a
eet
a
~ >"
>
m4
‘ ty tet tee “yet t
: ; : ' : ‘ '
A soehed po okt Ay Uae oe Bu fe BRE BN eh oe on
H is PRPS E SRE | +449 ‘ o>
ig ; FetaeRAIER 8, bef CAD bott balk By °
a teed i bas
THE UNIVERSITY
OF ILLINOIS
LIBRARY
SN26 »
M42 Me
Return this book on or before the
Latest Date stamped below. A
charge is made on all overdue
books.
University of Illinois Library
I eyt F Por
M32—30715
i ee
" Stee > eee tee a “
- « coe r- : lat
o .
s
2
Bet
ar Gs
d
MEASUREMENTS OF
MECHANICAL ABILITY
BY
Joon L. STENQuIsT, Pu.D.
TEACHERS COLLEGE, COLUMBIA UNIVERSITY
CONTRIBUTIONS TO EDUCATION, NO. I30
PUBLISHED BY
Teachers College, Columbia Aniversity
NEW YORK CITY
1923
Copyright, 1923 | :
é By . 4
JOHN L. STENQUIST
G Was hard law
etter te es F180 Cr
AT, WG
Jt +2We.
Pbcee+
ACKNOWLEDGMENTS
To Professor E. L. Thorndike, who is at once the inspiration
and guiding genius of all who are so fortunate as to be associated
with him, is chiefly due whatever merit this study may have, and
grateful acknowledgment is here made of my great indebtedness
to him. Very great credit is also due Professor H. A. Ruger for
his unfailing personal interest and constant helpful counsel.
Professor W. A. McCall has given much help in the statistical
treatment of the data. To the principal, assistant principals,
and shop teachers of Public School No. 64, Manhattan, credit is
also due for codperation in the giving of many tests.
Ajai ue
Digitized by the Internet Archive
in 2022 with funding from
University of Illinois Uroana-Champaign Alternates
https://archive.org/details/measurementsofme0Osten
IV.
VII.
VIII.
CONTENTS
PART |
A DESCRIPTION OF THE TESTS *
. INTRODUCTORY
. TESTS OF GENERAL MECHANICAL ABILITY—
Definition of Terms; Nature of Tests Used
: ae Pome OF ASSEMBLING T&sTS—ORIGINAL SERIES [
. Models Included in Original Series I
2. Method of Giving and Sagringy
3. Results with Original Series I
MEASURES OF 697 CHILDREN IN > eee eee ABILITY
Scores of Normal Children. en oe
. RESULTS AND Conct.ustons FROM. THE FIRST EXPERIMENT
VI.
Sune ae OR ASSEMBLING Test—Orictnat SERIES II
. Models included in Sefies II 5
2. Method of Scoring ¥
3. Results with ae Serie el
4. Conclusions ke
Re
RECOGNITION OF Mecuanican, DEVICES OR MECHANICAL
INFORMATION TEST— * #
1. General Nature. #.2.%
2. List of Mechanical Devites i in NRecneniion Test 7
3. Results with Recognition Test
4. Correlations
5. Relative Gaminenness? A Rach Device
6. Conclusions
SINGLE MODEL SERIES .
I. Single Series I .
2. Models Included in Shiela ae ar
Preliminary Trials with the Single Model Ste
3. Single Series II ENS
4. Models Included in Single Sho IL
Preliminary Trial of Single Series II with Sidoie Series I
5. Correlation of Each of 20 Models with Criterion .
31
33
33
33
* The Mechanical Assembling Tests herein described may be obtained from Chas.
Stoelting Co., 3037 Carroll Ave., Chicago.
The Picture Tests of Mechanical Aptitude are published by the World Book Co.,
Yonkers, N. Y.
v
Vil
is
XIII.
XIV.
eV
XVI.
XVII.
Contents
Vigan : .
. A New Method of Sealine: The McCall Method .
Advantages of the Method hea
2. Relative Difficulty of Each Model .
3. Old Order and Final Order of Models
4. Difficulties in Obtaining Certain Models :
5. T-Scale Values for Each Raw Score of Series I and Setes
lie ree EM §, Tk ee
a) The Binal peoane a 2
b) The Adult Norms
c) Grade Norms
d) Girls’ Records
. FORM OF DISTRIBUTION OF MECHANICAL ABILITY
. THE PARTIAL SCORE FACTOR
The Short Form Test
. SERIES III, ASSEMBLING TEST, FOR LOWER GRADES .
Models of Series III for Grades 3, 4, 5 and 6
SUPPLEMENTARY MODELS
RELIABILITY .
CORRELATIONS
1. With General Latellivences ;
2. With Other Criteria of General Mechanical Ability ;
SUMMARY OF ASSEMBLING TESTS .
MEASURING MECHANICAL APTITUDE BY MEANS OF ILLUSTRA-
TIONS; PICTURE TESTS OF MECHANICAL APTITUDE
1. Aim and Purpose .
. Description
a) Selection of Siibece Matter
b) Scoring; An Improved Method
3. Picture Tests 1 and 2. Jipree
a) Scale Difficulty Values . ;
b) T-Scale Values for Each Raw Store :
4. Reliability of Picture Tests
5. Correlations with Assembling Tests bad with Ehon Regis
6. Summary of Picture Tests of Mechanical Aptitude
NS
PAA I LeeeL
THE NEED FOR A BROADER DEFINITION OF GENERAL INTELLIGENCE
XVIII. ILLUstrRious SCHOOL FAILURES
XIX. THe LARGE PERCENTAGE oF ‘‘Low INTELLIGENCE”
XX. WHAT IS GENERAL INTELLIGENCE?
76
78
79
XXI.
¥Y XXII.
¢ XXIII.
XXIV.
XXV.
XXVI.
XXVII.
Contents
OTHER KINDS OF INTELLIGENCE
stapes INTELLIGENCE AND MECHANICAL ABILITY
. The Intelligence Tests .
/2. The Mechanical Tests . ; ‘
a) Analysis of Total Peecnhition é
4b) The Trustworthiness of the Meagure mente
/c) The Validity of the Measurements .
THE RELATIVE IMPORTANCE OF THESE Two KINDS OF
ABILITY
FICTITIOUS STIGMAS .
SUMMARY OF ParT II
APPENDIX
ASSEMBLING TESTS—DIRECTIONS FOR THEIR USE .
MECHANICAL APTITUDE TESTS—DIRECTIONS FOR THEIR USE
Vii
PAGE
81
82
82
82
86
86
87
89
90
gI
92
99
XIII.
XIV.
XV.
XVI.
viii
TABLES
. Frequencies of Scores Attained by 432 Children in the Original
Series I Mechanical Test .
. Illustrative Results with Original Series I .
. Distribution of Scores for College Students for Each Model—
Original Series IT
. Time Per Model—Original Series II
. Distribution of Scores in Case of 100 Eighth Grade Pupils .
. Coefficients of Correlation Between Recognition and Con-
struction Tests and School Subject
. Percentage of Right Scores for Each Model with S. D.
Equivalents .
. 5S. D. Distances of a Given Per Cent Above Zero
. T-Scale Scores for Each Number eae Series I, with ey
Distributions
. T-Scale Scores for Each Number esi Series II, with Age
Distributions
. Correlations Between Scores When Counting Only Models
Perfectly Solved, and When Counting Partial Scores
. General Scale Values in Terms of S. D. for Grades 6, 7
and 8 .
Average Difficulty of Each Element of Picture Test I
Average Difficulty of Each Element of Picture Test II .
T-Scale Scores for Each Number Right for Test I with Age
Distributions age UC mo, Get ih. Ue ee eae
T-Scale Scores for Each Number Right for Test II with Age
Distributions EUS LT REREAD OR CMs! 2
PAGE
9
13
19
19
20
24
38
44
45
46
53
55
67
68
79
71
FIGURES
Cut showing Original SeriesI . . . . . . +. . . facing
Gut showing Original Serieslie.. 4. a) ee ee
Cut: showing’ Recognition Test. 6. ks Gk eens)
. Cut showing Single Series I—Final Form. . . . ._ . facing
. Cut showing Single Series [I—Final Form... . . facing
Scale Difficulty Distribution of Models for Series I and Series II
Form of Distribution, Series I, for Grades 6, 7 and 8
Form of Distribution, Series II, for Grades 6, 7 and 8
Form of Distribution, Series I, for Grades 7 and 8, Individually
Form of Distribution, Series II, for Grades 7 and 8, Individually .
Form of Distribution, Series I, Men in Army .
Cut showing Series III for Grades 3,4,5 and6 . . . _ . facing
Form of Distribution, Picture Test I, Grades 6, 7, 8, Combined
Form of Distribution, Picture Test I, Grades 6, 7, 8, Individually
Form of Distribution, Picture Test II, Grades 6, 7,8, Combined .
. Form of Distribution, Picture Test II, Grades 6, 7, 8, Individually
Correlation of Four Intelligence Tests with Four Mechanical Tests
Correlation of Four Intelligence Tests with One Assembling Test
Correlation of Four Intelligence Tests with Picture Test I
! | . | r ; } a tv ON ae wil sie vale ie av Vieoure de "
76! " : f
7 iia
| et. Fe y) ig iy Jee AR! ey }
Ne arpa ae OP Sa Ye # ae . LK da? aa the, in Mi AY ie
y eS Pe , ‘ :
’ a v . . wa ' t
‘bh eee a: a Ra MS |S Ani SAE Ma ag nays
ay 4) ‘ 1 ’ Coa) : hh y ‘ Ne
y ' \
, id re! 1 Ps f A
at "
is i J orw : , ¥
in, ‘ Feat “a ¥) ij vy
oy - NF. 94 { t vA
! CAN? ea Ritise j
‘ 7
* é ~ * 7% ; . > te "
\ ; { Avy ‘ er Lara hs s eta Uailieg a y are
ALN ‘
iv ee ’ : 4
+ P ; , P a. socuh'® i
f re 7; x ‘nia Yin eee Ae Tes AS?
‘. : os) omer e = i’ mers | : an
ah | ! Hi » daa ERAN ae
ae i . yi %A
fi. : : us io ty, al .
Bie byrae. | | i Bae That & Wr cule) Bo is eek a,
iy ry Jr): ~ 4, > } , * ivi, iM i; at Sa a i
‘ pai ey Ms
PAR Dict
SECTION I
INTRODUCTORY
Tuis study presents descriptions, results, and conclusions
resulting from experiments with mechanical tests carried on over
a period of four or five years. The important feature is probably
that it deals mainly with a new type of test material, namely,
common mechanical articles of everyday life adapted for use as
tests under standardized conditions. In addition to this, how-
ever, are the results obtained in the use of picture tests dealing
with similar mechanical objects, and mechanical situations,
designed to test mechanical information, aptitude and ability.
Little has thus far been done to make mental tests less aca-
demic and verbal, despite the great interest that has sprung up in
the general field. Yet it is well known that a large percentage of
the population is ill adapted by nature and by training to excel in
the verbal, pencil-and-paper tasks that are imposed by the aver-
age mental test. By general agreement many of these are called
measures of general intelligence, but it is certain that many abili-
ties which could well be termed general are not measured by them.
Any means of examining into the more or less unexplored abilities
otherwise not reached is therefore important and this has been
the guiding notion in the present research.
e
The tests described touch but a small portion of mechanical ~
activities that can be tested, but within their range they are
believed to be significant. They deal specifically with the world
of objects,—real things, as distinguished from words, and involve ,
both mechanical skill and abstract mental ability. While their
nature is essentially mechanical they are in no sense trade tests,
but should rather be considered tests of general mechanical in-
telligence and manual aptitude. The picture tests do not, of
course, test skill in the sense of providing objects for manipula-_
tion, but the ability to answer the problems correlates well with
such skills.
2 7 Measurements of Mechanical Abtlity
But the use of actual objects or mechanical devices as test
material involves disadvantages as well as advantages,—disad-
vantages in that physical objects are always more cumbersome
to handle and to manage than printed forms. They are more
expensive and require more time to use; they involve various
minor difficulties, such as differences in supposedly identical ar-
ticles due to minor details; such, for example, as the differing
tension or stiffness of supposedly identical springs, etc. Models
also wear out, are broken, bent or otherwise spoiled.
The importance of measuring this ability, however, far out-
weighs the obstacles met in the mere nature of the materials. It
is well to keep in mind that modern life is permeated with ma-
chines and mechanical devices on every hand, and that the ability
to handle them is daily becoming more and more important to
every one. We should also keep in mind that while but a small
fraction of the population is engaged in the manufacture of this
multitude of devices and machines, every individual in modern
civilized life is concerned directly or indirectly with their uses.
Ability in this direction is therefore of increasing importance.
The past two or three decades have forced recognition of the
importance of the general field of manual or industrial education
and there is now scarcely a school that does not make some
provision, no matter how inadequate, for manual work. An in-
creasing number of elementary schools also now provide so-called
prevocational courses for pupils above the 6th grade. The
choice or rejection of mechanical courses by the average boy is
apt to be on the flimsiest grounds, and it is here that standardized
tests of general mechanical aptitude will be useful. Enormous
differences are found among children of the same age or grade and
it is believed that tests, such as those herein described, will prove
useful in more intelligent, educational and vocational guidance of
pupils.
No claim is made that the whole problem of measuring me-
chanical ability has been solved,—only that a small but specific
contribution has been made. In the use of these tests, as in the
use of all others, it is necessary to continually counsel the need of
careful interpretation of results obtained, liberal use of common
sense, and due consideration of all other factors involved.
SEcTION II
TESTS OF GENERAL MECHANICAL ABILITY
DEFINITIONS OF TERMS; NATURE OF TESTS USED
The term Mechanical Ability as here used means general
aptitude in the management and manipulation of things me-
chanical. It implies a general knowledge of mechanical princi-
ples and usages, but does not imply any special trade skill. The
tests described have been designed to measure the general me-
chanical ability of young people of school age, who have learned
no trade, but who may have much or little potential ability of this
kind.
Possibly it would be more appropriate to designate these tests
by some_other name for they are mechanical only in a limited
sense. The only mechanical skill involved is that of assembling,
and this is, as every one knows, but a small part of the multitude
of mechanical skills. -On the mental side they call for the ability
to recognize parts of ordinary mechanical devices, for the ability
to make judgments as to the reasons for the particular size, shape,
weight and nature of the parts,—in short, for the mental ability
to think through in some degree the same steps as those employed
by the designer of each machine. Manually, they call for the
dexterity required to put parts together to form the completed
machine or device after it has been decided how they should go.
Much of the performance of a typical child is, of course, mere trial
and error manipulation, in which he hopes somehow to make the
thing work. But the nature of the various models is such that
only a very low score is possible for the individual who depends
merely upon thoughtless manipulation of the parts. A generous
amount of the best kind of thinking is thus required to make a
high score. It involves accurate perception, reasoning and judg-
ment, applied to each model, In so far, therefore, as these mental
processes are of general importance in everyday life the ability
demonstrated in assembling these models perfectly could well be
called general intelligence. But since this term has been largely
3
4 Measurements of Mechanical Ability
accepted as meaning a more abstract ability, it is not thought
advisable to refer to these tests as general intelligence measures,
but rather as tests of the general mechanical ability here de-
scribed.
Two general kinds of materials have been tried: 1. Assembling
tests, in which actual disassembled objects are put together.
2. Picture tests, calling for judgments as to what parts belong
together, and including questions on mechanics and machines.
