therefore

,are all easily understandable

,

and occur because the subj ect has failed to note thatthere is another block that would make a still bettercompletion of the picture than the one he has chosen .

If we were to make a classification of errors into logicaland illogical ones

,as Healy has suggested

,it would

seem to us that these cases are the ones to merit theformer title . Healy gives nine moves as logical errors .We have discussed above the thirteen most frequenterrors of our subj ects and according to our point ofView, these ought to be called the logical errors, if aclassification into logical and illogical errors i s to bemade . Six of our

“ logical ” errors correspond to

Healy’s , namely the C . Wind . for the B . Wind .,the

Baseball for the Football , the S . Bird for the F . Bird,

the D . Cat for the Cat , the Purse for the Hat, and theHatchet for the Log . We have eight logical ” errorsthat are not mentioned by Healy . They are the Baby


in the Cat and Hat spaces,the Cage in the F . Bird

space , the M . Bottle and Chicken in the Cat space,

the Cherries in the Basket and F . Bird spaces,and the

Cat in the Chicken space . Two of Healy’s logicalerrors do n ot occur in our list of the 61 most frequentmoves to be discussed later

,and these are the Baby in

the Basket space— a move made by but one per cent .of the total

,73. c .

,by only 15 of 1479 people ; the Mouse

in the Cat space, which occurs only in per centof the total

,i . c .

,8 cases . These two so-called logical

moves eviden tly did not seem logical to our subj ects .

These do not appear at all in our list of common moves,

since the number wh o make these moves is in everycase less than 22

,or per cent of all the subj ects

tested . They therefore do not warrant any score,a o

cording to our method (see Chapter V,The Determ in

ation of the Scores) .The rest of the blocks are inserted in the spaces soseldom as not to warrant special consideration . Inlooking thr ough Tables II to XLII it will be notedthat most of these scattered cases tend to occur inthe lower ages .Changing slightly our point of view

,we may n ow

examine Figures 2 to 10 . These graphs show thepercentage of responses at each age for the nine spaces,and the curves denote the blocks that were most comm on l y inserted into these spaces . The solid line ineach case represents the block that was originallydesigned for the space . For example, Figure 7 showsthe three most common responses to the Basket space .

The solid line shows that in th i s space the correct block,73. c .

,the Basket

,was put in the Basket space by 30 per

cent of the five-year-Olds, by 37 per cent of the six

year-olds,and so on . Further that the Cherries block

RESULTS

FIGURE 2

41


FIGURE 3


FIGURE 5

THE RESULTS

FIGURE 6

45


FIGURE 7

THE RESULTS

FIGURE 8

47

48 THE ‘ PICTURE COMPLETION TEST

FIGURE 9

THE RESULTS

FIGURE 10

49


(broken line) was put in the Basket space by 10 percent of the five-year-Olds

,by 15 per cent of the six

year-olds,by 12 per cent of the seven-year-Olds

,and

so on . And further,that the th i rd most common

response that was made to th i s space was the insertionof the Bucket block (dot and dash line) , wh i ch occurredin per cent of the six-year-olds , 5 per cent of theseven-year-olds , per cent of the eight-year-Olds

,

and so on .

Sim i larly Figure 2 shows the three blocks mostcommonl y inserted in the B . Wind . space

,namely the

B . Wind .

, the C . Wind .

, and a blank block . Notethat blank refers to a block wh i ch h as n o picture onit . There are 10 blank blocks as can be seen fromthe pictur e on Figure 1 .

All the other graphs are to be read in the same way.

In general , the correct block for the space is distinctlyabove all the other curves . Indeed, in most casesthe percentages for the other responses are so low asto be relatively insignificant (see Figures 3 , 6 , and 9 inparticul ar) . Three graphs differ decidedly from thi sgeneral tendency

,in that the curve for the correct

block falls below some one of the other curves at someages . This is the case in Figures 2 , 5 , and 10 . It ismost marked in Figure 2

,where the curve for the C .

Wind . remains above the B . Wind . curve at Ages 5to 10 , inclusive . Th i s means that at these ages it ismore common for the observer to in sert the C . Wind .

rather than the B . Wind . It marks th i s space out asbein g definitely the most di fficult, but it does n ot

prove that the C . Wind . is really the more logicalblock . Th i s question was raised after the test hadbeen completed by asking a great number of the subj ects wh o made this error to state which block was the

CHAP TER V

THE DETERMINA TION OF THE SCORES

P ercen tage Di str i buti on

The total group of subj ects tested was made thebasis for the determination of the scores . TableXLIII shows the percentage distribution of the totalnumber of 1479 cases used for the determination ofthe scores . The vertical column gives the name of

the block and the horizontal row the name of thespace . There are nine spaces in the horizontal rowsand 41 blocks along the vertical columns . The tableis to be read as follows : the B . Wind . space (firsthorizontal row) was fil led in by the B . Wind . blockby per cent of the subj ects (upper figure) and thi smove was given a score of 100 ; the B . Wind . space

(continuing along the row) was fil led in by the Dogblock by per cent of the subj ects and received ascore of and so on . Passing to the next row wefind that the Dog space was filled in by the Dog blockby per cent of the total number of individualsand was given a score of next that the Cat wasput in the Dog space by per cent . of the subj ectsand was given a score of an d so forth .

In determini ng these scores , an examination of thenine correct moves was made fir st . This shows thatthe B . Wind . in the B . Wind . was the hardest, becausethe fewest subj ects were able to do it ; it was given ascore of 100 . The other eight correct moves weregiven scores in inverse proportion to the percentage Ofcases recorded . If per cent are able to put theB . Wind . in the B . Wind . space and receive a score of100

,then per cent wh o put the Dog block in the

52

53DETERMINA TION OF SCORES

TABLE XLIII .

Di stri bution of Tota l Number Tested

PERCENTA GES AND SCORES

B . Wind .

0 07 0 07 0 07

0 3 0 07 0 14 0 14 0 14 0 070 2100

P er Cen t . .

Score

9

6

75

0

0

0

0

3

2

0

0

6

4.

0

0

3

2

0

0

4

6

2

1

77

261

1

75

0

0

79

73

400

75

0

0

9

6

0

0

2

2

0

0

5

3

0

0

00

6

0

0

64.