The idea of presenting a disassembled actual commercial
article, such, for example, as a bicycle bell or mouse trap to be as-
sembled, was first suggested by Professor E. L. Thorndike as a
promising method of reaching certain capacities more or less un-
touched by the more common verbal pencil-and-paper tests. In
order to make them practicable as group tests in schools only such
models as can be given to whole groups of pupils have been in-
cluded. To meet this requirement it has been necessary that all
models be relatively small, light and unbreakable, so that they can
easily be carried about and used over and over, as well as that
they be of such a nature that they can be readily disassembled or
assembled. The final Single Series herein described probably
represents the best types of models. They can be quickly and
positively scored, and easily disassembled by boys after taking
the test.
While it would be desirable to include other operations besides
assembling, this one activity was chosen as representative of many
mechanical tasks and calls less for special trade skill than most
mechanical operations. Thus, assembling is of a more general
nature than, e.g., chiselling, chipping, filing, sawing, soldering,
forging, etc., all of which require at least some trade training.
The picture tests, however, cover a much wider range of ob-
jects and operations, and include questions pertaining not only to
simple and small objects but to large and complicated machines
and processes. .
"[ SOLS [PUISUQ ‘I ‘OT
SEcTION III
DESCRIPTION OF ASSEMBLING TESTS—ORIGINAL SERIES I
The first test tried consisted of seven very common mechanical
contrivances placed in a corrugated cardboard box, 16 by 16 by
2 inches, which could be placed on an ordinary school desk. This
has been generally called the “‘Stenquist Construction Test,’’
Original Series if Fig. 1-shows-its-essential nature.
I. MODELS INCLUDED IN SERIES I
The objects placed in the box were:
2 Carriage bolts with nuts, 2 by 3 inches.
2 Pieces of safety chain containing Io links.
2 Small bicycle monkey wrenches.
2 Round wooden mouse traps.
2 Models made of three angle irons bolted together with
Screws.
2 Small rim locks.
2 Bicycle bells.
In the upper compartment was placed one complete set of the
models, fully assembled. In the lower half was placed an exact
duplicate set, completely disassembled.
The task consists in assembling each model as rapidly and
perfectly as possible.
2. METHOD OF GIVING AND SCORING
Twenty-four children were arranged, one in a seat, in an
ordinary classroom. After a record blank had been filled out, the
following instructions were given: ‘‘Lay the paper which you
have just filled out on top of your desk near one edge where you
can get it easily later.’’ The twenty-four boxes containing the
test materials were then distributed. Holding up one of the
' This test is described also in Stenquist, J. L., Thorndike, E. L., Trabue, M. R.,
““The Intellectual Status of Children Who are Public Charges,’’ Archives of Psy-
chology, No. 33, published by Department of Psychology, Columbia University.
5
6 Measurements of Mechanical Ability
boxes before them, directions were given as follows: ‘‘Turn the
box which you have on your desk so that the letter ‘F’ is toward
you.! Do not look into the box till I say go.?
‘‘Each of these boxes is divided into two parts (indicating by
gesture how the partition extended across the middle of the box).
In the compartment or part farthest away from you there are
seven mechanical models, i.e., seven mechanical things; one of
them is a bolt with a nut on it; another is a small wrench; another
a small chain; and there are four other things.
‘‘In the part nearest to you there are seven mechanical things
just like the others except these are all taken apart. I want you
to take all the parts in the compartment nearest you and make
seven mechanical things exactly like the ones in the compartment
farthest away from you as quickly as you can. As soon as you
have finished them all, raise your hand; and we will write on
your record sheet just how long it took you to do them all.
‘Begin with the one that looks the easiest.
‘“‘If you want to take apart any of the models to see how they
are made you may do so, but you must put them together again.
Screw all the nuts up tight; don’t leave them half on, but don’t
use the wrench to tighten them with. Do you understand?”
(Repeated if necessary.)
“You will now get ready. Grasp the sides of the box so that
you can take the cover off quickly when I tell youto. Are you all
ready? Go!”
The instructions being somewhat long, we found it necessary
after the children began to work to give also the following in-
structions. This was done after three minutes:
“Do the ones that you think are the easiest first. Screw all
nuts up tight with your fingers but do not use the wrench.”
We found that two examiners could manage twenty-four sub-
jects. Assoon asa hand was raised, the examiner noted the time
from his stop-watch, walked over and entered it on the record
sheet of that pupil. The pupil then replaced everything in the
box and put his record sheet in the box ready to be graded.
‘At the end of 30 minutes all children were required to stop
work.
1“*F’’ means front.
'2 We found it necessary to be very vigilant in keeping the subjects from opening
the boxes before the signal was given, as the pressure of curiosity became very great.
A Description of the Tests 7
3. RESULTS
The pupil’s achievement with each of the seven models was
graded on a basis of 0 to 10, by an arbitrary schedule of partial
score values. All perfect scores were given 10 points each. All
seven models assembled perfectly in the full 30 minutes then gave
a score of 10 X7,or 70. An arbitrary value of I was given every
“‘gain-minute,’’ i.e., for every minute of the 30 that remained
after the pupil had completed the test. For example, if the sub-
ject completed the test in 16 minutes, I2 seconds, 14 points were
added to his score. Fractions less than one-half minute were
neglected. Fractions of more than one-half minute were counted
as I.
We found that after a little practice, and with skilled manage-
ment of boy helpers, one examiner and four boy helpers can grade
the twenty-four sets in about 40 minutes.
We had then for each child a record like the following sample:
Score Attained with Each
ge Model Credit for | Total
Br PPPS PP Te yy OP RT Ea FC Eee Time Score
Pie Ge a lo ARC Sau Ca
Waid ic ee tach atl Ee ED hd: 3 e) co a 9 e) 27
Path ay fe arc at FOr Ae (LO [LOW rel 10 3 fe) 63
ac ae ee een 10 | 10] 10 | 10 | 10 | Io | Io 8 78
SECTION IV
MEASURES OF 697 CHILDREN IN MECHANICAL ABILITY
Although the results obtained with this series, Original Series I,
have, as already indicated (page 5), been reported elsewhere, the
essential facts are here repeated for the sake of making this ac-
count complete.
SCORES OF NORMAL CHILDREN
The test was first given to 432 unselected children in a New
York City public school, and the scores tabulated. as shown-in
Table I to yield age norms.
From these norms true norms were estimated to be as follows:
Age
6 7 8 9 10 II ue 13 14 15
to to to to to to to to to to
| 8 9 Io II I2 13 14 15 16
Median Score..... SAMS AAS eS) (RAs A Shoes eS OLGn | O2e50100875 | 76.4 | 77.5 | 82.5
Estimated True
SCOPES «cuss busbers 20 32 42 50 ef 63 69 75 19 82
The discrepancies between the obtained and estimated medians
are due to the allowance made for especially bright six- and seven-
year children.
Having these norms the real work of the first experiment was
begun, namely, to measure the ability of 265 children who were in
institutions for dependent children. Four tests were given—
Binet, Trabue Language, Thorndike Reading, and this mechanical
test.
By utilizing the median score for ages 6, 7, 8, etc., and inter-
polating the scores for each intervening month, a table of age
norms was built up.. It was then only necessary to read the table
to determine the degree of over-ageness or under-ageness of any
child subsequently measured. (Since the test in this original
form has been discontinued the table is not here reproduced.)
8 |
A Description of the Tests 9
TABLE I
FREQUENCIES OF SCORES ATTAINED BY 432 ORDINARY CHILDREN AS TESTED
IN A PuBLIC SCHOOL OF NEW YorRK CiTy—ARRANGED BY AGES
6=6.0upto7;7=7.0 upto &, etc.
Age
Score
| |, es | | eS
_
_
=
_
pb
= me NH WN
IN)
LS) tN
a | on
A
s NW & & WD
tN
—
O
i)
_
NS
Ww WwW
No om
=
~~ = WD = =» me WD
_
Ww Ww
NO
NON
No oN
-~
-
wW
_
_
ioe)
‘©
~
on
10
as eo *
Sse 6.2
4) oom, @ 6
oP. e =
60s ae 3S
é 8 bp a %
Cla Pay Sito mC
we yale |.
Ce ie Ce
ele ee »
oe! 9,6
eee ee
Rs ak ee
6 we le
aye wie
aie! weve
Se. er Ge
hk ape we
CP ae ie
o/ ce. tape
Measurements of Mechanical Ability
TABLE I—Cont#'d.
Age
a i | | | | |
I I I 4 I
I I 2
3 2
I 5 I 5 I 2
2 I 4 I I
I 2 3 I I
I 2 I I I
I I I I I
I 2 2 @ * I 2
2 2 I I
I ee 2 r
I I I I
a. 6 I I
I 2 3
2 I I
I I I I 2 2 I I
I I I I 3 I I I
2 2 I
I 2 I 2 I
2 2
I 3 I I I I
I I 2
I I I I
I 3 B 2
I I 3 2 I
I I 4 2
2 I
4 I 3
I I 2 I I
2 3 I
2 I I I
I I 2 I I 3
2 I I
I I I
I I
2 I 3
I I I
3 I I I
I 3 I I 3
s°e) 6m '%
a A ps
“re @
A Description of the Tests II
TABLE I—Cont'd.
Age
I I I
I 2 2 3 I
I I I
I I 6 2 I I
I I I
I I 2 2 I
I 2 I
I I
2
I 3 -
I
I
I
SECTION V
RESULTS AND CONCLUSIONS FROM THE First EXPERIMENT
As'a measuring device the experiment demonstrated the
practicability of utilizing such materials as have been described.
The interest displayed by the children was intense, and even
those children who were almost complete failures at it were anx-
ious to try. The test as a whole proved too easy (the more able
finishing within 10 minutes with perfect scores) and hence was
probably unreliable for individual deductions, but general
averages are sufficiently reliable. The marked differences be-
tween the type of ability measured by the mechanical test and
the abstract intellect tests is significant. The records of 50 boys
and 50 girls selected at random from the total results for all the
dependent children are reproduced in Table II.
The results show that the dependent children are as a group
about 12 years behind in mechanical ability, but considerably
more so in abstract intellectual ability. The 11- and 12-year-
olds are about 2 years behind; the 13- and 14-year-olds about 23
years; and the 15-and 16-year-olds about 43% years behind.
But the pupils behind in abstract ability are not always behind
in mechanical ability.
The percentage of unlike signed deviations is for the cases cited
about .31, which is equivalent to a correlation of but about .5.
Pupil 29, e.g., is 1.8 years behind in abstract ability but 3.3 years
ahead in mechanical ability. Correlations with subsequent and
more perfected mechanical tests show that the true correlation
between intelligence tests and the mechanical tests is seldom over
.4. Thus it is confirmed that a pupil may be inferior in academic
school work and yet have marked ability in manual activities.
But there is no evidence to support the popular notion of a law of
compensation,—which assumes that low abstract intellect signi-
fies high mechanical ability, or vice versa. Our correlations are
low—but always positive—between the two abilities. .. If we know
that a pupil is above average in abstract ability all we can predict
with regard to his mechanical ability is that he is more likely to be
I2 ;
A Description of the Tests 13
TABLE II
ILLUSTRATIVE RESULTS WITH ORIGINAL SERIES [|
Boys
Under-Ageness in ;
Age Used in Com- | Three Tests of Ab- | Under-Ageness in
puting Under- ___|stract Intellect Com- Mechanical Test.
Identification Number
Ageness Estimates | bined. (+ Equals | (+ Equals Over-
Over-Ageness) Ageness)
tA hte Mit cain. e abies eee snare’ ease TA —2.7 —8.7
TPA cic le dPibiens s+ o's a Wf AE ate Kee Tait — .I +2.1
a: ee ER 8 Ses len 10.0 —1.3 +4.5
hs SR ee arm oping Ai dd 13.8 —3.0 —3.0
= tes Sor eirtqcc merece 10.4 + .1 —1I.5
PIE Rae ow oie aera Ay 10.2 —I.I —2.9
ae ee ee aE ee Abe a P 9.8 + .4 +3.4
set & Me ce die ace ae etee hates 14.? — .8 ae
NS ee Peg ae tis, SV Pe tea 9.8 —1.8 —3.1
fet oe Ee at i RR) S8 eee 12;2 20 +1.4
ny Eee eee Rte. = Ted — .8
ee eMac wae Me O:Slosdis as 14.0 —3.1 —2.2
SAM oe Te ee Se, eee e TAsa: —1.6 +1.1
SURiaAe WAGE ai ON oiehe ig tea a oe 3433) —I1.2 +2.3
a te eat, oe eet ar ee aia T2a0 —2.7 wate
gc Ae Ce ES ay 13.0 + .2 sie
ee 5 ey ee ee CRASS, 13.8 +1.8 —1.3
fe AR An ee, eee ie! 10-7 —1.6 —I1.4
AEE de Re, 2 oad ns Tait —2.7 —1I.0
So Re ee Ea we r27 —I.1I + .3
tr Oe OO ee ces 10.3 + v1 +1.3
eer renee te Oke ce ies Sea eA 13.6 — .3 +3.4
SNES Sd aunt essa) ake tion Siamiteere es 9.9 + .I +3.5
Cie) RONR nes, ara aes ah 4 eee Md, # — .9 — .4
np 20 RO on ER he Ca Ee 12.8 =2):9 WERE
Sn ct! St ERR SA Peer 12*4 —2.4 ae
POM Ee canes, Maciek e. 14.? —3.0 ae Bes
Pos 6 Bll eae ae Gee ee TIZ0 — .3 —3.4
ER a SNE ara, 2. 0 «ois stecale si'ehe wes 13.0 —1.8 +3.3
98S ok re eee A ESP —3.9 —3.1
Payee Aa ticle oss Chee hes 14.3 —2.5 oa ot
piet ie “ie eek et age aie SESE Ee a a 13.9 —3.3 a:
Bet 1 de OO ee 16.6 —3.3 + .7
Nei ine Mirycioe Pa Py aa 10.? —2.1 Stat
ein, ade ete erate at eee cs. BBecees ae 10.9 — .5 + .4
ad Sale a feokrn she Seeletae ate ne ats Tie + .3 — .9
APP Oe nce eerie os oe 12.0 +1.3 +2.3
} hglgne Wiel eels, oe Aa she ies os oes 10.5 —1.0 —3.6
SU ee re She Meee 9.8 —1.8 —1.5
AG ee ic > 2 Bos cbr 10.0 —2.5 —2.7
BIN i gal ol ok, « CR a eke eer T3i +1.9 —1.5
eee te Sh) as ied ch Nous Ate OE I4.3 —2.4 —1.3
hid Sha Pain SAR ater 10.5 — I —2.7
“lacs daly, Sateen Saeeaeaeaes oe! 12.4 —2.1 —2.1
1 A EMAL dies yp: Cae SPR ape dale oe 14.0 + .6 —I1.2
(1h 3 ae eee T4 nf — .4 ide
“ha: 3 A, ean 10.4 “52 —2.6
a A re 13.2 —3.9 —1.6
Men OES Ieee ee ee 10.9 —I.5 —2.2
STEPS 26 ss apis ds es cee ee i023 —I1.2 — .6
Measurements of Mechanical A bility
Identification Number
Ge 6 a Fee 64 eS bl © Letra wb re © a wT we
a a6 we & © to 6 eo eh ee is Wels
ed @ 6 0 Oo 810m 6p Wie e 0 Bw o's ele one
C40 6 She B® © iw Ste B06) 6 a ere mie « Bw
eo 6) ve 6 6 fee © DS Be me 16 0 6) 6 6b
a 10) 6.16, 6. O.6 Siw ie em Ae te. Be ale means
ays © 6 a OG 6h e 6 #06 (sc 8) a oF el «ecm 6
& (7) bh .0 6 lelGre 6) 616) 6 fe 6.0 Ue) © 6 bln ie volte
oo (© 6 @ B10 a Sb 616660 eo 000. © 2.8 0
oS ela Beale, & We. ee 6 OOM RS hel ee Uae
e660 @ 08810) O06) © 0! Gia eke tae © 676
ve ® ele v6 6)0'0' #.6 O10:6. is \eimle we 8 +
o 16 * (6 428 6 0» 6,e eye 6) 6)01 6 4.” & <0 0:8
Ww 6 ie 8 60 8 6 8 wee) 0 ie a8) 0 4 wl 8s
ele, 2/00 * © 00e 6 6502 0 0 8ine 0 6m © a6
oF @y'e bes be 6, Big, 0 sic aww fe wees
coe eeceareeeeeouereeecee
Swe lee o Glee ard) we (lee a) @@ (0 6.618) 6
oe ye) 6) fh V6) ware Bete Ay. .& fee else 616
eee er ee eeeeer es eee evrevens
6 eels @ © oe 6 6) eos & ne Oe 8 0 eo
Sigel rh Oe 6 @. Wh6 oO eW@ Le eee! ere je,
a 6,0 Be 4 6) © 4 109) 6 en! lee ele 6 616 »
ee ee ey
O95. ave) we 16; 018 ee (9 ef) e..8/8w ee eee
bie \¢)'e. 9 0'@ ‘pi 0m /0 4u a wie! wie) e018) 2
Cie) 1c. S 8 6 PB) Cl 0 6 ew sere eis) & bie
sims eh BMC ©. © 6) w6\ ee 6m Oe) 6.0, 6
#8) \ 6, oS lal Se in eel eo ws sie /b.6 ed, 2 pis
id! (¢/, 8) 18 (9) 9S 6 Be # 6 658 1018 (8 ew 6 a ©
uae! eco 6 8s, 0 6 es» Cig 610 0 pee ce
See) ee ww Oe ea he wipe oO) & 0) '6 Ve
6.4) id. 6. @ fo @ A) SO. 8 10-19), bie 0 6 O68
OSES .e bh SB, Fs BOW eels @felg av) ene aye
6b. Vb) ae 60 eae Rie ele lols bo Oy
See! 0 oe War @ 476) e.e wine je) site) » te
o. 6 0 4) ole el 19, O10. (h Ol 8 6 16 wisielle. ap
o 40 B86 SP SLs oe @ 3 ec bie ele (> «0 6
aime Whe w bie 618 Sie 9 ‘0 oie. A Sem ie) © (e
o 2 ©) e @ 6 Ge) 618 Ace, Se 1 ee a e186)
Bitte 6° m 8 OFO O10 e 66) e/etenere i@ 6.0 sb
© ee eke (2 0) 6.9) ©, © oes ae el ele a als
$y ewe (6:4 6) \0 0.0 ee) mw ate, ele mb me
ee © a) 6 6m 6 O90 dee ec he a 48 Se
Sie 3 lols 5) dS 8) 6.6 0 6) 8.0.6 61s lela ee) =
We 6. 8 10 p16 S 8 he © Be 6 we \plenetp ie fol
Bie wiley aE Sep (9) O10 6 © ele oie) 0) oa! ee
BO eee a Oe O66 6) 818) Sica tee wae 6 ae
% 6 0.8 6) ©) om 6.9) 8) 6 0.6.6 ir pe 19) « ele
TABLE II—(Cont?'d)
GIRLS
Age Used in Com-
puting Under-
Ageness Estimates
Lal
La]
NnNOoOCnUW
©
NUOhwW
i ha
°o .) Ve) -
lo He -Romome -) MW CwNH wOOWR wih oO lo meat NRHN OW
©
HO ORN
Lan!