0

0

2

2

0

0

2

2

0

0

3

2

0

0

4

4

0

0

2

2

0

0

5

5

0

0

COMPLETION TESTPICTURE54

TABLE XLIII— (Continued)

Pa

m.

m

0

95

2

Sa

ga

n

hna

m

Ha

ga

n

06

0.

m

B . Wind .

2

2

3

2

5

0

1

1

70 .2

1

0

0

Dog2

2

3

2

2

7

72

1

1

2

1

8

2

1

1

4

3

3

2

5

3

0

0

5

3

0

0

1

1

4

3

0

0

P er Cent . .

Cat4

9

9

6

4

9

2

7

1

0

270

1

0

0

4

2

1

5

3

0

0

Per Cent . .

Footba l l0

0

1

7

1

0

O

4.

0

0

3

2

0

0

O4.

0

0

00

5

0

0

6

0

0

1

3

2

3

2

0

0

3

2

0

0

0 07

Hat

P er Cent . .

1

0

0

71

0

75

0

0

2

2

3

2

9

6

0

0

5

1

5

3

0

0

3

2

0

0

Basket

P er Cen t . .

5

3

0

0

2

1

0

0

A

4

0

0

P er Cen t . .

Score

5

0

1

1

3

2

0

0

3

2

0

0

3

2

0

0

00

5

5

3

0

0

3

2

0

0

4

3

0

0

9

6

200

0

0

1

0

72

72

1

1

0

7

1

2

5

9

4

2

1

600

~

2

1

5

0

0

0

9

3

1

1

9

3

3

2

1

1

0

0

7

2

1

0

0

0

0

9

6

3

2

0

0

0

0

72

2

1

1

1

0

0

71

0

00

5

0

0

0

0

mH

mn

e

u

e

_

Cm

_

Cm

er

O

e

c

d

e

c

k

P

SHP

S

m

B

C

F

£35DETERMINA TION OF SCORES

TABLE XLIII— (Con tin ued)

”firm

“

axaqnxm

20

30

5

oonm

96

230mm

.

2

230

m

e

bo

ny

.

m

700 .9

6

0

0

2

1

0

0

7

5

0

0

Footba l lP er Cent . .

ScoreHa t

P er Cen t . .

ScoreBasket

2

2

0

0

75

0

0

P er Cent. .

Score

3

2

0

0

3

2

0

0

0

0

3

2

0

0

P er Cent .Score

Ch i cken42

0

0

2

1

0

0

3

2

0

0

P er Cent . .

ScoreF . Bird

P er Cent . .

Score

THE PICTURE COMPLETION TEST56

Moo

fio

Z

o

9m

Eo

fio

m

o

ak

goo

n

TABLE XLIII— (Con tinued )

3

2

0 07 00

P er Cen t . .

Score .

P er Cen t . .

P er Cen t . .

Score

ScoreBasket

P er Cen t . .

ScoreHa t

P er Cen t . .

Score

Score

Footba l l

Cat

0

0

3

2

0

0

0

0

0 1

7 0 07

DETERMINATION OF SCORES 57

Dog space sh ou ld receive a pro rata score of 64100 : and so on for the nine correct blocks .

The easiest of these nine moves is the Log in the Logspace

,since the largest proportion of successes,

per cent,is with th i s move .

Now in regard to the other moves made, which wemay call errors

,it is obvious that the most pardonable

error is the one made by the greatest number of people .

We might term such an error the “ most logical ”

error,meaning by logical ” that it occurs most

frequently . Therefore all the other moves meritscores in proportion to the number of people makingthe move . The greater the frequency of the

‘

move,the higher the score . We cannot

,however

,allow any

of these so-called errors to exceed the score of any ofthe correct moves

,hence all of them must fall below

a score of 52,the score for the easiest correct move .

The relative merit of these errors can be determined ,then

,by using 52 as a starting point . The greater

the frequency of the move the larger the score . Ifper cent (Log in Log) is equal to a score of 52 , thenby direct proportion

,47 .1 per cent (C . Wind . in B .

Wind .) is equal to a score of 32 . These scores areall given in Table XLIII . The second figure

,the figure

below the percentage,denotes the score in each case .

For practical purposes the scores recorded in thetable were needl essly fine

,so the decimals were dropped

and the nearest integer was taken . Neglecting a ll

scores below we arrive at the scores given in TableXLIV .

Sixty-one moves have been found to merit somescore by th i s procedure . To make the lowest score 1is , of course, purely an arbitrary procedure . It wouldbe perfectly feasible to go beyond th i s and include


TABLE XLIV.

R. WINDB. WindC . Win dBl ankCag e

Mouse

III CHICKEN

FOOTBALL

F . Bird

5? F . BIRD . ffifl

’

fffifi.

6

2 Basket

BASKET Ch ermesBasket


frequency of the position of any block in any of thesquares was noted . Upon these frequencies the scoreswere allotted . That is to say

,the method of scoring

is based entirely upon what is actually done by thesubj ects . The performance of the subj ect is a measureof the ease or difficulty of any position

,and this in turn

determines the value allotted to that particular move .

Now,in View of the fact that the number of tested

individuals is not constant at each age,the Obj ection

may be raised that any age possessing a proportiona tely larger number of subj ects might influence undulythe scoring values . This is the first point that calls for acorrelation between the relative ease or difficulty of eachm ove at each age and the same move for the total group .

Another and more important question also demandsth i s kind of a correlation . If the type of performance

,

73. c .,the relative ease or di fficulty of the various moves ,

is radically and characteristically different at any onea g e or group of ages from the type of performancecomm on to the other ages , then the method of scoringa s determined by the total of all tested may be unfai rto that age or group of ages

,since the method of seor

i n g is supposed to be determined by what the subj ectsa ctually do . If

,for example

,the relative ease and

d ifficulty of the various moves for Ages 6 and 7 wereto vary from the relative ease and difficulty for theo ther ages, then the scoring values as determined byt h e total tested would be unfair to Ages 6 and 7,m eaning by unfair that they would n ot be based uponwhat six and seven-year-olds actually find to be easyor difficult, but upon what a group of in dividuals, ofwhich the six and seven-year-olds form a small number

,find to be easy or difficult . In other words , if

any age or age—group were to deviate in th i s respect

DETERM INATION OF SCORES 61

from the total group,that age-group would be scored

by values obtained from other sources . Such a methodof scoring would be perfectly legitimate and defensible .

In fact,the best possible move for the nine correct

squares is determined by such a method . And inregard to the other moves

,excluding the nine best

possible ones,it would have been entirely feasible to

have decided upon a method of scoring based upon theperformances of a select group of adults . This methodwould imply that such a group Of adults makes movesthat are the most logical or the least illogical . Thismethod was not adopted Since it was thought thatthe ch i ld ’s performance might differ so radically fromthe adult ’s as to make a method of scoring derivedfrom adults ’ performances artificial when applied tochildren . And further

,a practical consideration enters

in, namely that a select group of adults would make SOfew illogical moves as to limit the range of scores verymaterially . As a matter of fact

,the total number of

moves , other than the nine correct ones , chosen by theadults amounts to 54 . Since there are 369 possiblemoves in addition to the nine correct ones

, 54 is arelatively small number .To arrive at these correlations

,the rank according

to difficulty of the 61 moves whi ch have been alloweda scoring value was taken . These 61 moves are themost frequent , as determ ined by the total number ofindividuals tested . It was not deemed feasible towork out correlations for all the possible 369 moves ;most of them are seldom met with . These 61 mostcommon moves will give us a sufficient indication ofthe correlation between the total group and the different age-groups . It is to be noted

,however

,that this

method of ranking excludes in certain age-groups


some moves which do n ot appear in the total groupand which would have ranked higher than a few of

the less frequently chosen moves . To thi s slightextent

,therefore,

'

our table of ranks is incomplete .Table XLV Shows the ranking in order of frequencyfor the 61 moves, for the total group tested, and thenfor each g roup separately. From this table Figures1 1

,12

,and 13 have been constructed . These three

graphs Show the ranking at each age of the twelvemost frequent moves as determined by the totalgroup . Some idea of the Similarity in .ranking for thedifferent age groups can be Obtained from them .

Figure 11 shows that the Log in the Log space was themove that received first rank at all ages except Six andseven . The other three moves

,namely the Dog

Dog , the Football— Football, and the B . Wind .

-B.

Wind . do not display such a constant rank , but therange of variability is small when we consider thatthere are 61 possible ranks . The four moves givenin Figure 12 Show very little fluctuation in rank ; theChi cken— Ch i cken, the F . Bird— F . Bird, the CatCat

,and the S . Bird— F . Bird moves vary onl y from

on e to thr ee places . In Figure 13 the same relativelyconstant tendency is observed in the ranks for theBasket— Basket and Hat— Hat moves

,but when we

come to the two “ logical ” errors,the C . Wind . in the

B . Wind . space and the Baseball in the Football space,there appears a somewhat greater range of ranks,although this is comparatively slight i f we considerthe total length of the series . The change in rank ofthese two moves is somewhat abrupt as we proceedfrom the lower to the higher ages .The correlations determined from the table of ranks

(Table XLV) are, fir st, between each age-group and

P osi ti on

Log—Log

Basket—Basket .

Ch icken-Ch i cken .

Hat—Ha t

Footba l l—Footba l lC . Wind -B . WindF . Bi rd—F . BirdB . Wind —B. Win d .

Basebal l—Footbal l .S . Bird—F . Bird .

Ch erri es—Basket .Ha tch et—LogBaby-CatM . Bottl e—Cat .

Baby—HatCh erries—F . B i rd .

Ch i cken—Ca tCat-Ch i cken

B l ank—DogMouse—Hat

D . Cat—DogBasket—F . B ird .

Ch erries—Footbal l

S . Cat—Hat

S tool—LogCh erri es—Ch i ckenCat—DogBucket—Basket .

S . Bird—Ch i cken .

F . B ird—Ca t .

MouseL Dog .

S . B ird—Ca tBaby—Ch icken .

Ha tch et-Ch icken

DETERMINATION OF SCORES

TABLE XLV.

A g e Groups

1 1 1 1 1 1 1

2 2 3 4 2 3 4

3 3 2 3 3 4 2

5 4 4 2 4 5 5 2

4 5 5 5 4 5 2 5

8 7 6 6 8 7 76 6 8 8 5 6 8 6

7 8 10 10 10 10 10

9 9 7 8 5 7 6 9

10 10 9 7 9 9

12 12 12 11 11 5 11 12

11 11 11 12 11 5 12 11

14 16 28 20 13 5 26 5 22

14 13 13 13 5 13 5 14 13

(A

C/1

01

q (I'

14 14 15 5 13 5 20 13 14 5

16 5 17 18 23 18 5 2 1 49

18 5 15 14 15 15 5 2 1 16 522 5 47 5 28 17 15 5 16 5 49

18 5 18 18 20 48 5 2 1 32

20 5 28 36 57 36 5 16 5 22

5 25 5 24 18 32 28 5 32 49

38 40 5 2 1 5 40 17 2 1 22

31 20 28 32 18 5 32 32

20 5 33 20 32 48 5 43 5 49

28 55 5 36 57 28 5 43 5 49

5 16 5 55 5 23 5 26 5 48 5 43 5 22

51 5 24 28 17 28 5 56 5 16 5

22 5 47 5 2 1 5 48 5 24 15 49

45 5 20 53 5 40 36 5 43 5 22

38 24 23 5 17 28 5 2 1 22

45 5 40 5 28 20 24 32 49

5 51 5 24 15 5 23 2 1 5 32 14 5

31 33 53 5 32 36 5 56 5 49

25 5 28 36 40 36 5 43 5 49

5 38 47 5 44 5 40 2 1 5 32 32

5 38 40 5 36 48 5 28 5 43 5 3231 33 53 5 26 5 36 5 56 5 49

5 25 5 33 36 57 36 5 26 5 22

5 38 47 5 36 40 36 5 43 5 32

5 38 55 5 44 5 48 5 28 5 43 5 49

5 31 20 60 23 48 5 2 1 49

5 51 5 28 60 57 36 5 43 5 49

56 5 24 28 48 5 48 5 43 5 32

5 45 5 55 5 53 5 32 58 5 56 5 49


TABLE XLV (Contin ued)

A g e Group s

P osi ti on

Stool-Ca t

Stoo l—DogCh i cken—Hat

Cup-Cat

Ha tch et—DogCag e—Ch ickenDog

—HatD . Cat—Ch i cken .