Lal
OR NV Y
Under-Ageness in
Three Tests of Ab-
stract Intellect Com-
bined. (+ Equals
Over-Ageness)
|
LI
ONDOnD
|
ONAUNDA
N
AKUMA Awo~ar.8
a: | | |
pe Perc ae? a Bags eee weg ey OM Roa)
CORWE WRUNG BWRHACR BSHHND CHIHD
4:
Ans ow
Under-Ageness in
Mechanical Test.
(+ Equals Over-
Ageness)
|
>)
RO HH Ob
~I
MOISCA Annan
|
peed
NWR H
| |
ies) .
HUAAWW += RWON
|
sheath >
WO: hy
=P
cel
ICNP
hROHD
|
woo! |
A Description of the Tests 15
above average in it also, but there are many chances for him to be
below. It is clear that mechanical ability is not measured by
ordinary paper mental tests and that it is worth while to further
develop the type of test materials here tried out.
With this in mind a second series of models was accordingly
designed and tried out. This series is called Original Series IT.
SECTION VI
CONSTRUCTION OF ASSEMBLING TEST—ORIGINAL SERIES II
Experience with Series I indicated the need for a series of more
difficult models, in order that the test might be extended upward
into high school and college grades, and also the desirability of
more substantial boxes. Accordingly, after much search for
suitable models six considerably more difficult than Series I were
selected. These are shown in Fig. 2, together with the improved
box.
I. MODELS INCLUDED IN SERIES II
The models are:
Model H. Two straps buckled together in a complicated
way with two buckles, four slides and two rings.
Model I. A wall electric switch.
Model J. A large rim lock.
Model K. An ordinary electric bell.
Model L. The works of a pendulum clock.
Model M. An electric light socket.
As in the case of Series I, a duplicate model not fully assembled
was included so that the problem here, as in Series I, was frankly
one of copying each model by building up a second model from the
parts. In this series each assembled model, together with all the
parts of one disassembled model, was placed in a separate com-
partment provided in the special reversed box, and not mixed as in
Series I. :
This improvement eliminated the miscellaneous sorting of
parts, although, of course, it also eliminated that feature of the
test which called for identification of the particular parts of each
model out of the entire mass of parts. But this sorting process,
while no doubt a valuable test in itself (later tried out in a different
way—see Recognition Test, page 21) was not the kind of reaction
which was sought, besides it is wasteful of time. The object
16 ;
‘[][ SOMas [eUISIIQg *% “DIY
A Description of the Tests 17
here was to test more strictly for manipulative skill. The cover
of the box was designed to open toward the person being tested, to
form a tray in which to work to avoid losing parts. A large and
small screw driver and a pair of tweezers were included in this set.
The test was given in the same manner as the preceding Series I,
except that at least 50 minutes were found to be necessary.
2. METHOD OF SCORING
The credit given for each model, when perfectly or partially
assembled, is shown by the standard score sheet below. After
the test the scorer examined each model and entered the score on
record sheet which had been signed and placed inside the box by
each person examined. The models were then disassembled to be
used again. Boys who scored high in the tests were found to be
ideal helpers.
CONSTRUCTION OR ASSEMBLING TEST—ORIGINAL SERIES II
STANDARD SCORE SHEET
Grade All Models on a Scale of 0 to 10
Mover H (Strap) Score or Deduct
BERET TONNE Pe tucson) Os Vie BR sk feed hens a < oom,
peatiier tran Tevereed 01,5 «dU AMEE uh o's 5. ee e's sees 8
RATIO DAR VOTE Ct SOE ak ties UK a rice 6
SeuErOHCIe Wie Il aly WAY 20). 047 ec es. Alcw ne a ded: a
Eee ILL Or, Wrong, fOr CACM (ant poh iis ne ae + 5s 2
IRE ee eI ok.) oc Son ee tied Sik gee es cde: ITO
Mope I (SwitcH)
INO Sitters 7c na. aie Biioe cee ee ob sO. we oO
One contact wrong or omitted...
Both contacts wrong or omitted. ....
Bracket Wrong ci pchea cea ee ee es die oe eee oes 3
SRELIOCE ST, Vy 3. «1s SG EE TE Gi antes din eh hi ins se Wiltewrada lL
oor
MopeEL J (Rim Lock)
No attempt. Seats
Spring loose, Nae Over tepicttea helt GE rare a oe. Ae er
Spring all wrong or omitted. ali. Be gee BRN st, 5
Revolving cam not oa, in eaietting ath carck i ihaien elS
OD et EE RR ES SEA REO Bs ee ena ys ORE ee eRe? 10
18 _ Measurements of Mechanical Ability
MopeE.L K (BELL)
INO AEDES DE. Aksiooe & ae Meine alan peice iets Gop e Ser
Wires wrong with respect to washers, for each....
For each washer omitted or misplaced. .
For omitting or misplacing small square nocierione each.
For each case of wrong screw used...............0..00-
The whole thing about half solved....................
MopeEL L Nair
No attempt.
ae a
21.6 wo
For works parity Assembled Allow for Beh Binion in nalace We
Works all assembled but top frame not in place.....
PReteet (ORO aa ni A ns 1D ceberanars (eee ibe eRe
Novattempti. ay as neste eis GMD nian Senn tery ante
Lower disktinvertede ux irc ons see ee eae ae
Upper disk invertedtes (0) GA ae ee a ae
Small nut omitted or misplaced...
For omitting small black center pin bearing...........
All properly assembled but no tension in spring... .
This model frequently occurs in a very mixed-up condition; in such a case
judge as to whether the whole effort represents that the problem is one-half,
ee a i”
one-fourth or three-fourths solved and grade accordingly.
NotE—There will be cases where the degree of achievement does not cor-
respond to any of the values given. In this case the obvious procedure is to
judge it in terms of the case most like it.
TimME.—The standard time is 50 minutes, although this has been varied.
3. RESULTS
Records were obtained from the following groups, the highest
score possible being 60:
No.
Freshman engineers, Columbia (1915)... 35
Teachers College graduate students (1915) 29
Efficiency men, silk factory (1915)....... 30
Freshmen, Mass. Institute Technology
(1916).. es 40
Freshmen and econ year, Moy encanart
Tastitute (1996) calhy Wawa ae 58
The results soon demonstrated that the idea of utilizing these
Too
Av.
43-4
48
very difficult models is impracticable for school purposes.
much time is consumed in both giving and scoring the material.
It is too bulky and awkward to handle in classrooms.
= Nom ms
m= WH WH
It is also
A Description of the Tests 19
difficult to assign proper partial scores to a model that may re-
quire 30 minutes and be greatly affected by luck. This series
was accordingly never extensively used.
The distribution of scores for 190 cases is shown in Table III:
TABLE III
DISTRIBUTION OF SCORES FOR EACH MODEL
[ORIGINAL SERIES II, IN 190 CASES OF COLLEGE STUDENTS AND OTHER
ADULTS}
Score
Model rrr 1 Otal
fe) Iv 6-9 10
FISCSIOE) Ge ah ek as ee 3 7 47 133 190
PRS osteo) OY a eae Oe ca ar & 14 16 14 146 190
LIC Res ha neti oe 15 12 62 101 190
Rethiectria’ Bell igs aces Ss 19 32 65 74 190
EAECIOCIE) odbc ids Gas a ee 56 24 17 93 190
M (Electric Socket)........ 120+, |) ~29 15 26 190
The order of difficulty is shown to be approximately the order
in which the models were arranged in the box, i.e., the order in the
table. The frequency of zero scores is exactly in this order. The
scores indicated above cannot be taken, however, as entirely re-
liable for models L and M, asa large number of persons worked so
slowly as to leave little or no time to try these models.
The average time required by 35 freshmen engineers per model
was as shown in Table IV:
TABLE IV
TIME PER MoODEL—35 FRESHMEN ENGINEERS
H I J K i M
Strap | Switch | Lock | Elec. Bell} Clock | Socket
Av. minutes.... 7.4 12.5 7.4 12.8 9.2 8.3
A further group of 100 8th grade boys were later examined by
Mr. Hazen Chatfield in a New York City public school. From
these cases the distribution of each partial score was as shown in
Table V:
20 Measurements of Mechanical Ability
TABLE V
DISTRIBUTION OF SCORES FOR EACH MODEL, IN 100 CASES OF 8TH GRADE
Boys OF 11 EXPERT TEACHERS ”
H I J K L M
Score Strap | Switch | Lock Bell Clock | Socket
hs 7 St, See 8 II 10 16 58 79
Ree Mp 19> Le pate 3 3 I a 3 5
7 MRR 5 Siding Cnc ae 10 12 6 ‘3 fs 5
2 to Ty cia, NG I I a 8 4 I
PRAM vents eo) icy Sp ie 5 5 6 8 5 3
Rha! eee eee I O 6 5 4 2
PER 1h a See I 4 3 4 O I
Tce Sere e re ys.§ 10 6 j I2 2 fe)
BNE Cele, Be I2 4 7 I2 I fe)
a Wray adel» Robs 9 2 29 13 I oO
TORI eee ae 40 52 22 12 I5 4
100 100 100 100 100 100
Approx. Median
SCOres hear: 8.8 10.0 9.0 6.7 O. aye
This table shows that 58 per cent did not reach Model L, and
79 per cent did not reach Model M in 60 minutes. The total
scores reported for each class are therefore largely the result with
four models tried, which is a meagre basis for drawing conclusions
about relative mechanical ability. The time for 8th grade boys
should be extended to, say, 90 minutes, to obtain the benefit of all
models. r)
4. CONCLUSIONS
This series requires more time than is generally practicable in
school testing, and apparently does not yield as valuable (per-unit
-of time-spent) diagnosis as sets composed of longer series of easier
models. It seems doubtful that as good a measure of this type
of ability is obtained in 60 minutes with Original Series II as in 30
minutes with Single I or II (developed later). The labor of scor-
ing is also greater in the former.
‘JSaT UOTTIUSOIDY «LOI
i
Bn ee
Section VII
RECOGNITION OF MECHANICAL DEVICES OR MECHANICAL
INFORMATION ‘TEST
I. GENERAL NATURE
Following out more specifically the idea of identifying me-
chanical objects and mechanical parts by name, a series of small
mechanical objects ranging from the very simplest obtainable to
those comparatively technical, e.g., from a common wood screw
to the parts of a spark plug, were fastened on an 8 inches by 15
inches stiff cardboard, to fit into a flat cardboard box about 14
inches high. Fig. 3 gives a general idea of the appearance of this
test.
2. LIST OF MECHANICAL DEVICES IN RECOGNITION TEST I
The list of names which follows was given to each person to be
tested. The subject was instructed to find the name of each
article in the box and to write its identification number opposite
the name:
a. Bushing for packing nut of t. Fuse wire
spark plug u. Gasket or washer for making
b. Cabinet door hook hose coupling
c. Carriage bolt v. Gimlet
d. Catch for cabinet door hook w. Glazier’s point for fastening
e. Central insulation for spark plug glass
f. Center punch x. Glass cutter
g. Common ten penny nail y. Hack saw
h. Common washer z. Hinge
4. Curtain rod fixture at. Insulating plug for electric light
j. Cotter pin br. Jam nut or first nut for top of
k. Coping-saw blade spark plug
1, Cut nail c1. Lock washer
m. Dowel screw dit. Machine bolt
n. Drive hook et. Main body of spark plug
o. Drill fi. Nail set
p. Eight penny finishing nail gt. Packing nut for spark plug
q. Expansion lug nut hit. Patent box or mitre frame
r. Flat head harness rivet fastener
s. Flat head wood screw 41. Picture nail
21
22 ‘Measurements of Mechanical Abthty
ji. Pipe reducer bushing ut. Stove bolt
kt. Plumb bob vi. Tar paper cap to prevent paper
lt. Roller skate wrench and key from tearing
m1. Round head rivet wt. Thumb nut
ni. Saw screw x1. Wedge for tool handles
o1. Shade fixture for nonrevolving yi. Wedge to prevent window from
end rattling
pi. Shelf stop or support z1. Trunk caster
qi. Set screw a2. Window sash fastener
ri. Small hasp b2. Window lift
st. Soft solder c2. Window shade fastener, non-
t1. Staple for small hasp revolving end
3. RESULTS WITH RECOGNITION TEST
This test was given to 205 pupils of the Forest Park School,
Springfield, Mass., in codperation with Mr. J. L. Riley, then
principal, and Mr. W. R. Cole, in charge of industrial arts courses.
The pupils had been divided into selected classes as indicated
below. The average scores and average deviation of each, ob-
tained in 30 minutes, were as follows:
Average
Score Out | Average
No Grade Group of a Possi- | Deviation
ble 55
20 ori OB upoys Regular TAcT Ses
LOnsiy,. oe) 7B i Practical Arts 20.7 7.5
porcine Mehr ee Muy de. u Regular Manual Training 16.4 G27
By fig MEADS sy 9: i Especially Bright 19.8 4.9
22D AC eR STs Regular Manual Training 20.0 8.0
89 .......| 9B and 9A | Boys, Regular 28.0 6.5
60 ne plies Oeitas Regular 9.4 5-5
Of these, the Practical Arts group were boys who had elected to
take the maximum shop work available, spending much more
time in the shop than any other group. The Regular Manual
Training group spent much less,—14 hours per week in the shop,
—while the Regular group spent even less, and was composed of
undifferentiated pupils.
The Especially Bright class ie composed of pupils selected by
teachers as able to progress faster than the others, being promoted
at shorter intervals.
The average scores for each group given above show that the
~
A Description of the Tests 23
task is too difficult for pupils of all these grades. Even the 9th
grade has an average score barely over 50 per cent perfect, while
the others fall much lower. As is to be expected, the Practical
Arts group score slightly higher than the others of same grade,
The grade progression from 6th to 9th appears to be constant,
suggesting that the experience needed to recognize these 55 ob-
jects and their names is gradually gained more and more by all as
they become older. Judging from these data the average 9th
grade boy knows about twice as many of these objects and their
names as does the 6th grade boy. The Practical Arts 7B grade
group scores slightly higher than the Regular Manual Training
8B grade,—a gain of one year. Original mechanical interest and
ability, as well as the extra training, no doubt contribute toward
producing this result. The girls’ scores show that the test is
entirely too hard for 6B girls.
But only very limited inferences can be drawn from averages of
a single unstandardized test. To obtain checks on these scores
the assembling test, Original Series I, was given to a number of
the same groups. The number who took this test and average
scores were as follows:
Average Average
;
ie. Fata: Sgn Score Deviation
2 ee eee 7B Practical Arts 73.0 7.9
By ae ee eo 7B Regular Manual Training 68.9 6.4
Li 7B Especially Bright 72.8 eee
29 thy, Reoeeorae 7A Regular Manual Training 65.2 14.9
The Practical Arts group again scored higher than the Regular
Manual Training group of same grade, and were again followed by
the Especially Bright group. This test, however, was found to be
much too easy for these grades. The scores are therefore largely
a comparison of the speed with which each pupil could assemble
the models.
4. CORRELATIONS
To obtain a still further check, teachers were asked to rank
their classes in several school subjects. The order of merit in
algebra, geography and literature was combined (tentatively
weighting all equally) into one composite ‘‘school subject”’ rank.
3
24 Measurements of Mechanical Ability a
From these three measures a number of coefficients of correlation
were computed. These are shown in Table VI below. Since
each group was unavoidably small, and ne essarily ranked by a
different teacher, the identity and number in each group is
indicated to avoid giving misleading figures.
TABLE VI |
COEFFICIENTS OF CORRELATION BETWEEN RECOGNITION TEST,
CONSTRUCTION TEST ORIGINAL SERIES I, AND SCHOOL SUBJECT
No. of Boys | Grade Group r=
19.2% Mute 9B Recognition and School Rank ~.08
20. oe ek 8B Recognition and School Rank — .39
Wy eee ey St ys 7B Especially Bright Recognition and School
Rank .02
FO eee anata 7B Practical Arts Recognition and School Rank aat
16) eps 6B Regular Recognition and School Rank .O1
ZO 2vohnd eaR 7B Manual Training (Regular) Recognition
. and School Rank A
Tos acu feat 7B Especially Bright Construction Test and
School Rank 12
TORO OTe: 6B Regular Construction Test and School Rank .08
TOM saute. cas |e Practical Arts Construction Test and School
Rank — .08
ZO Deh cE ant: 7B Manual Training Construction Test and
School Rank .24
DON an tien 8B Recognition and Rank in Manual Training — 31
by Pa ae a AV a 7B Especially Bright Recognition and Con-
’ struction Scores 55
TOF eRe ee 6B Regular Recognition and Construction
Scores .42
LO othr sneenaeon 7B Practical Arts Recognition and Construc-
tion Scores .22
Oras diate eae 7B Manual Training (Regular) Recognition and
; Construction Scores .19
TS eee, ee se 7A Practical Arts Recognition and Construc-
tion Scores .49
BT owe es 6B Regular Recognition and Construction
Scores .47
ZU cts uted 7A _| Manual Training (Regular) Recognition and
Construction Scores ‘7k
Kate ar Sarak ae 6B Girls Recognition and Construction Scores .26
A Description of the Tests 25
The groups being in each case small, the probable error is large,
but the agreement between similar group correlations tends to
obviate this. While the data are inadequate and the measure-
ments crude, there is evidence that the true correlation between
rank in school subjects and the Recognition Test is near zero.
Between the Construction Test and school subjects the correlation
is alsolow. Other data not here available indicates that it is not
generally over .40. There is, however, some evidence of correla-
tion between the two mechanical tests, but the coefficients are
too low to be significant, the average of the coefficients here re-
ported being 41.4. But while it is probable that there are similar
elements in the two tests, mere inspection shows that the two
tasks are of different character. A boy may assemble a dozen
devices without knowing the technical name of any of them.
5. RELATIVE “‘COMMONNESS”’ OF EACH DEVICE
One other tabulation is of interest, namely, the relative fre-
quency of right answers for each of the 55 devices, or the degree
of ‘“‘commonness”’ of each.
On the following page is a tabulation of the numbers of right
answers for each device—for 57 boys, arranged in order of dif-
ficulty.
The results are somewhat surprising in several cases. The
hack-saw blade ranks second, while the coping-saw blade ranks
forty-second. The roller skate wrench and key is the easiest of
all, and the first one on the list, bushing for packing nut of spark
plug, is the hardest of all, while the jam nut or first nut for top of
spark plug is no more difficult than the window lift. The cotter
pin is no more difficult than the glass cutter, and so on.
RELATIVE FREQUENCY OF CORRECT ANSWERS FOR EACH DEVICE IN
RECOGNITION TEST ARRANGED IN ORDER OF DIFFICULTY
57 Boys. 7th to 9th Grade
Forest Park School, Springfield, Mass.
Number
Name of Device Correct
ermeeccer: SICAte Wren aris Bess 5 tale al alba at eat 3 oe 50
MERSOCORTw DIAG eo ae auton ic ntna Gael tert orate ss 48
OE Set The gee BSD oy Ee Pom PR eA e Rename RAN 47
2 OE A i ARIAIRESS OS Coes tba REAP gt gk RA 47
26
Measurements of Mechanical Ability
Name of Device
. Cotter pin. . ae lees wae
. Tar paper cap to apres paper ween tengene Pe ee
4 lst head wood ecrew eo ca, op ie ee ee ey ee ee
> Common washers ts bac: oa ic eee ee ee ee
‘ Curtam tod Ireture ves oe ys eee a eee ee
. Common ten penny nail. . 0 PRE TOEE, PEL SORRY Phe 2
. Window Imit.64 oho RA ee eee 2S ee
J Hight penny finishing naile. J... peseacres )\ demas aa
. Picture nail. ; :
. Glazier’s mee for Fastening ee
Window: sash fasténer’..34) ko tea ee ae eee gee
. Gasket or washer for making hose coupling. ............
. Window shade fastener, nonrevolving end. . rey
op Mette OTB hi tat Nl A dee eahis (Oh cet Nae es
~ Wedgeitor tool handles. 25. Syn. vain ce amet ee cee
. Insulating plug for electric lighta22 os) 07 nee
SPT DOD Sune Re er cae ee ee cee ee Ree
pi FOUSCSW IEE. 727, 3.2L Tp eine A ee ee ee
a CENTEr DUNCD ie Bak Rk oot ere noes Reno aoe eae
mC abinet COOL MOOK, (225. sid hottest
wonadesixtire 1Of NONTeVOLVING .ef1d y chose ee ee ee
AUCSITIMIEL, Age BR coerce Se eles We cc vis Bae ot ERY CT a eet ee
. Central insulation for spark plug....... Piekae Rata.
wiGarnage Bolts. 207 Wes ei, os em nes rane Pelee
SOTAPICU Gi SIAM RAST oe ric Me ee eee. Seren ee eee
Wai bodyor Spark pltig 4. isc. wee otters. Ge eke cee
: Catch for cabmet door hook: & eee 2s ee
. Jam nut or first nut for top of spark plug... 4 REARS
» Pott Soldering cAPee se ca cca oem ieee: TL ee re
« Coping-saw: bladel/./ccy Gon) oo ee i ee ee
BSW BCL OW or). ele aay Uk Una dans ee eel) pee dr SO
SUDSEE SCTE W ics cde eae nek mee Unt Al eRe) ce ee ee
Ry OtOVE: DOME 2 226 eV 2/2 ae at. cle ne ed
g shelf stop Or SUPPOLts = wow lp lce see wa een eee
ae Na chines boltei 2) on vce See Rgee hte ee ree ee
mi DILL DAS vate ie-'a eo sits a nies heii, cote ee ee
Number
Correct
47
46
46
45
44
44
42
42
41
A Description of the Tests 27
Number
Name of Device Correct
AieLemrent. Dox Or mitre frame fastener. coon as ks ae ae a 9
q. Expansion lug nut. Pe oe 9
yt. Wedge to prevent wihiow Ere Patines 7
ji. Pipe reducer bushing. . i
gt. Packing nut for spark ie ' 4
a. Bushing for packing nut of sore Bing 2
6. CONCLUSIONS
While the results of the Recognition Test are interesting froma
research point of view, they are of doubtful value in practical
educational testing work.
The experiment was largely to determine the practicability of
the method, and while there is no doubt but that there is a certain
value in this sort of a test, it has serious limitations, the most im-
mer nek {Tb
portant of which is that it does not give promise of measuring |
general mechanical ability of the kind in which we are most in-
terested, such, e.g., as is measured by the assembling tests. It is
purely a test of certain technical information and, moreover, it
seems very probable in the light of later results with picture tests
that this kind of measure can be obtained with infinitely less labor
and expense by the use of pictures,—and these can be increased in
range almost infinitely, which is not possible with actual objects.
The incidental educational value in the handling of actual me-
chanical objects, of course, is higher than that in looking at their
pictures, and for any purpose, misperceptions will be less frequent.
Actual objects also afford a better basis for what may be called
mechanical reasoning. But the range of objects is limited. It is
extremely difficult to cover a representative field without having
at the end an impossible collection of large and heavy objects,
impracticable to manage. Its usefulness is therefore largely con-
fined to the laboratory.
SECTION VIII
SINGLE MODEL SERIES
Experience with the Original Assembling Series I and II showed
that such sets must be made more convenient and more workable,
if possible. It was accordingly decided to attempt the develop-
ment of a series that would eliminate as many as possible of the
faults of the former sets and add possible improvements. The
faults were in the main as follows:
The Original Series I was too easy, being adapted only for the
lower grades, and was exclusively a copying test. There was no
way to insure beginning with the easier models and progressing
toward the more difficult ones, as all parts were mixed in one large
compartment. Moreover, the boxes were of an awkward shape
to handle, and being made of cardboard were not sufficiently
strong. : ‘
The models of the Original Series II required an average of from
IO to 20 minutes each for most persons. Thus in one hour less
than six models could be tried. The element of luck entered into
this arrangement, and it is particularly difficult to give just and
proper credit for a few partially finished difficult models. The °
sets were also cumbersome to handle and the models difficult to
disassemble. The boxes as designed were about 8 by 43 by 20
inches. ~
I. SINGLE SERIES I
x
After much search and experimentation ten models were se-
lected,—each one simple enough to be solved by an average 7th
grade boy in approximately 3 minutes. From the Original Series
I those models which had proved most satisfactory were taken,
and these supplemented by others, better chosen in the light of
past experience. A smaller, narrower, though longer box was
next designed,—a group of eight of which when strapped up for
carrying are not materially larger or harder to handle than a suit
case. In selecting models all that were not ‘‘fool proof,’’ easily
scorable, and easy to disassemble were rejected. It was also
28
A Description of the Tests » 29
decided to eliminate the extra assembled ‘‘copy”’ model in each
case for the reason that even simple objects would then im-
mediately become sufficiently difficult to constitute a test.
Moreover, it eliminated mere ‘“‘copying”’ and introduced what
was believed to be a somewhat “‘deeper’”’ sampling of the kind of
ability it was desired to measure. It also cut the cost nearly in
_ half, made the test only half as heavy, and easier in every way to
manage. The idea of using the cover as a tray was retained, but
all tools except one small screw driver were eliminated. These
purely physical features may seem irrelevant and unimportant,
but after a considerable experience with this type of tests it seems
clear that if any such test is not perfected mechanically so that it
is easily workable by any competent examiner,—and is also eco-
nomical of time in scoring,—it defeats its usefulness and is, for
practical purposes, valueless. .
The chief improvement hoped for in Single Series I, however,
was increased measuring power, through a wider range of samples,
better control of conditions, and the elimination of copying. The
reduction of time to 30 minutes, as against 50 to 90 minutes in the
Original Series II was also important since it made it possible to
give the test conveniently within an ordinary school period.
2. MODELS INCLUDED IN SINGLE SERIES I
The models selected were as follows:
/
Ordinary cupboard catch
Six links of safety chain
Three-piece Hunt paper clip
Bicycle bell
Wire bottle stopper
Clothes pin
Shut-off for rubber tubing
Push button
Small rim lock
Mouse trap
SOMO O WD
The general method of scoring previously adopted was retained,
in which each model perfectly assembled was scored 10 points, and
partial scores assigned each model according to an arbitrary
schedule of values ranging from I to 9.
smerce
30 Measurements of Mechanical Ability
Thirty minutes was found to be sufficient for at least 80 per cent
of 6th grade children, and was adapted as standard. A perfect
score in 30 minutes was accordingly scored 100 points. In addi-
tion a speed bonus of one-half point for each minute under 30
which was not used was added. (This, however, occurs but
seldom.) Fig. 4a shows this series in its final form after the
models had been scaled.