S . B ird—DogF . Bird—Ha t

Cag e—B . Win d . .

F . B ird—Footba l l .

F . B ird—Ch i ckenBl ank—LogMouse—Ch i cken .

B . Wind —DogBooks—Ha t

the total group . They answer the question as to therelative difficulty of the moves for each age-group ascompared with the total group . These correlations

,

as computed by the Spearman Foot-Rule method , aregiven herewith :

Averag e

DETERMINATION OF SCORES

FIGURE 1 1

65


FIGURE 12


The P . E . varies from .03 to .04 . These correlationsare high and range only from .74 to .90 . These highcoefficients and their uniformity at every age justifythe use for subj ects of any age of the scoring valuesdetermined by the total group . They show that

,on

the whole,the relative ease or difficulty Of the moves

i s pretty much the same for any age-group .

In addition to these correlations,we have computed

by the same method others between each age andevery other age for the ranking of these 61 moves .These correlations are Shown in Table XLVI . It will

TABLE XLVI

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

be seen that for the 61 moves under considerationthere is at all ages a high correlation . The coefficientsfluctuate between .55 and .82 . This means th a t

_

for

the 61 moves under consideration the relative difficultyat all ages i s much the same . In other words no ageor group of ages is strikingly different from the others ,and therefore we may conclude that n o injustice isbeing done to any particular age by our method of

scoring . Before computing these coefficients it wasthought that there might be an increasing differencebetween the ages as we proceed from the lower to the

DETERMINATION OF SCORES 69

hi gher ages . For example,Age 6 might be thought to

correlate most highly with Age 7 , less with Age 8 , andso on up to the adults , with which it might be expectedto correlate least . A n inspection of the table showsthat this tendency is present

,but only to a very Slight

degree . We cannot find any very uni form decrease .

Looking at the first line of the table,we note that Age

6 correlates most highly with Age 7,then with ages

13,9,adult

,8,10

,1 1

,15

,12

,and 14

,in the order

named . An inspection of the other ages,taking care

to begin with the ages closest to the age under consideration

,shows much the same results . On the

whole,then

,the correlations seem to Show that n o

particular injustice is done to any age by the methodof scoring adopted .

CHAPTER VI

NORMS

By the foregoing method the 1520 cases werescored, and a table of distribution for all the cases byage was made . Th i s table is too extended to givehere ; only a summary can be shown . Table XLVIIShows the actual number of cases di stributed by ageand by score in units of fifty points . A perfect perform an ce re

'

ceives a score of 646 . Table XLVIIIShows the same facts as Table XLVII

,except that

percentages are used in place of numbers,thus making

possible a'

di rect comparison frOm age to age absolute .

Without going into detail,we see that the larger per

centages in the table tend to run diagonally from theupper left-hand corner down to the lower right-handcorner . At n o age-group does the largest percentageoccur in the 600 to 646 score . The largest Single percentage of adul ts and fifteen -year—olds lies in scoresbetween 551 and 600 . It is interesting to note thateven adults do not find th i s test easy to completeperfectly . A percentage of adults falls below ascore of 2 51 . The greatest number at any age-groupreceiving -

a very hi gh score,above 600

,is found in the

fifteen -year-old gr oup . From this group down,there

is,for th i s very high score

,a steady decrease in per

centage to zero at Age 6 . Similarly, with the verylow scores we fin d the largest percentages at the lowerages and a more or less steady dim inution in percentage as we proceed to the higher age-groups .

P ercen ti les

From the extended table of distribution not includedhere

,percentile scores for every ten per cent were cal

NORMS 71

TABLE XLVII .

Di stri buti on of S coresNUMBERS

Score 5 6 9 1 0 1 1 1 2 1 3 1 4 1 5 A d .

1 1

2 0 150 150 152 150 150 155 150 150 109 52 92

TABLE XLVIII .

Di stri bution of S coresP ERCENT A GEs

T otal


cu l ated for each age-group . These are shown inTable KLIX.

l The fifty percentile or median givesthe age-norm for the test . The median scores forchildren Show a steady increase from age to age

,and

would seem to denote sufficient difference from age toage to make this an excellent test for diagnostic purposes . The drop of ten points in the medi an of theadults as compared with the fifteen -year-olds raisesthe question whether the test i s adapted to adul t

TABLE XLIX.

Per cen ti le Di stri buti on

Hig h est 394 525 646 646 646 646 646

259 354 463 507 577 578 583207 260 364 446 499 522 578169 2 19 318 422 5 450 509 519

126 185 263 381 440 478 50 189 153 239 328 407 435 45549 99 189 296 372 411 44328 76 153 261 325 367 411

5 62 107 20 1 5 297 313 3692 14 60 129 240 251 296

0 0 O 3 62 35 122

20 150 150 152 150 150 155

207 245 348 439 477 515 534

89 153 239 328 407 435 455

5 71 5 131 235 310 339 380

10 1 87 108 102 83 88 77

intelligence . A S noted in another chapter, Healyreports a poor performance with adults . In explanation : it may be that a far greater number of possibilities occur to them than occur to children, or it may be,that they do n ot try as hard as the children to do their

1 T h e quarti l e on th is an d oth er tabl es 18 th e sem i-interquarti l e rang e .

We h ave fol l owed Th orn dike i n cal l ing i t th e quartil e for purposes ofbrevity.

NORMS 73

best,although this was not obvious from their behavior.

Or it may simply be that our group of fifteen -year-oldsdid actually possess slightly more intelligence than ourgroup of adults

,if this test is a good test of intelligence .

The other percentiles Show an almost steady increase from age to age

,although thi s increase is not as

uniform as the increase of the medians . The curvesof the percentiles for each age are Shown in Figure 14 .

From this graph we note that the 30 percentile, the40 percentile, the 60 percentile, the 70 percentile, the80 percentile and the 100 percentile as well as the50 percentile (medi an) show no decided breaks fromage to age

,whereas the curves for the lowest score,

the 10 percentile, and the 2 0 percentile do Show verydecided breaks at certain ages . For example

,the 10

percentile of the th i rteen-year—olds drops slightlybelow that of the twelve-year-olds Similarly

,thefifteen

year-olds at th i s percentil e drop 37 points below thefourteen-year—olds . In the same way

,the 20 per

centile Shows a distinct drop at fifteen yea rs . It is tobe noted that this lack of a uniform increase in thepercentil es is characteristic of the upper ages . It ismost marked at Age 15 where there was the fewestnumber of cases . None of the breaks occur below theage of eleven . It is to be noted in reference to the100-percentile curve (i . e .

,the best score at each age)

that thi s reaches the maximal score at Age 7 and re

mains at th i s point for all successive age-groups .This means that at Age 7 and above there is alwayssome indi vidual who can complete the test perfectly .

The curve Showing the lowest score at each age isdecidedly irregular

,and n o individual in any age-group

above seven makes a zero score as in Ages 5,6,and 7 .

The curve attains its highest point in the adult group .


FIGURE 14


FIGURE 15

NORMS 77

Year-S ca le Norms

In case the test might prove suitable for inclusionin a year scale

,we present below the necessary values .

We have taken the arbitrary 75 per cent standard fora year scale— a standard which we believe the best indealing with a test of this type .

2 The lowest scoremade by the upper 75 per cent is taken as the yearscale value and the age interval for any one age extends from a point mi dway between this score and thescore of the age below to a point midway between thisscore and the score of the age above . These intervalshave been determined from the 25 percentile columnin Table XLIX . These limi ting scores

,or age inter

vals , are as follows :

5 6 7 8 9 1 0

5—38 39—10 1 102—183 184—2 72 273—324 325—359

A g e 1 1 1 2 13 14 1 5 A d .

Score 360—381 382—409 410—436 437—441 442-449 450 an d over

That is to say , all children wh o make scores of fromabout 5 to 38 are to be given five-year-old credit ; allchildren making scores from 39 to 10 1 are to be givenfive and Six-year-old credit ; those between 102 and183 receive five Six and seven-year-old credit ; andso on up to age fifteen

,where those who make 442 or

better receive a “ pass” at each year from five to fifteen,

in clusive .

The presentation of our results in the form of medians

,percentiles and year intervals makes possible the

incorporation of t h i s test into any type of scale or

group of tests in general use at the present time .

'For a di scussion of standards for yea r sca l es an d of th e way inwh l ch a test of th i s kind can be in corporated in to a sca l e, see P in tn er ,R . , an d Pa terson ,

D . G., A Sca l e of Perform an ce T ests, D . Appl etonan d Company, 1917.

CHAPTER VII

SEX, SOCIAL STATUS , AND SCHOOL STANDING

The differences in the scores of the ch i ldren according to their sex, according to the social status of theirparents

,and according to their school standing have

been briefly investigated . It is not the intention ofthe writers to enter into an exhaustive study of thisquestion

,but it seemed desirable to study the influence

of these three factors,particul arly of the last two .

There are an increasing number of studi es showingthe difference in general intelligence between childrenfrom better class homes and chil dren from poor homes . 1

If the same di fference were to be found in th i s on e test,

it was felt that it would give added reliabil i ty to thetest as a test of intelligence . In the same way

,some

idea of how the accelerated and the retarded pupilsperformed the test might increase our confidence init

,if the test showed a difference in the performances

of these two groups .

Sex Difieren ces

No minute analysis was made of the sex differencesin this test ; it was deemed sufficient to find the medians for the boys and the girls at each age .

The median scores are presented in Table L, andFigure 16 shows the same facts graphically. Fromthe table we see that the median for the boys is greaterthan the median for the girls for Ages 5 , 7 , 8 , 13, andfor adults , while the median for the girls is hi gher atthe remaini ng ages . From the graph it will be seen

1 Yerkes, R . M . ,Bridg es, J. W.

,and Hardwick, R . S . , A Point Sca l e

for Measurin g Men ta lgAbi l i ty . Warwi ck an d York, 1915 . Also ,

Bridg es, J . W . , an d Col er, L . The Rel ation of In tel l i g ence to Soci a lStatus. Psych . Rev i ew, —3 1 .

78

SEX,SOCIAL STATUS AND SCHOOL STANDING 79

TABLE L .

S ex Difi'

eren ces

MEDIAN SCORES

that the two curves cross and re—cross each otherseveral times , Showing no constant and uniform sex

difference . The slight variations of the medians ateach age are evidently not due to sex . We may,therefore

,conclude that the test is equally well adapted

to boys and to girls .

Difieren ces i n Soci a l S ta tus

The schools in wh i ch the tests were made weredivided into a good

,a medium and a poor school .

The good school,however

,was not very superior to

what we have called the medium school in regard tothe social status of the childrens ’ parents . The poorschool contained ch i ldren from the laboring classes

,

but could not by any means be considered the poorestschool in the city . There are relatively few ch i ldrenin the medium school

,so few in fact that at some ages

we omi tted the median as being too unreliable .

The median scores according to age for the threetypes of school are shown in Table LI . It will benoted that the medians for the good and medi umschools are very much alike . In fact at three agesthe median for the medium school is slightly abovethe median for the good school . There is

,then

,no

radical difference between these two schools on this


FIGURE 16

SEX ,SOCIAL STATUS AND SCHOOL STANDING 81

TABLE LI .

Soci a l S tatusMEDIAN SCORES

test . On the other hand,a comparison of the poor

with the medium or good school shows a decided difference in medians . At every age the medians of thepoor school are below the other medians . This i sshown strikingly in the curves on Figure 17 . At n oage does the curve for the poor school reach the curvefor the good school . The difference is uniform overthe whole curve and fairly well marked at all ages .