The instructions which are printed on the cover of each box are
as follows:
DO NOT) OPENRIUHIS BOX sUNTILATORD
ODO RSG
Directions
In this box there are some common mechanical things that have all
been taken apart. You are to take the parts and put them together
as they ought to be; that is, you are to take the parts and put them
together so that each thing will work perfectly.
Do not copy what your neighbor is doing but work absolutely by
yourself. Turn the box so that the hinges are towards you. When
opened in this position the cover forms a tray in which to work.
Do not break the parts. Everything goes together easily if you do
it in the right way. Begin with Model A; then take Model B; then
C;andsoon. If you come to one that you cannot do in about 3 min-
utes, go on to the next. The person who gets the most things right
gets the highest score.
Preliminary Trials with the Single Model Series. Experience
with this series quickly demonstrated it to be an improvement
over the earlier ones. The extended range of models, each of
which can be solved in a comparatively short interval of time
(averaging from I to 5 minutes) was found to offer a better chance
for mechanical ability to show itself. It afforded a better sam-
pling of a pupil’s ability since he had ten chances instead of four or
less (as was the case with the Original Series II) in a period of 30
minutes.
The advisability of continuing the “single’’ model idea, that is,
the eliminating of the extra assembled copy model was considered
both on the basis of the administrative advantage, and on the
basis of the resulting efficiency of the test. In order to test the
“ULIO
EEG
why
Ts
IMIS
OSU
“DV. “OI
+ / ‘ Sy
Vw 44 ul re
FAR 5 7
A Description of the Tests 31
latter point a group of 62 pupils were given a special test as
follows:
From the twenty models later available, ten, which were of such
a nature as to lend themselves advantageously to being dis-
assembled as well as assembled by the pupils, were made up into a
first series, called the ‘‘disassembling-assembling criterion set.”
Here the pupils were first permitted to take apart each model,
and, after this operation had been scored and boxes inspected, the
pupils were immediately required to assemble the models which
they had previously disassembled.
This probably constitutes a more thorough test than either the
assembling with, or without a copy model alone, but is of course
much more laborious and costly in time. A single series of dif-
ferent models was then given the same pupils, to afford an op-
portunity for comparison. The correlation between these two
tests was estimated from these results to be between .6 and .7,
indicating a fairly high correspondence. In order to afford an-
other check, shop teachers’ ranks were obtained for the groups in-
cluded. Fortunately, it was possible to obtain the independent
rankings of two such shop teachers, the intercorrelations of which
averaged .g1, justifying considerable reliance in these ranks as
criteria by which to judge a test. The correlations between the
shop rank and each of the tests was then computed. Between
disassembling-assembling and shop rank, r= .58+.06, and between
assembling only and shopwork, r=.61+.06, indicating that the
single series probably is at least equally as good a measure as the
disassembling-assembling series. More experimentation should,
of course, be carried on to establish more precisely these points,
but it was not practicable in this instance. From the administra-
tive standpoint the single model series are in every way advanta-
geous,—unless it be that the opportunity for cheating is somewhat
greater. But by ordinary precautions this factor was easily
controlled.
On the whole, therefore, it seemed justifiable to continue the
further development of the single model series.
3- SINGLE SERIES II
Following out the success with Single Series I, the next
task undertaken was accordingly to form: a second similar set
supposedly about parallel in difficulty with Single Series I.
32 Measurements of Mechanical Abithty
This was called Single Series II. Here the attempt was again
made to select only models which in the light of past experience
seemed thoroughly practicable for this purpose. This means
they must be sufficiently difficult to present a real problem, and
yet be workable. They must be of such a size and nature as to fit
conveniently into a series,—must not demand too much mere
physical strength, nor special assembling tools, must represent
considerable variety, and must correlate fairly well with the same
criterion. | Particularly only those which can be very quickly dis-
assembled should be included. In the preparation of this series
one further step was taken than before in the search for suitable
models. Certain stock commercial articles were partially remade
in such a way that they can with this modification be disassembled
and assembled; for example, a rivet may be replaced by a re-
movable pin without destroying the identity and essential char-
acteristics of the article. A screw may replace a rivet in the same
way. This makes available many more models.
As previously pointed out, one of the difficulties met in employ-
ing physical objects of this kind as test material, as opposed to
printed problems which can be produced at minimum cost, modi-
fied ad infinitum, and reproduced at will, is that the former are
lacking in just these characteristics. Thus, while the models
selected seem commonplace when found, the task of finding ob-
jects that will meet all requirements is considerable. A trouble-
some point has been met repeatedly in the fact that articles of this -
character are continually disappearing from the market, so that it
frequently happens, after a model has been standardized, that it is
‘ unprocurable except at the exorbitant expense of buying new dies
“6
or patterns, for ‘“‘making it up special.’”’ The most practical
method of overcoming this circumstance has been to continually
standardize new models in terms of old ones, so that a com-
paratively large number of known difficulty are available. In ad-
/dition a practice has been made of selecting fairly staple articles.
One reason each model must be standardized individually is to af-
ford units or models of known difficulty to be used as substitutes
for articles unprocurable after they have been standardized.
This introduces difficulties, but cannot well be avoided.
Fig. 4b shows general appearance of Single Series II in its final
form.
‘WAOY [eULT ‘[] Seles asurg “qh ‘Oly
A Description of the Tests 33
4. MODELS INCLUDED IN SINGLE SERIES II
The list of models as first tried out was as follows:
A. Elbow catch F. Calipers
B. Rope coupling G. Rubber stopper
C. Toy pistol H. Four-piece paper clip
D. Expansion nut ' JT. Double acting hinge
E. Sash fastener J. Lock
Preliminary Trial of Single Series II with Single Series I.
Preliminary trials of this series indicated the models all to be
serviceable. Preliminary scaling indicated also that they were of
a slightly better ‘‘spread”’ or distribution as regards relative dif-
ficulty. The two series were now given to some 300 pupils and on
the basis of these data the further refinement of the material was
undertaken. Asa preliminary it was thought advisable to check
up the question of the contributory value of each new model.
5. CORRELATION OF EACH OF 20 MODELS WITH CRITERION
The criterion here adopted was the total raw score in 20 models.
With this each model was correlated with results as shown below
for 50 thirteen-year-old boys.
First SINGLE SERIES I
A. aden CALCH aL. cake tes ores. Bes Clothes, Tith.:.janind ae eee .68
B. Chain. Seu eat emo Ram aa, (ECODEr NOSE yo). ie) tn eee AS
C. Hunt paper ine ei Fee ee Ry ee CE Ee Tiett: LILCONLs a tira cae ee ah SE
Dithirycie pel... seo. ee HON be SOCK ONG, 1a. oy vaeme Wee en ee AS
E. Exp. rubber stopper....... TOU tie VWVITEIBLODDCE by is cau cet set he
FirsT SINGLE SERIES II
Fit RADOW CALEY ox nc! aia att Ce PD) EA Os 75 «USDA IMC on i Eeiraba (21 ig, W
By ROPE COUDUINS ss. oi bes OO MILA RA OSCE cure lt Crime Gah 7
C. Pistol. be elo ya Ue inane de TOM Sepa ah ai A ats | .68
D. Bansrinien ant Mea Ahh ta Gee .64 I. Doublehinge. . 32 Sage ea
i! Sash fasteners. Aewis,) tds CePA Zs SGN C2 Br), a Nea eee .48
While it might be theoretically desirable to retain only models
correlating very high with this criterion, the practical considera-
tion of the difficulty of obtaining suitable models made it seem
advisable not to discard any model which had been found to work
well in the series. Moreover, a low correlation with this criterion
is no evidence for assuming a low correlation Pe other equally
valid criteria.
SECTION IX
SCALING
As in the case of Series I,! arbitrary, partial and perfect score
values were assigned in Series II? for various degrees of excellence
in attempted solutions of each model. Each model correct was
counted 10 points, as before. Thus with the models roughly in
order of difficulty within each set, and with these partial score
values, a working method of scoring each individual was es-
tablished. But at best this procedure is crude. The difficulty-
distances between models are by this method unknown,—that is,
the exact difficulty of each model, as compared with any other, is
undetermined, and no account is taken of the form of distribution.
However, in dealing with this special type of problem a large part
of the task consists in the experimentation necessary to discover
and perfect models, as well as in the special technique involved in
managing them. A series of mechanical objects highly perfected,
in so far as finesse in scaling and theoretical treatment is con-
cerned, might still be impracticable and largely useless for actual
work. But having previously taken up these points, and having
selected material so as to meet these requirements, the next
logical step is the refinement of the mathematical technique.
I. A NEW METHOD OF SCALING: THE MCCALL METHOD
In the matter of scaling each individual model to determine its
relative difficulty, and in the scaling of each series as a whole, a
number of methods were possible. The theory of scaling material
of this type is not different from verbal material, except for pecul-
iar items such as the short series of problems necessitated by
physical limitations. But these are incidental. The literature
of test making contains abundant examples of ways of scaling.
In fact, it is the variation in methods and technique that is now
most disconcerting, for since much of the procedure is arbitrary
1 See sample score sheets in Appendix.
* For sake of brevity, and since previous series have been discontinued, the term
‘“‘Single’’ series will henceforth be dropped, all series being single unless otherwise
specified.
34
A Description of the Tests 35
it becomes more and more confusing as each scale comes out,
based on some new modification in procedure. Fortunately, at
the time of this research a growing movement, fostered by Pro-
fessors Thorndike, McCall, and others, has developed for the
standardization of technique in the scaling process. Even >
though that standardization be based largely on mutual agree-
ment to adhere to an arbitrary procedure, the important thing is
the agreement on some one definite method.
In the interest of uniformity, therefore, as well as on the basis
of the advantages incident to it, the McCall method of scaling has
been adopted.!
Advantages of the Method. As has been suggested, the chief
advantage lies in the direction of adopting uniformity of method,
making possible direct comparison of final scores for tests of vari-
ous abilities. Just as a series of Fahrenheit thermometers used
respectively for measuring the temperature of one’s bath, blood,
room, automobile radiator, baby’s milk, etc., etc., will record the
final results in comparable and meaningful units ¢(which-we-call
‘““degrees’’), just so it should be possible to compare units of any
number or variety of mental abilities.
Adopting a uniform procedure involves at least three important
items:
1. The agreement as to a basis for scaling, that is, what grade or age
should be used in determining scale values.
2. The agreement as to a common unit.
3. The agreement as to a uniform zero point, or point of reference.
Scales have in the past been constructed on the basis of this
grade or that, or on the basis of several grades combined. Units
have been of all kinds,—the number of right answers, per cent
right, probable errors or standard deviations of various grades and
ages. Zero points have been located at practically as many dif-
ferent points as there are scales.
Professor McCall’s method proposes to standardize these points
as follows:
a. The basis of scaling adopted by mutual agreement by a number of
investigators is the total distribution of children whose ages range from 12:0
to 13:0 years—no matter in what grades found. The reason for the choice
1Wm. A. McCall, How to Measure in Education, Macmillan Company. Also
Teachers College Record, March, 1921.
36 Measurements of Mechanical Ability
of 12-year-olds in preference to others is that it has been found through
researches by Thorndike, Kelley, and others, that with this group a more
normal distribution is found than for any other age, since this group is least
affected by the factors of school elimination.
b. The standard unit adopted is one tenth of zr S.D. of the 12-year-old
distribution, which unit McCall proposes to call ‘“‘'T”’ in honor of Professors
Thorndike and Terman, early advocates of some such standard practice.
c. The standard point of reference is to be the mean 12-year-old, with
the zero point arbitrarily (but apparently reasonably) located at 5 S.D.
below the mean. Scale values thus defined will henceforth in this report
be referred to as “T-Scale”’ values.
d. Each test scaled as a whole. The important departure in this method
is that the test is scaled as a whole. Each possible “‘number right”’ on the
whole test—no matter which elements are included—is given a difficulty
value, first in terms of “per cent of 12-year-olds who exceeded plus half
those who reached”’ that partial, and then, to take account of the form of
distribution, this percentage is converted into the corresponding S.D.
value of 12-year-old by means of a table. Sucha table appears on page 44.
The two extremes of this table, it will be noted, represent such minute
percentages that in practice the ends of the scale are never actually
reached. The table will, of course, be recognized as a representation in
round numbers of the normal surface of frequency, whose two extremes are
infinite, but are here arbitrarily placed at —5 $.D. and +5 S.D. For
most scales the table range will lie between, say, 15 to 20 and 75 to 80, and
this is a sufficiently large range to provide adequate differentiation.
McCall has thus adopted the methods employed by Bucking-
ham, Trabue, Woody, and others, for determining the difficulty
of each scale element, to the determination of the difficulty of each
possible percentage of right answers for the test asa whole. This
ignores the relative difficulty of each individual element as
stressed by previous scale makers,—except for the general recom-
mendation, advising placing the elements in the general order of
difficulty for all grades to be tested, to best insure that the pupil
will attempt all the problems which he has any chance of solving.
The method takes advantage of the fact that because: a given
element is most difficult for the greatest per cent of pupils in
general, there is no certainty that it will be most difficult for any
particular pupil. Some other element may for him be the most
difficult. 7 |
The method avoids the more or less precarious and especially
laborious procedure of measuring inter-grade distances which is
based on assumptions which have never been adequately sus-
tained. It is also much simpler than the former methods, and
A Description of the Tests 37
makes it possible to standardize easily many achievement tests in
terms of T-Scale values. It avoids the other laborious and
somewhat involved 20-80 per cent method used by Thorndike, in
scoring the Alpha Reading test or the 50—50 per cent method used
by Kelley in the scoring of the Kelley-Trabue Completion Ex-
ercises.
2. RELATIVE DIFFICULTY OF EACH MODEL
The next task would then logically be to determine the T-Scale
values of each possible number right for Series I and for Series IT.
Before doing this, however, it is necessary to examine more closely
whether the order in which the models were at first placed in each
test is in accordance with their real difficulties. To enable us to
observe this point the percentage of correct answers for each
model for grades 6, 7 and 8 were computed with their S.D.
equivalents. For convenience all the models, that is, both Series
I and Series II, were thrown together and all the results tabulated
in Table VII. A glance at this table shows at once that the most
striking fact is the similarity of difficulties for all of the 20 models,
for any given grade, or on the average for all the grades. It
means that the 20 models,—selected on the basis of personal esti-
mate as being of a variety of difficulties, are really not very dif-
ferent,—the total range of either series being (on the basis of the
average difficulty for the three grades) only about 2S.D. Fig. 5,
showing this fact, also shows that there are ‘gaps and bunchings’”’
of the models of each series, with Series II a little more difficult on
the whole. Theoretically, it is desirable to have a larger range in
scale values, but in this case we must keep in mind that there are
but ten elements, and to spread ten problems out over a long
range of, say, 4 to 6 S.D., results in a very ‘‘thin”’ scale, with
great unreliability at any one point of the scale.
There is, therefore, a justification for accepting the series as
. they are, rather than beginning again and substituting, say, three
models much easier and three much harder than any at present
included, to produce a larger range of difficulties. Ten scale
elements grouped fairly close together tend to eliminate mere
“luck’’ scores, since the opportunity is provided to try more than
once, at about the same difficulty. So long as the number of zero
or perfect scores for the whole test is negligible or small it is likely
that the final score is more reliable when based on such a group of
38 Measurements of Mechanical Ability
models than it would be in the proposed long and thin scale.