Raci a l Difl'

eren ces

It was thought that the di fference between coloredand whi te chi ldren m i ght be Shown by means ofmedians at each age . In our records

,however

,there

were only 36 colored ch i ldren and,therefore, age

medians would have had no value . The small number of colored ch i ldren does not allow of any importantconclusions . We have presented the facts

,however

,

for what they are worth . Table LII shows the number of colored children above or below the median for

TABLE LII .Raci a l Difieren ces

COLORED CHILDREN— NUMBER ABOVE AND BELOW MEDIANA g e 6 7 8 9 1 0 1 1 1 2 1 3 14 1 5 T ota l

. 0 0 1 3 0 1 1 0 1 0 7Bel ow . . . 2 8 1 4 4 3 4 1 1 1 29

THE PICTURE COMPLETION TEST

FIGURE 17


ch i ldren.1n grades above the corresponding ages have

been counted accelerated ; all childr en in grades belowthe corresponding ages have been called retarded .

The results are presented in Table LIII .

TABLE LIII .S ch ool S tandi n g

NUMBER OF A CCELERATED AND RETARDED ABOVE AND BELOW MEDIANA g e 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 T ota l

Accel eratedAbove 7 7 18 10 14 11 14

3 2 9 4 4 5 2

Retarded2 3 8 1 1 1 1 9 14 58

Bel ow 2 5 7 12 19 2 2 19 19 105

The table is to be read as follows : At Age 6 thereare 7 accelerated pupils wh o made scores on this testbetter than the median score for Six-year-olds ; thereare 3 accelerated pupils wh o made scores below thatmedian and there are n o retarded pupils . Under A g e10 we find 18 accelerated pupils

,of whom 14 made

scores above the median for ten-year-olds and 4 madescores below the median . Of the 10 retarded pupils atAge 10 , 3 made scores above and 7 below the median .

Examining the results for the accelerated pupils , wenote that at every age a decidedly greater numbermake scores above the median than below . The totalShows 81 making scores above and only 2 9 below themedian . That is

,out of a total of 1 10 accelerated

pupils 74 per cent make scores above the median .

Turning n ow to the retarded pupils,

.we note that atevery age a greater number score below than abovethe median for their age . The total Shows 105 belowand only 58 above the median . The total number ofretarded is 163, 64 per cent of whom score below themedian .

SEX,SOCIAL STATUS AND SCHOOL STANDING 85

In general,therefore

,we find that there is an agree

ment between performance in this test and schoolstanding

,and it is an agreement such as one might

anticipate in view of what we have said of the differen ces between the intelligence of the accelerated andthe retarded pupils in our schools . An absolute agreement between the results of the test and the schoolstanding was not to be expected

,and would have been

suspicious if it had occurred .

CHAPTER VIII

OTHER METHODS OF SCORING

Two other methods of scoring have been used forthis test in work already published . These we havementioned in Chapter II and we now propose to present our results worked out according to these twomethods

,both for purposes of comparison with the

work already done and for a critical inspection of themethods . These two methods are (1) Healy

’s Methodand (2 ) The Method of Right and Wrong Moves , usedby Hall .1

Hea ly’s Meth od

Healy’s norm for children above ten years is : notmore than on e logical and two total errors with a timelimit of five minutes . His list of logical errors hasbeen given in Chapter IV . In another place he says :At eleven years this test should be readily aecom

p l i sh ed with not more than one or two final errors ,and certainly not more than one illogical error . Themedian or average performance for all in the group ofthose ordinary in ability above ten years , is on e finalerror and no illogical error .” 2 The latter sentencewould seem to make his norm even a little higher th anbefore .

We have worked over our results in the light ofHealy’s norm

,considering as passed all who made not

more than one logical and two total errors (logical inthis connection means the particular moves so desi gmated by Healy) . A S the results were being compiled ,

1 Ha l l , G., El even Men tal T ests Stan dardi zed, Eugen i cs a nd Soci a l

Welfa re Bu l leti n ,NO . V, Sta te Board of Ch ari ti es, N. Y . , pp . 74—78 .

2

8

Hea l y, W.

,Th e Indi vi dua l Del i n quen t, Boston , 1915, Ch ap . VII,

p . 3, p . 1 1 1 .

OTHER METHODS OF SCORING 87

it became obvious that Healy’s norm was much toosevere . It occurred to us that the time limit of fivem i nutes mi ght be one of the reasons . Therefore, wepresent the results with and without the time limit.

of five minutes suggested by Healy .

TABLE LIV .

Number and P er cen t. P assi n g at Hea ly’s S ta nda rd

T im e Limi t N O T im e L im i tP er C en t P er C en t

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

Table LIV shows the number and per cent at eachage who pass the test according to Healy’s norm . T h e

left half of the table gives the results with the five

minute time limit,the right half with n o time limi t .

The table is to be read as follows : At Ages 5 and 6no children passed the test ; 7 chi ldren, or per cent

,

of the seven-year-olds passed if a time lim it is adheredto

,and 12 or 8 per cent if n o time limi t is used , and

so on with the other ages . It will be noted that disregarding the time limi t increases the percentage of

passes only slightly .

According to Healy’s standard all normal childrenshould pass the test at Age 1 1 and above . At Age1 1 our table Shows only 34 per cent (time limi t) or

37 per cent (no time limit) passing the test . It is, of

course,impossible to conceive of about 60 per cent of

our eleven-year-olds as below normal . Continuing to


the twelve-year-olds,we find 47 per cent (time limit)

or 51 per cent (no time limit) passing the test . Healy’s

n orm is obviously too high for twelve-year-olds . Cont inu in g up to the 13 , 14 , 15 , and adult age groups, wefin d much the same state of affairs . Healy’s norm ismuch too severe for all ages . It is an example Ofi nadequate standardization leading to the establishm ent of a spurious norm . The danger of thi s in actualclin ical work i s obvious and real . That the danger isreal can be Shown from certain conclusions drawn byPorter

,3 who evidently has Healy’s norm in mind . We

Shall reserve a discussion of these cases until we havepresented the results for the other method of scoring, soas to be able to interpret Porter ’s results in the light ofour norms .

Meth od of Ri gh t a nd Wron g Moves

In this method of scorin g,suggested by Hall

,only

th e nine correct moves are taken into account . A movei s either right or wrong . The results for our Owndata according to this method are shown in Table LV .

T hi s table Shows a distribution of the cases according

TABLE LV.

Percen tage Di stri buti on of Number of Correct MOi es

NUMBER OF CORRECT MOVESA g e 0 1 2 3 4 5 6

3 9

0 7

3 .

3 Porter, F .,Difli cul t i es in th e In terpreta tion of Men ta l T ests

T ypes an d Exam pl es, Th e Psych . Cl i n i c. 9 : 1915 .

OTHER METHODS OF SCORING 89

to the number of correct moves . It is to be read as

follows : At 5 years 35 per cent of the cases made n ocorrect moves

,15 per cent made on e correct move

,15

cent made two correct,15 per cent three correct , 10

per cent four correct,10 per cent five correct , and n o

ch i ld made more than five correct moves . The otherages are to be read in a similar way. The table showsthe natural tenden cy for the number of correct movesto increase with the higher ages . The percentage ofpersons at each age wh o made a perfect performancecan be seen in the last vertical column . It is interesting to note that Ages 15 and 13 Show the highestpercentages for absolutely correct performances .Table LVI presents the medi ans and 25th and 75th

percentiles for our data for this method of scoring .

TABLE LVI .

Number of Correct Moves

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0

The first vertical column gives the age,the next the

number of cases, the next the 25 percentile score, thenext the median , the next the 75 percentile score andthe last the quartile

,i . e.

,the semi -interquartile range

,

which gives an indication of the amount of variationamong the middle fifty per cent . The median shows

90 THE PICT URE COMPLETION TEST

a steady increase from (interpolated) correct movesat age five to 7 correct moves at Ages 12 to adult

,

inclusive . For a rough method of scoring thesemedians will suffice , but they cannot compare withthe system adopted in th i s work

,Since no regard is

paid to the type of errors made . Two ch i ldr en,both

of whom get seven moves right and two wrong cannotbe di fferentiated by this method

,whereas the two

wrong moves in each case may be indicative of verydifferent types of ability.

Our data may now be used for a comparison withHall ’s results . Hall presents her results in number oferrors . Subtracting our medians from nine, we getthe median number of errors and we can compare ourmedian errors with Hall’s median errors for the Sixage—groups tested by her . This comparison is asfollows

The similarity of the results is strikin g ; the onlydifference occurs at Ages 10 and 12 , where it amountsto on e error . It is to be remembered that Hall testedonly 30 children at each age and that her twelveyear-old children are notably poor on all the tests reported by her . We think we may say that our medianis a better index of twelve—year—Old ability in th i s test .In the light of these two methods of scoring , we

may now examine Porter’s special cases, which wereevidently published before any other norms , with theexception of Healy’s

,were obtainable . We are able

.92 THE PICTURE COMPLETION TEST

standard— 7—year ability . Just what we Should expectin a boy of : th at mentality according to the Binet .Other cases Show a similar agreement or di screpancy

between our interpretation of the result of the testand the comments offered by Porter . At times wefeel that her interpretation of the result of the test istotally inadequate and wrong , and all the way throughat best we can have little better than good guess-workwithout an adequate standardization for purpose ofcomparison .

This discussion of other methods of evaluating perform an ces in this test leads us to the conclusion thatthey are lacking in many respects in comparison withthe method of scoring proposed by the authors . BothHealy’s norm and the evaluation by number of errorsare too crude for a complex test of this nature . Thereare so many possibilities that we ought to score theseaccording to their ease or difficulty, and thus differentiate between many different types of performance .

CHAP TER IX

THE T IME

As stated previously,a record was kept of the exact

time taken by each Observer to complete the test .The time was taken for the first period , i . e.

, until theobserver had signified that he had fini shed , and alsofor the second period

,i . e.

,at the completion of the

entire performance after he had been asked by the exam in er whether he wished to change any of the blocks .In this work the results for the complete performanceonly have been used .

For each age a di stribution of time in minutes wasmade . A summary of th i s distribution is given inTable LVII . In this and in Table LVIII the time

TABLE LVII .

Percen tage Di stri buti on of Cases for Each Peri od of T ime

TIME IN MINUTES

0 25 0 20 0 10 0 10 0

6 23 3 16 6 14 0 5 3 6 7 2 7 27 17 3 18 0 13 3 10 0 12 6 5 3 6

8 19 8 15 1 16 5 13 8 7 2 5 9 1

6 28 0 17 3 12 0 8 7 6 0 4 0 2

0 29 3 22 0 10 0 6 7 2 0 2 0 0

4 30 8 18 2 11 3 3 1 2 5 1 3.

3 30 0 2 1 3 6 7 3 3 0 70 28 7 16 6 2 0 3 3 0 7 0 79 30 2 10 1 2 7 1 80 27 0 3 8 5 80 17 4 8 7 7 6 2 2

intervals are all expressed in minutes . Minus onemeans all fractions of a minute from 0 seconds to 59seconds ; on e means all records from one minute up to


on e minute and 59 seconds , and so on . It was deemedunnecessary

,as th e results will Show,

to give fractionsof a minute .

In Table LVII the percentage of cases at each ageis shown according to the time taken . Glancing downthe vertical columns it will be noted that practically n oobservers finished before one minute

,and not more

than per cent in any one age-group finished withinon e minute and 59 seconds . In the third column (2minutes to 2 minutes and 59 seconds) we note at eachage a fairly large percentage of cases ; the lowest is 8per cent at Age 7 and the highest 50 per cent at Age15 . In the 3-minute column we find again a largepercentage of cases at each age ; the lowest is 17 percent at Age 7 and adults

,and the highest 30 per cent

at Ages 11,12

,and 14 . In the 4-minute column the

percentage on the whole drops off as compared withthe two preceding columns . In the five mi nute column the percentages are on the whole still lower.In the remaini ng columns the percentages continue todecrease very rapidly. From th i s point on ,

there arealways zero percentages at each age .