While the range of difficulties is short, the ten tasks are by no
means identical in difficulty, and less so in their nature. We
might actually have a scale of ten elements of identical difficulty
and identical nature and yet obtain a measure by considering the
speed score. This, of course, is not our purpose here, although
account is taken of the speed, and hence the score is partially
in terms of it. The differing nature of each model makes it
TABLE. VIE
PERCENTAGE OF RIGHT SCORES FOR EACH MODEL wITH S.D. EQUIVALENTS
Zero=—5 S.D. N=Sertes I: 452, Series II, 459
8th Grade 7th Grade 6th Grade
pela Ah IE PBS aa a Aver
Model Per Per Per Se
5.D. 5.D. Cent 5.D. Equiv
ee ee Eee AS Re
Cupboard catch .| 665 457 714 443 560 485 | 462
Catia ye.) 286 557 220 wer 203 583 572
Hunt paperclip..| 340 541 300 553 252 567 554
Bicycle bell.....| 243 570 || 9422 577 185 590 | 579
Expansion rubber
Stopper: hey wich 182 591 134 611 147 605 602
Clothes pin..... 445 514 | 464 509 318 547 523
Rubber hose....}| 231 574 249 568 34 611 590
Push button ....| 206 582 131 612 096 631 608
Lock Nati... 142 607 114 621 062 654 | 627
Wire stopper....| 231 574 168 596 086 637 602
Elbow catch ....| 525 494 | 562 484 | 380 531 503
Rope coupling...| 695 449 592 477 202 583 503
DIStolee sche 61 472 615 471 386 529 | 491
Expansion nut ..| 51 497 | 562 484 | 228 574 | 518
Sash fastener....| 251 567 266 563 189 588 573
Gahiperso. i. Pte ly, 128 614 146 605 ae: 632 610
RP TAD eo ae yams 105 625 115 620 028 690 645
Paper clip No. 4] 146 605 115 620 | 050 665 | 630
Double hinge....| 073 646 | 094 632 027 693 657
BOEKING! 20 h7F 1050 665 023 700 a: 741 683
Totalace vor 11,201 EE,323 12,136
AVETAIE) 201 Vane 560 5 ack 566 ome 607
A Description of the Tests 39
particularly hazardous to attempt to say that the mechanical
ability of a certain boy is, say, 30 in Series I, because he can
assemble the cupboard catch, clothes pin and Hunt paper clip,
but not the other models. In Series II he may score 60 because
of special experience, and the accidental nature of the particular
objects included. It very frequently happens that three difficult
(as by this determination) models are solved and many easier
ones (as by this determination) are not solved. This, of course,
occurs in other scales as well, such as reading scales and language
scales, but not so frequently because there is greater uniformity
and continuity in the nature of the scale elements. It was partly
to provide some statistical method of interpreting such scores that
the 20-80 per cent and 50—50 per cent methods previously referred
to were devised, and partly to provide a simpler method for ac-
SERIES I SERIES I1
n * 462 n #459
Lock #2 }
;
Trap Double Act. Hinge 4 \
Lock #1
Push Button
Wire Stopper
Shut-off
Bicycle Bell
Chain
Defiance Paper Clip
Calipers
Exp. Rubber Stopper
Sash Fastener
ak
Exp. Nut
Hunt Paper Clip
Clothes Pin
Rope Coupling & Elbow Catch
Pistol
Cupboard Catch
i
V
Fic. 5. Scale Difficulty Distribution of Models for Series I and Series II.
Av. S.D. Difficulty Values for Each Model for Grades 6, 7 and 8.
4
40 Measurements of Mechanical Ability
complishing the same purpose that the McCall method was
proposed.
Before going further into this, however, the matter of the order
of the models within both series should be settled. Fig. 5 shows
that the order of difficulties is not the same as that determined in
the beginning by the preliminary trial with a few cases. On the
other hand, the differences are not very great.
3. OLD ORDER AND FINAL ORDER OF MODELS
Following is the old order again repeated with the final order
for both series:
SERIES |
OLD ORDER FINAL ORDER
A. Cupboard catch Cupboard catch ™
B. Chain : Clothes pin «
C. Hunt paper clip Hunt paper clip ”
D. Bicycle bell Chain-{
E. Wire bottle stopper » Bicycle bell \
F. Clothes pin | Shut-off ,
G. Shut-off | Wire stopper
H. Push button” Push button 4}
I. Lock No.1 Lock No. 1 }
de LraO ae Trap
SERIES II
OLD ORDER FINAL ORDER
A. Elbow catch Pistol
B. Rope coupling Elbow catch ,
C. Pistol Rope coupling
D. Expansion nut Expansion nut
E. Sash fastener Sash fastener
F. Calipers Expansion rubber stopper
G. Expansion rubber stopper Calipers
H. Defiance paper clip Defiance paper clip
I. Double action hinge Double action hinge
J VLOCK No, aan Lock No. 2
It will be noted that the shift in position is slight in terms of
scale distances, as shown in Fig. 5. The question now comes up
whether to leave each test as it was originally, in order to preserve
its identity, which is desirable in the McCall method of scaling,—
or to rearrange the models in terms of the final values obtained.
It seemed best to do the latter. Shifting the position of scale
A Description of the Tests 41
elements, however, introduces an error in that the difficulties
have a tendency to change when placed in a different position on
the scale. But the changes here made are so slight that it is
believed no serious change in difficulties will result.
In comparing the two scales in Fig. 5, it is clear that the spacing
of both the scales would be improved by shifting models from one
series to the other, and this could be done since all twenty models
were given to the same pupils. But there is an objection to
destroying the identity of Series I in that all other records ob-
tained with it then would be lost. The chief body of data col-
lected with this series was that obtained in the Army, where
14,000 cases were tested. This seems sufficiently valuable to
justify preserving the identity of Series I, and doing so automati-
cally preserves that of Series IT.
4. DIFFICULTIES IN OBTAINING CERTAIN MODELS
In this connection an unfortunate circumstance, illustrating the
‘annoyances incident to working with this type of material, may
here be considered. After all the records of the Army experiments —
were completed for the 14,000 cases, with Single Series I, and the
task of scaling and establishing norms for age and grade was taken
up, it was discovered that two of the models used in that series
were unprocurable because they have been discontinued by the
manufacturers. The two articles in question were (1) a small
bicycle wrench and (2) a coin safe for holding pennies, nickels
and dimes. It was therefore impossible to preserve the exact
identity of the series used in the Army, and the only possible
alternative was the substitution of other models. Accordingly,
this was done. For the bicycle wrench, which was Model A of
the Army series, the cupboard catch of our present series was
substituted, and for the coin safe, Model E, the wire bottle stop-
per, as of probably similar difficulties.
These substitutions must therefore be kept in mind when
considering the Army series. In order to evaluate them the dif-
ficulties were carefully compared. From data in hand the fol-
lowing comparisons were made. For a group of 7th and 8th
grades (supplemented by adults, as shown) the difficulty values
of the discarded and of the new models were found to be as
follows:
42 Measurements of Mechanical Ability
OLD (DISCARDED) MODELS
A. Bicycle Wrench
No. Group Per Cent Right S.D. Equivalent
95. 7th and 8th Grade Boys ...... .516
220, ,, ooldiersiy (jae tr ee eke .525
A Verace saa Ghats okt 5200 Cah RESUS
B. Coin Safe
95 7th and 8th Grade Boys ...... .408
220) Soldiers faa. gab es ibd a ea! .440
LS” Con, Rea ee 42d) osc a ae
ge s
2” Average Difficulty of Old Models... .. . . 50.0
ef New Mopets
- A. Cupboard Catch
544 7thand 8th Grade Boys ....... .689 .450
‘ B. Wire Bottle Stopper
544 7thand 8thGrade Boys....... . 196 586
Average Difficulty of New Models............... 51.8
Difference int Diflicult ye. ey iets ot ec LO-LOte tao)
Thus it is seen that the average difficulty of the two new
models exceeds that of the old discarded ones by .18 S.D., or 1.8
points on the T-Scale. From Army scores obtained with the
series, including these two easier models, this amount should be
subtracted to make them comparable with the scores herein re-
ported, which were obtained in the final series. This correction is
of course only the most probable one. To substitute one model
for another without altering the scale values as a whole would
require perfect correlation and identical difficulties. All we know
here is that the difficulties are reasonably equivalent (we have the
estimated differences). The correlation of the four models in
question with the total score was found to be for fifty cases as
follows:
Wrench with Total Score, 10 models..............7=.5I
Coin Safe with Total Score, 10 models............r=.49
Cupboard Catch with Total Score, 20 models......r= .67
Wire bottle Stopper with Total Score, 20 models . ..r= .48
A Description of the Tests 43
5. T-SCALE VALUES FOR EACH RAW SCORE
OF SERIES I AND SERIES II
Having determined the scale difficulties of the elements of these
two tests, and having arranged them in what seems to be the best
order, we may now consider the matter of scaling each instrument
asa whole. This is done by calculating from the distribution of
the 12-year-olds the percentages exceeding plus half those reach-
ing each possible raw score value, and then converting these per-
centages into T-Scale equivalents, in the same way that elements
of scales have been treated by other investigators.!
The distribution of scores for the two tests as rearranged and
scaled is given in Tables X and X. Because of the small number
of cases of 12-year-olds, it was decided to utilize as a check upon
them the scores of ages 13, 14 and 15. By computing the dis-
tances between the median of the 12-year-olds and that of the
13-year-olds in terms of the percentage of one group which reaches
or exceeds the median of the other group, and transmitting this
into an S.D. equivalent, and then correcting all of the 13-year-old
values by this amount, the 13-year-olds may be utilized as 12-
year-olds. This of course assumes a normal distribution for all
age-groups thus utilized. Ordinarily it is inadvisable to thus
make use of neighboring age-groups, especially those more than
one year removed from the 12-year-olds. In this case, however,
no marked differences are discernible in the form of distribution
for ages 13, 14 and 15, and since the number of cases is small it
was thought best to utilize all of the data.
The exact method followed in Tables IX and X is as follows:
The S.D. scale values, with —5 S.D. as a zero point, were de-
termined for each age group exactly as for the 12-year-old group.
The distances between the 12-year-old group median and the
median of each other age group were then calculated by the per-
centage of overlapping method. Thus the percentage of 13-year-
olds who fell below, plus one-half those at the median of the 12-
year-olds, was found to be for Series I, . 26. Reference to Table
VIII shows the nearest S.D. equivalent in round numbers to be
56.55.D. Subtracting this from 50, the midpoint of the 12-year-
olds, gives a difference of 6.5 T. That is, the difficulties of at-
taining each of the various numbers of models right for the 13-
* Buckingham, Trabue, and others.
44 Measurements of Mechanical A buity
year-olds is on the average 6.5 T less than for the 12-year-olds.
Similar differences have been computed for each age-group and
utilized as a ‘‘correction.’’ Adding this correction to the S.D.
values of each age group we obtain the 12-year-old equivalents.
That is, the older groups are thus utilized as 12-year-olds in order
to increase the reliability of our data. By taking the averages of
TABLE VIII
S.D. DISTANCE OF A GIVEN PER CENT ABOVE ZERO; EACH S.D. VALUE
Is MULTIPLIED BY 10 TO ELIMINATE DECIMALS
The Zero Point Is 5 S.D. Below the Mean
S.D. $.D S.D S.D
Value Per Cent Malge. 4 hon GeDt | ervatyen [ener Cent Value Per Cent
oO. 99.999971 25. 99.38 50. 50.00 75. 0.62
0.5 99 .999963 25.5 99.29 50.5 48.01 75-5 0.54
TE 99.999952 26. 99.18 flee 46.02 76 0.47
1.5 99 .999938 26.5 99.06 51.5 44.04 76.5 0.40
2. 99 .99992 27. 98.93 52. 42.07 77 0.35
2.5 99 .99990 27.5 98.78 52.5 40.13 Lie 0.30
Zi. 99.99987 28. 98.61 te 38.21 78 0.26
3:5 99.99983 28.5 98.42 53.5 36.32 78.5 0.22
4. 99 .99979 29. 98.21 54. 34.40 79 0.19
4.5 99 .99973 29.5 97.98 54-5 32.64 79.5 0.16
5. 99 .99966 30. 97.72 55. 30.85 80 0.13
5.5 99 .99957 30.5 07-44 55.5 29.12 80.5 oO.11
6. 99 .99946 31. 97.13 56. 27.43 81 0.097
6.5 99 .99932 83045 96.78 56.5 25.78 81.5 0.082
P(e 99.999I5 32. 96.41 tty 24.20 82 0.069
7-5 99.9989 32.5 95-99 S75 22.66 82.5 0.058
8. 99.9987 33. 95.54 58. 21.19 83 0.048
8.5 99.9983 33-5 95.05 58.5 19.77 83.5 0.040
9. 99.9979 34. 04.52 59. 18.41 84 0.034
9.5 99.9974 34.5 93.94 59.5 17.11 84.5 0.028
Io. 99.9968 35. 93.32 60. 15.87 85 0.023
10.5 99.9961 ake 92.65 60.5 14.69 85.5 0.019
Poe 99.9952 36. 91.92 OI. T3a57 86 0.016
Ers'5 99.9941 BGa5 QOI.15 61.5 I2.51 86.5 0.013
I2. 99.9928 sti ie 90.32 62. II.51 87 O.OII
r2.5 99.9912 37 25 890.44 62.5 10.56 87.5 0.009
DSi 99.989 38. 88.49 63. 9.68 88 0.007
L325 99.987 38.5 87.49 63.5 8.85 88.5 0.0059
I4. 99.984 39. 86.43 64. 8.08 89. 0.0048
14.5 99.981 39.5 85.31 64.5 7-35 89.5 0.0039
15. 909.977 40. 84.13 OB 6.68 90 0.0032
I5.5 99.972 40.5 82.89 65.5 6.06 90.5 0.0026
16. 99.966 41. 81.59 66. 5.48 QI 0.0021
16.5 99.960 41.5 80.23 66.5 4.95 91.5 0.0017
Lig 99.952 42. 78.81 OF 4.46 92 0.0013
17.5 99.943 42.5 77.34 67.5 4.01 92.5 0.O0II
18. 99.931 43 75.80 68. 3.59 93 0.0009
18.5 99.918 43-5 74.22 68.5 3.28 93-5 0.0007
IQ. 99.903 44 HLA 69. 2.87 94. 0.0005
19.5 99.886 44.5 70.88 69.5 BAT) 94.5 0.00043
20. 09.865 45 69.15 70. 2.28 95. 0.00034
20.5 99.84 45.5 67.36 7025 2.02 95.5 0.00027
Gh Os 99.81 46 65.54 7 Ree I.79 96 0.00021
21.5 99.78 40.5 63.68 7 feel Ix5S 96.5 0.00017
22% 99.74 47 61.79 yee I.39 07 0.00013
2275 99.70 47.5 59.87 72.15 L422 97.5 0.00010
234 99.65 48 57-93 73. I.07 98 0.0008
23.5 99.60 48.5 55.96 7325 0.94 98.5 0.000062
24. 99.53 49 53.98 TAs 0.82 99. 0.000048
24-5 99.46 49.5 51.99 74-5 0.71 99.5 0.000037
Brapanel SAEED Gatet eke Peak Nat MMT Ane Secs. 2) | Uae T OR ets RS Aan Cad CVn 100 0.000029
A Description of the Tests 45
TABLE IX
ASSEMBLING TEST—SERIES I
T-SCALE SCORES FOR EACH NUMBER RIGHT. WITH PERCENTAGE OF EACH
AGE Group WHO REACH OR EXCEED EACH SCORE
Total Number of Cases—1,361
Ageiz2 | Ager3 | Agerq | Age 15 Pkeayy
No. of Problems T-Scale
Right X10 Score Per Cent | Per Cent | Per Cent | Per Cent | Per Cent
Exceeding | Exceeding | Exceeding | Exceeding | Exceeding
+ One-half} + One-half} + One-half} + One-half} + One-half
Reaching | Reaching | Reaching | Reaching | Reaching
TCO ES ere ae hee 24 99.6 fe) re) 00 100
REO Sirk late he eters 30 98.7 r¢) 99.8 r¢) 99
ASUOs Sihcisiaie ts oes 31 97.3 99.6 06.8 te) 99
ORTON Pace oe ieee a O50 i 98.8 95.6 99.5 99
STO 30 ae iiges s copes 35 91.9 97.6 94.8 99.5 99
TOSCOAT Ec iron nia 38 88.8 94.0 93.6 07.4 98
ESOC DS i wee, eens 40 85.2 90.8 OI v2 94.1 97
Tt. COu Loe tate. es 42 81.6 88.7 87.9 Oras 96
50 (GL EFE as cee 43 78.9 85.9 85.1 890.8 95
TS tOutOr. fois. sh 44 7Sind 82.3 82.3 88.2 95
20 tOPSIe so eee 45 Ghee. LIAS 80.3 86.6 94
SCO; BSE eerme ene 46 O72 77.9 78.3 85.5 93
BAy CONES e sive ee woe el 47 G25 73.8 75.8 83.4 92
20 COUA7R ee ote 48 58.1 TA ane 81.2 QI
28; COndOU a tes 49 55.0 69.0 67.8 Out 90
SO tO8STe. ee 50 49.6 62.9 65.4 95.9 88
PICO eae re eae 51 43.3 57.7 63.7 ays 87
SALONS S i Oe ea ee 52 sO 54.9 60.6 68.5 87
SGRCOT SUE Was eo acted ig 32.9 51.6 56.9 65.6 87
SERLO” 40%": antas asians 54 28.4 48.0 54.9 63.0 87
AONUCG) Alas cs ele iter 55 23.9 45.6 51.6 61.3 78
AP AOCAS Re. os fe 8 56 20.8 43.2 40.8 58.1 75
AATUG AS es sd t= CO Py | 18.5 38.7 42.0 54.4 72
AOLCOBA GT Oe 5 acces 58 16.7 Baar 37.9 52.2 69
BSLLOMAQ) ca ar- 5 cs 4.54): 59 14.5 28.7 35.9 Raghie 66
SOSLOES Cte cs actos 60 ba ae 25.4 33.9 47.9 63
Ere Wh te eras epee ae 60 9.0 22.6 STEE 44.7 60
RANCORG Sere eae. ca cnths 61 9.0 19.8 20515 42.0 57
BROMO Sif te mae ete aie a 62 742 Wee 28.3 42.0 54
SSutGeSOu cee wrcite my: 62 7 hor: TOMs 2erh 38.2 51
GO tOLOTE: cece kus: 63 Eee. 14.5 22.6 35.0 47
OP ICO OS ate a eine 64 SA I2.9 22.6 31.8 44
OARtOROS cer. oe cee: 64 5.4 Lies 19.0 26.9 4I
OGGEOL0 7. ore ee: 65 sat 9.7 19.0 22.6 38
OSi:t0 CO 9 sn 2. eyes 66 4.1 8.5 16.2 20.5 35
A Yast 0 tr ty Gey 3 oe ee et a 67 ee | 7 hak TQ7 E723 31
721 OR 73h ae Cols 68 232 HaR 1H ONY 13.5 28
SA LOFFS 2 ae re 68 ye | Oat TOnr 10.8 25
AO SCOm 97) eta. cena 69 Baw 6.1 8.5 S7, 22
9S 5tO. 70. aot eee 70 2234 Sas ery Sea 19
SO. t0uST2 2. renee 72 r34 5.3 257 6.0 7
82.:£0' 83's eis Sek oie 74 A Chay 2.9 3.8 14
S43t0/ 85556 - stad 74 Pan a7 2.0 3.8 II
BOOB 7s dee ee ee 75 2.4 Le 3.8 9
BS tO S03 4 on8 sei ce 75 AY} Tia Bind 7
DOLCOL OL keisha aan? 79 8 ree ne 6
MEZA AiG les i a ciate 80 8 afl Ue 5
SP ETO INR AS & Ae tele 80 fies 4
OOLEORO Fi. «chide aac 81 3
GOEL OO saan ec. 81 2
cou tele toe CA sae Renae Rae 82 I
EGAtOP LOS 25s + wes 82 snete
PO ALT HIOS oy aes esis 83
PAMINGOLLOT 5 Winis es 4). 83
TOG7107100 «25h. 50's 84
IN a VEN RG
46 Measurements of Mechanical Ability
TABLE X
ASSEMBLING TEST—SERIEsS II
T-ScALE SCORES FOR EACH NUMBER RIGHT, WITH PERCENTAGE OF EACH
AGE Grouprp WHO REACH OR EXCEED EACH SCORE
Total Number of Cases =450
Age I2 Age 13 Age 14 Age 15
No. of Problems T-Scale
Right X10 Score Per Cent Per Cent Per Cent Per Cent
Exceeding Exceeding Exceeding Exceeding
+ One-half | + One-half | + One-half | + One-half
Reaching Reaching Reaching Reaching
0 to I 27 99.6 99.6 te) (9)
22tONS yore ars 29 98.7 99.6 00 (3)
A» CONS ner: 32 97.0 98.8 99.6 99.0
6) toe7. 35 94.0 97.6 98.8 96.9
BtO Oey ee ee eee ay 91.0 97.6 98.0 O5 ea
TO: CORTE Shae ee 39 87.5 94.8 06.4 O37
12 tOMLS VL eee 41 84.1 91.6 93.6 OS
TALtONIS LEY ode ee 42 80.2 91.6 Ola 92.2
16. tO3072 4 ee 44 ao 890.5 88.3 90.0
T8touLos 45 fits) 86.2 85.1 87.9
20 \tOM2 Tia sepshaaie ae 46 66.0 Bens Srey 86.9
22° tOG23\ tone 47 60.8 83.1 76.6 86.9
24 1025... 48 56.1 fetes EO 85.8
20: tO02 7). aes 49 50.5 74.6 Oa 82.2
28 tO820.. ci een 50 43.0 73.8 68.2 80.0
ZO stORS Ts oe renee 5I S775 WTA 64.9 76.4
32 tos33s 52 eye ees 67.4 62.5 rane
ZA StORS 5 Coe eerie 54 28.1 6227 60.9 67.9
ZOetOes 7
| Right | 5-2? | Richt | 5-D-
31. Finger clip....| .105 .625 9) 0) .075 .644
32. Ford timer rol-
eR AG en hn Ba) .625 .068 .649 fe) fe)
33. Spring hinge ..|_ .158 .60 13 .63 .025 .696
34. Coin safe.....| .47 .508 sea soe Bods:
Anpewretcn ..;.'.. .842 .40
56 Measurements of Mechanical A bility
Some of these models were discarded for the reason that they
are at the time improcurable. This was the case, as has been
explained, with models No. 34 and No. 35. Of the other models
some were found to involve too much mere physical strength, asin
the case of the “finger’”’ paper clip and spring hinge.
SECTION XIV
RELIABILITY
The self-correlation of a test is commonly utilized as a measure
of its reliability. If the reliability of a test were perfect, any
number of measurements of the same individuals taken with that
test would yield precisely the same results. This never is the case
with any measuring instrument yet devised. The reliability of
various tests, however, varies greatly, and in order to interpret
intelligently correlations obtained with a given test the reliability,
or self-correlation, must be known. The self-correlation co-
efficients obtained for these tests are as follows:
Considering Series I and Series II, scored in the regular way
(counting partial scores) for 369 cases, 7th and 8th grades,
r=.59+.02. For 23 graduate university students, men and
women, 7 between Series I and Series II =.75.
For Series I alone, alternate models were correlated as follows:
Models A-C-E and B-D-F were each considered as a test,—that
is, the two halves of the test were intercorrelated. The co-
efficients found are:
SUE ASES NCTE PREITY A te Rove ala tiene sate aise agate ceed reo hs r= .68
20 cases, Ist year high school boys, 22.0465. 20s. r= .80
116 cases, 7th and 8th grade boys.................. 7=.7§
Pa S EA NOC PAUASIIIG AGasiats sciet ete seed o' vidi ate aie» ol oe r= .79
eeamtenel ese COLIN TRACE ICS woh cect oh Nv whesl fy Ai he r= .06
PMY eels OLIVET AUS MOVE ein ax ata Sorc ei Mapa ee itidiuile ere aed r=.45
It is probable from the above coefficients that the true reliability
is between .6 and .7.. For two groups, the high school class and
7th and 8th grade boys, it runs up even higher. This degree of
reliability is probably as high as can generally be obtained with
such material, but it is not all that could be desired. It is to
be hoped that further experimentation will result in scales of
higher reliability.
57
SECTION XV
CORRELATIONS
The correlations of most interest are those with general in-
telligence and with other available criteria of mechanical ability.
I. CORRELATIONS WITH GENERAL INTELLIGENCE
The most reliable of the former were obtained from the Army
records, which between Army Alpha intelligence test and Series
I are as follows:
“Camp Taylor, 109 unselected men....................-. r= .323
Camp Devens, 107, foreign eliminated, but largely inferior
PTT wei Mie Pet | | ae ead mE daa ace lnaat Mat nde ecb phe Aa alge r= .35
Camp Bee, 76cunselected: men, 0. aks fag eh aaa eee eee r= .30
Camp Lee, 30 men below 501in Army. e00 2005. .20. 2 r=.00 approx.
Camp Lee, 216 men low grade, individually examined... .. r= .00 approx.
Camp Dix, 909 men, 303d Engineers, unselected ......... r=.51
Massachusetts School Feeble Minded, 30 cases, with mental
Sh RN os UGE MPs Ge or aici ated APE Otte ee ARAN ae ee OLE gen ame r= .32
Same group with officers’ ratings. .... r= .25
For 100 7th and 8th grade boys, New vor Puphe schasis!
between Series I and composite intelligence score, made up
of Haggerty, National 1 and 2, Otis, Kelley-Trabue, and
OMY ELENA ane, cre arisen Tu oe hea fe teen Oy auth ge ote SER cheer t= .397
For same group, same tests, with Series IJ............... r= .338
2. CORRELATIONS WITH OTHER CRITERIA OF GENERAL
MECHANICAL ABILITY
The best available criteria of general mechanical ability of the
kind supposedly measured by these tests has been manual training
and science teachers’ ranks.’ It frequently is true, however, that
these ranks are too unreliable to be trustworthy, because the
pupils’ abilities are not well known to these shop instructors.
An effort has therefore been made to obtain classes having two
shop teachers, making it possiblé to intercorrelate the two rank-
ings for reliability, before considering either of them as a criterion.
The coefficients obtained are as follows:
58
A Description of the Tests
SHOP TEACHER RANK AND SERIES [|
27, 7th and 8th grade boys in Lincoln School..................005:
15, 8th grade boys in New York City public schools..............
24, 8th grade boys in New York City public schools..............
14, 6th and 7th grade boys, Horace Mann School. ................
18, 6th grade boys in Horace Mann School........
17, 6th grade boys in Horace Mann School........
59
r= .83
r= .80
r=.42
r= .81
r= .90
r= .88
~\
SECTION XVI
SUMMARY OF ASSEMBLING TESTS
We have then as a result of-our experiments three instruments
for measuring mechanical ability of the kind herein described.
Two of these, Series I and Series II, are of practically equal dif-
ficulty and can be used interchangeably for Grades 5, 6, 7, 8, high
school and adults, generally. Series ea is much easier, iets
adapted to Grades 3, 4, 5 and 6. | eeanentiiie a
The norms given are admittedly based on a pons aie
number of cases, but because of the method of scaling adopted
these can be quickly and continuously substantiated or revised as
more records become available. The correlations show that the
reliability of any one of the tests is reasonably high as compared
with other tests. More than one series now being available, this
can be increased by retesting.
The advantages in the method adopted in scaling are chiefly
that scores are reported in well-defined terms—namely the
variability of 12-year-old boys—and that the scores are directly
comparable with T-Scale scores of other tests, as well. The
short form of scoring permits the rapid testing of large numbers.
As to what the test measures our correlations show that it selects
ability markedly different from that discovered by verbal tests of
general intelligence,—the correlations never ranging over .5, and
for most groups being nearer .4. On the other hand, it does
detect those qualities that cause a pupil to excel in the opinion of
manual training and science teachers. Whatever this ability is,
it is not, however, trade skill, any more than it is verbal intelli-
gence. It is rather a composite of common sense and skill in
managing physical objects of a mechanical nature. It might be
called general mechanical intelligence and ability. The origin of
this ability is not here considered, but its distribution is shown to
be largely regardless of ordinary school classification.
Ordinarily we are most interested in determining whether a
pupil is unusual in this type of ability, and this the tests show us
admirably. As for making hairbreadth distinctions between
60
A Description of the Tests 61
pupils because of slight differences in scores in these tests, caution
must continually be counselled here, as well as in the use of other
mental tests.
We have in these tests, then, instruments for obtaining a
definite measure of a trait which is generally estimated with
great inaccuracy by school authorities as well as by parents and
pupils themselves. The shortcomings of the tests have been
repeatedly noted in this report. Their advantages and the uses
which can be made of them are obvious.
SECTION XVII
MEASURING MECHANICAL APTITUDE BY MEANS OF ILLUSTRATIONS:
- PIcTURE TESTS OF MECHANICAL APTITUDE !
I. AIM AND PURPOSE
The natural limitation in any “ material test,’’ i.e., one requiring
physical apparatus, is of course that such tests are somewhat
difficult to administer in large school systems where thousands of
individuals are to be tested. This is chiefly because the scoring .
must be done after testing each class before the material can again
be used. While a large number of outfits may be available, it is
out of the question to have a set for each pupil as with paper tests,
and it is therefore not possible to test large numbers in a short
time, as can be done with paper tests. Moreover, physical ap-
paratus, while of far more intrinsic interest to the pupils, is of
course more cumbersome to handle than mere sheets of paper,
and requires somewhat more mechanical skill in scoring and
managing.
To meet this increasing need for some means whereby a teacher
or principal may quickly obtain some measure of the mechanical
abilities of large numbers of pupils in great school systems such as,
for example, in New York City, and in survey work generally, the
writer set out to develop a series of paper tests of general me-
chanical aptitude, and to evaluate these in terms of mechanical
ability as shown by the assembling tests; by shop ability as shown
by rank given by teachers of manual training, and in terms of
general intelligence.
These tests involve judgments of mechanical relationships and
a general knowledge of things mechanical,—their principles,
operation and use. While the actual trial at manipulating de-
vices, such as in assembling tests, is sacrificed, many of the same
general mental processes are called for. Because of the difficulty
1For samples see Stenquist Mechanical Aptitude Tests, published by World
Book Co., Yonkers, N. Y.
62
A Description of the Tests 63
in obtaining suitable models for assembling tests, they are limited
in range, but the moment the problem is transferred to paper an
enormously larger range of applications is opened up. Thus,
while it is impracticable to use the assembling of a lathe or engine
as a test, it is quite as easy to treat such devices by means of
pictures and questions as is a paper clip or mouse trap.