There is therefore little di fference in the time takento complete the test at each age . The table Shows usthat in general there is a slight ten dency for longerperiods of time among the lower ages . Th i s tendencyis by n o means uniform . The largest scattering ofcases occurs at Age 6 . The general impression ob

ta in ed from the table is the close Simi larity of thetime period requi red to complete the test from age to

ag e . The number of the cases in the 2 and 3—mi nutecolumns seems greater at all ages than the number inthe other columns . This can be seen from the following summary. Taking the cases that took from 2 to


age to age very slightly. Again we see the tendencyfor the median to decrease as we approach the higherages . The percentiles bear out what was noted inthe table of distribution , that the greater number ofcases lie with i n relatively narrow limits of time . Therange indi cates the extreme variations .It is obvious from this discussion that we cannottake the length of time as a measure of the in dividua l ’sexcellence in the test . The distinct di fferences inscore denoting the excellence of the performance as weproceed from a lower to a higher age-group are so

marked as to make any difference in time of n o con se

quen ce in th i s test . It would seem to be en tirelyunjust to penalize a performance because of the lengthof time taken or to award more credit for a performancesomewhat quicker than the average . The differencesin time from age to age are not sufficient to warrantthis . It may be desirable for practical purposes tolimit the time to eight or to ten minutes . This limitation is to be recommended in the use of the test in ascale of performance tests . It would be a justifiableprocedure

,since we must remember that the examinee

is doing many other tests and is n ot to be fatigued byany on e test . If a limit of 10 minutes were imposedvery few of the cases would be affected by it , sincenever more than 5 per cent of the cases at any agetook more time .If we examine a little more closely the cases thattook ten or more mi nutes to complete the test , wefind that the maj ority of these occur in the lower ages .All but two occur at nine years or below. The excellence of the performances taking ten or more minutesis as follows : At Age 5 there were n o cases ; at Age 6there were 13 cases , of which 10 scored above average

THE TIME 97

for that age and 3 below ; at Age 7 there were 8 cases ,4 above and 4 below the average score ; at Age 8 therewere 10 cases , 6 above and 4 below the average score ;at Age 9 there were 8 cases , 6 above and 2 below theaverage score ; at Age 10 there were no cases ; at Age1 1 there was one case below the average score ; atAges 12 to 15 there were no cases ; among the adul tsthere was one case , and it fell below the average score .

Th i s seems to indicate that there is little relationbetween the long time taken and the score made .

Most of these long-time cases scored somewhat betterthan the average for their age . Th i s fact again em

ph asi zes the uselessness of the time taken as a measureof the excellence of the performance . The conclusionis , therefore, that time Should not be taken into a o

count in evaluating the excel lence of a performancein the test . The onl y restriction in regard to timethat might arise woul d be a practical on e, and, if atime limi t of 10 minutes were imposed

,some Slight

penalty would attach to those taking too long a time .

CHAPTER X

CONCLUSIONS

The justification for this lengthy and minute analysisof on e test lies

,we believe

,in the increasing demands

made by clinical psychology for more effective andreliable measuring instruments . It seems to theauthors that these demands cannot be met except byvery detailed studies of individual tests . Studies ofthis sort are beginning to appear on all sides . Thetime of rough and ready standardization is passed,because we are beginning to realize its dangers andbecause we are expecting a test to give us more than avague indication of a child ’s abil ity.

It ought not to be necessary to say that this test asstandardized by us must take its place simply as on eamong a group of tests used for clinical purposes . No

child’s mentality can be described on the basis of asingle test , however accurately standardized .

A summary of the chief results of th i s study follows

(1) The test has proved to be excell ently adaptedfor children at all ages and to some extent for adults .

(2 ) The method of presentation used by Healy hasbeen found to meet fully the requirements of practicaltesting .

(3) The 152 0 subj ects tested have proved Sufficientto determ i ne a method of scoring and to arrive atreliable norms for Ages 6 to 14

,inclusive .

(4) The method of scoring has been determinedfrom a purely obj ective standpoint . It depends uponthe relative ease or difficulty of any of the 369 possiblemoves . The ease or di fficul ty of a move is determinedby the number of individual s that make that move .

98

INDEX

Log ica l errors, discussion of,

Cattel l , 2 39, 57

Col er, 78Compl etion Meth od, Ebbing Moves, Rank of, 626 .

h aus, 9T rabue, 9 Norm s, 701i .

Corre la t ion, 2 , 59ff.Year Sca l e, 77

TCube est , 6Otis, 5

Ebbing h aus, 9

Farrand,2

Fem al d, 9

Fish er, 7Form -Board , Segu in, 5, 6

Ha l l , 86, 88, 90Hardwick, 5 , 78Hart, 2Heal y, 6, 9 , 12 , 13,19, 2 1 , 2 2 , 39, 40 ,

Henri , 4

Intel l i g en ce 'uotient, 7 Ra ce, Differen ces due to, 8 1f.Coefficient of, 7 Recording , Meth od of, N i .

Pa terson , 77Percen ti l e Meth od, 7, 8Percen ti l es, 701iGraph s of, 74ii

Pi cture Compl etion T est , Auth or’s resul ts for , 2 2if .

Del inquents tested by, 13Description of, 9fl’

.

Hea l y’s resul ts for , 121i .

Ha l l ’s resul ts for , 14ff .

Psych opath i c cases tested by ,13

Universi ty studen ts testedby , 13

Pin tn er , 6 , 7 , 77

Porter, 88 , 90 , 91 , 92Procedure, Meth od of, 16ff .

INDEX

Sanctis, de, 5Sca l e, Binet-Sim on,Knox, 5Sanctis, 5T erm an , 5

Yerkes-Bridg es, 5Schm itt , 6Sch ool Standing , Influence of,

83ff .

Scores, Determ in ation of th e ,

52 ff .

Distribution of, 71 , 72

T abl e of, 58

Scorin g , Hea ly’s Meth od of

,

866 .

Meth od of Ri g h t and WrongMoves, 88fl

’.

Meth ods of, 86ff .

Sex, D i fferences due to , 78f.S im on , 4

Sim pson, 3Soci a l Status, Differences in,79if .

Yerkes, 5, 8 , 78

10 1

Spa ces and B locks, Con tra c

tions for , 2 2 (footn ote)Spearm an

,2

Standardi za tion , probl em of, 3, 6

Stern , 5 , 7, 1 1

Sub'ects, Description and n um

ber of, 19ff .

Syl vester , 5 , 7

Th e Picture Completion Test - Forgotten Books

Documents

Transcript of Th e Picture Completion Test - Forgotten Books