If a paper test of mechanical aptitude, even partially as effective
as the actual manipulative tests, could be invented to measure the
same general trait, it was thought to be quite worth while because
of the ease with which it can be utilized for large numbers. The
need for something of this kind is particularly urgent in connection
with vocational and educational guidance.
Here, as in the assembling tests, the aim is to measure individ-
ual differences in that certain general ‘‘ mechanical bent”’ or “‘ turn
of mind” of children of school age,—well recognized by all,
though but vaguely defined in the minds of most persons. The
marked distinction between pupils in this kind of ability is, how-
ever, well known to every parent, and to every teacher,—partic-
ularly to teachers of any form of shop-work: but unfortunately,
almost nothing has been done to obtain an exact measure of it.
But this ability must not be confused with trade skill, or trade
knowledge. The Army trade tests are better adapted to select
skilled mechanics. This, however, is not the problem with boys
of the upper grades and high school. The problem there is to
discover differences in general mechanical interests and abilities
which will constitute reasonably intelligent bases for guidance.
2. DESCRIPTION
The technical names and language involved in mere verbal
questions on mechanics,—including descriptions of mechanical
devices and processes, defeat their usefulness as tests of general
mechanical ability. Advantage was therefore taken of what is
probably the best substitute for objects to actually handle—
namely, illustrations of such objects. By means of these it is
possible to present a great number of mechanical problems with
the utmost ease, without the use of any language, and in addition,
a large number of problems in non-technical language by simple
questions referring to illustrations.
The method of arranging the illustrations in such a way as to
call for a judgment of relationship between two or more ideas has
|
64 Measurements of Mechanical Ability
previously been employed with marked success by psychologists
in verbal and picture tests of general intelligence and other traits.
By this method it is often possible with pictures to present a more
pertinent and telling question than by technical, verbal descrip-
tion,—and always more easily.
The comparison of the mental processes involved in actually
manipulating parts of mechanical devices, with those involved in
answering the questions presented by the illustration tests, is best
portrayed by the correlations shown in actual trial. This is
treated statistically in a following section.
Selection of Subject Matter. No test can do more than sample
the almost endless variety of mechanical contrivances of man.
In complexity, they range from the absurdly simple to the almost
infinitely complex—from the stone axe of primitive man to a
Mergenthal linotype, or a modern battleship. But generally
speaking a few principles and laws of mechanics govern them all,
and each new invention is for the most part but a novel combina-
tion cf old principles for new purposes.
The specific devices selected to be used as the basis of test
questions may not therefore be of as great importance as seems
apparent at first thought. In these tests a consistent effort has
been made to select on the following bases:
1. Devices must be of general interest, and not pertain to very highly
specialized trades. (Common household articles that are of a mechanical
nature are most apt to fall within the experience of every one.
2. The question involved must be as mechanical as possible in its nature,
involving a knowledge of, familiarity with, or understanding of the pur-
pose, use, operation, construction, or reason for special size, shape, weight,
material, etc., of the device in question.
In the main the models chosen in these tests are common rather
than highly specialized devices. No trade or occupation is
singled out. But in cases where a somewhat special tool or device
is included the question asked is of a general mechanical nature,
that does not necessarily require acquaintance with that particu-
lar device.
While the present series are of a generalized nature, it is clear
that a large number of series, each of which, while not strictly a
trade test, would nevertheless deal with a restricted field, would
be of great value. Thus, for example, there is need for a stand-
A Description of the Tests 65
ardized test of carpentry, cabinetmaking, cement construction,
blacksmithing, sheet metal working, etc., particularly in connec-
tion with vocational education.
The answering of the questions of these tests involves a certain
type of information and ability in perceiving and judging mechan-
ical objects and their characteristics that seems almost instinctive -
in some individuals, and almost wholly absent in others. But
what the psychological processes and principles involved are,
is not within the province of this study to attempt to demonstrate.
It may not, however, be out of place to point out that the mental
thresholds between the type of mechanical ability herein treated—
and other skills and information, particularly general intelligence
and common sense—are not sharp, clear-cut lines. On the con-
trary, these abilities probably merge imperceptibly into each
other.
_ Scoring:AntImproved Method: Brief mention may be made of
the method of scoring, which has been so simplified that it can be
done efficiently at high speed by clerical help. In addition to
employing the ‘“‘key”’ method, a further expedient has been
introduced in binding the pages with overlapping margins® By
placing all the answers at the edge of the page they are exposed
without the necessity of opening each page and repeatedly read-
justing the stencil, which, though simple, is wasteful of time.
Thus, while as much as five minutes is sometimes required in
scoring such a booklet by the old method of opening each page
and adjusting the stencil key each time to scattered answers,—by
this method it can easily be reduced to from one to two minutes
per booklet. Keys are so designed that only one adjustment is
necessary.
Ease of scoring, while always subordinate in importance to
reliability and efficiency of the measuring power of a test, be-
_ comes of great importance to the practical administrator of tests,
and, in fact, in large school systems it becomes almost the sine qua
non of a usable test. For, if scorable only by experts and at great
expenditure of time, a test is practically worthless to school ad-
ministrators who face the alternative of ‘‘ putting it over’’ through
the machinery at hand,—the teaching and supervisory staff, or
else foregoing it altogether.
1Excluded-in-first-edition.
66 Measurements of Mechanical Ability
3. PICTURE TESTS I AND II OF MECHANICAL INFORMATION AND
APTITUDE
A total of 173 questions, some expressed in terms of pictures to
be compared one with another, and some in terms of printed
queries referring to lettered pictures of machines and common
mechanical articles were originally compiled into two tests, I and
II. In Test I was placed only non-verbal material. In Test I
the task is to determine which of five pictures “‘ belongs with, isa
part of, or is used with”’ each of five other pictures. The total
test has nineteen distinct group elements. The test is scored by
counting the total number of items right.
Test 2 is divided into ten parts, each consisting of from five
to seventeen questions. The first of these consists of nineteen
pictures of mechanical toys, and each of these pictures has been
cut into two parts. The task is then to find the missing part for
each picture. Parts 2, 3, 4 and 5 consist of a series of questions
relating to the mechanical properties of each of four lettered
pictures of typical machines: An ordinary electric bell, a
blower, a countershaft, a power drill press. The questions
asked are, however, answerable by competent persons, even
though they have not had direct experience with these par-
ticular machines, as they involve chiefly mechanical reasoning
and perception.
The last group of questions pertains to the construction and
operation of two ordinary derricks. Here as in the other groups
the ability to answer the questions does not depend so much upon
a direct experience with such machines as upon insight into
mechanical principles and usages.
Scale Difficulty Values. After a few preliminary trials had
showed that these tests correlated well with shop teachers’ ranks
and with the assembling tests, 664 of Test I and 1087 of Test II
were given to Grades 6, 7, 8 and high school. On the basis of
these records the average relative difficulty of each element was
computed. The results appear as Table XIII for Test I and
Table XIV for Test II.
T-Scale Values for Each Raw Score. The same method of
scaling as employed in the assembling tests has been adopted for
these tests. Tables XV and XVI give the T-Scale values for
each number right for each test. These tables also give the age
A Description of the Tests 67
distributions for other ages than the 12-year-olds, so that the
percentage of any age which exceeds a given score can be seen at a
glance. This is the same arrangement as in the case of the as-
sembling tests.
TABLE XIII
PICTURE TEsT I*
PERCENTAGE OF RIGHT ANSWERS TO EACH PROBLEM AND S.D. EQUIVALENTS
To Eliminate Minus Signs Zero Is Considered as at —5 S.D.
Grade 6 Grade 7 Grades
Problem eo esse | Average S.D.
Bos Per Cent Per Cent Per Cent Equivalent
Right S.D Right S.D. Right S.D
Lace kins A 621 47 .586 48. .675 45.5 46.8
Pe ee ere 3 fh 307 55 ¥235 iw fe .329 54.5 55.5
Pa ee ae 321 54.5 .288 Soa5 . 3890 53 54.3
Tak ORE dS .70 44.5 -534 49. .618 47 46.8
Le Ce rt 679 45.5 Te ABs .8II AI 43,0 —3
rc ees BS .362 es .316 56 ITA. 55 54.5
én eee .562 48.5 502 50. .50 50 49.5
OS ave cnke Paes 452 51 Sp 3 49.5 .508 50 BOs
oN bat. detene Sm 30 55 274 56. 297 Wiss oy 55-5
CO Ee Oe 262 56.5 .214 58 B22 Bes Laplne.
EL, sak Cos 286 56 .260 56.5 232 S75 56.6
ee ee See 421 52 386 533 472 51 52.0
1 Le ie ee a 242 SF, 379 is 3125 354 54. 54.8
CAP eats E252 560.5 . 309 She to2 Cee 55.0
| eit Alor .318 ite 393 53% 393 Sele 53.6
LO rane estiias .204 58. . 246 Cy be Sse Sse 56.6
4 EWN AR i ee . 238 577), .298 BSe5 . 3360 54.5 55.6
org eee aks .142 60.5 ie the S73 .193 58.5 58.8
Oe ce os ane 26 61.5 .214 58. .229 S75 59.0™
* One group of ten pictures to be matched is considered one problem.
68 Measurements of Mechanical Ability
TABLE XIV
Picture TEst II
PERCENTAGE OF RIGHT ANSWERS TO EACH PROBLEM, AND S.D. EQUIVALENTS
To Eliminate Minus Signs Zero Is Considered as at —5 S.D.
EXERCISE I
Grade 5 Grade 6 Grade 7 Grade 8
Prahienn n=168 Nn =314 nN =228 n =348
Per Cent} S.D.. |Per Cent] S.D. |Per Cent). S.D...|Per Cent; $.D.
i BAN cosa thy Batt 215 443 742 435 885 380 880 382
A Aen Sein at 57 482 682 453 .837 402 He 425
eee AER Pee ec 44 515 534 AOI -610 472 600 474
OSE ER 358 521 433 527 .470 508 511 495
CRA Sein et eh Tes a 53 499 622 469 . 767 427 765 428
O Sih ntktame oirepars 328 545 423 519 500 500 558 485
Oita e eeeelctones 590 477 623 469 790 AIO 760 429
TO... ...-eeee 405 524 459 511 482 505 495 501
LE era wee atelsiecs 547 A488 602 474 710 444 799 416
ERS ces etches .505 499 .604 473 1759 430 782 422
A Se eI oa 2 .62 469 741 435 .825 407 .852 395
DAVE AES orc sine si 482 .642 463 As 434 772 425.
Tic eee cree -46 510 -591 477 695 448 .719 442
LO a Bie sme § .50 500 .710 444 .729 439 -747 433
CT a oe .56 485 .699 448 . 730 437 .750 432
TS eek « pines .40 526 .470 501 -535 474 .578 480
TOR mere as eee .815 410 .853 395 .940 344 917 361
Average: 64 per cent right
EXERCISE 2
Figure 1
1 ciate Sheen oh ePe\ .388 528 -461 510 .579 480 2735 437
2 shia fe tatenaianeede .008 629 121 617 pane S73 -354 537
Wit prten is! Mt, Ae -490 509 NIP} 412 .769 426 .802 415
ARE ieee 047 668 .092 639 Dey 614 .189 588
cere sank .ok te .316 548 .484 504 .543 480 .705 446
(ae Liabe want ALAR att +167 597 . 268 562 . 399 526 -445 514
foe nh OW eA OREN .035 682 Sy 3 Ws) 620 Sate 612 .216 579
Soren aac .057 658 .086 637 .158 600 233 573
Average: 37 per cent right
Figure 2
Tiieieyewoaetess wate .442 520 -423 519 .570 482 -625 468
Pee 334 543 . 366 534 -519 495 -§25 494
Zalartetete eaters e . 238 571 .379 531 .430 518 605 473
Ae eat ge. .202 583 . 283 558 -399 526 495 501
SM tae -AII 522 -398 526 -456 ees .477 506
Ohare ci 8 ay3 594 -234 573 . 386 529 .460 510
PF el hina Va eee ees was: 583 .229 574 EP ee | 572 - 394 527
Average: 42 per cent right
Figure 3
1 GEA, Paar PEE A .405 524 -553 486 698 448 .705 428
mB Sei NS ee aoe .220 577 .32I 546 -390 528 -422 520
Bie vavele mudeseces .460 511 .465 509 .580 480 .564 484
Atpe cis eateries .I61 5909 1252 567 -329 544 435 514
enh ae Greens eae .185 590. Sevag 546 EY) 573 .216 579
Average: 44 per cent right
A Description of the Tests 69
TABLE XIV—(Cont'd)
EXERCISE 2—(Cont'd)
Figure 4
Grade 5 Grade 6 Grade 7 Grade 8
Problem a =168 mM =314 n=228 n =348
Per Cent}. S.D. |Per Cent} S.D, |Per Cent] S.D. {Per Cent!’ S.D:
ey ae ok ee . 430 518 .6.40 404 .830 404 .875 385
«Ry ASS Ee 280 559 321 546 482 505 .511 497
UME sao ted 6 3 107 624 143 598 280 558 314 548
PS eae ee .053 662 ray 614 236 572 -330 544
Leia er eee eee .047 668 137 610 149 604 161 599
Re .035 682 .044 671 127 614 181 5901.
71 Be ee 214 579 277 559 .490 501 .482 505
RE 2 ahd aera ee 202 584 204 583 .390 528 322 546
erate whee ek 077 643 140 608 .241 570 293 505
Io. -340 541 .360 530 456 5II A451 512
i Ry Re ee ae .088 635 .146 605 219 578 i 593
Os ee ea a es IOI 628 ob ars 594 202 584 .187 5890
Average: 31 per cent right
EXERCISE 3
Section A
Sites ae Goes .452 Sire .547 488 .629 467 .652 461
As ee ty cee Wiena as .256 566 277 559 .500 500 EN he | 496
oreg e Beare epee .179 502 .158 600 324 546 .362 535
See yo ee fe ate. 548 . 286 557 -447 514 457 SII
ie shee ae oe .250 566 . 261 564 .394 527 .402 525
a GH 3s Eten cp .244 570 -251 507 .486 504 394 527
RVG e et Seas eye ee 244 570 .267 562 -415 521 .385 529
POE e etd dcik . 262 564 .242 570 e507 496 -407 501
Thee a cet. e .208 581 .204 593 .405 524 2437 516
ONE tries ote .280 555 . 236 572 495 501 .500 500
bec thee a eee .328 545 .236 572 .552 487 uEas AOI
Average: 38 per cent right
Section B
Dn etas Mec: kta .274 560 .341 541 .517 496 . 569 483
Phare ah ates. 325 545 -353 538 .430 518 434 516
ea eres Care eee .O4I 674 .124 616 .184 590 .218 578
Ae Soa ian SI TO 618 .242 570 .258 565 3322 546
SR farvaga: sve .234 570 .302 552 .469 508 .506 498
Average: 35 per cent right
Section C
1 Ce SARE ML ah a eB 612 . 219 578 .302 552 .304 551
SE Rae Ee 185 590 .216 579 .280 559 aee5 566
= aS ee (220 545 Oy 582 .294 554 .282 558
Aer s, cto nha 5 bes .220 578 a ep 573 .368 534 .330 544
at cay) Weer es Se .O71 647 . 168 596 . 268 562 .290 550
‘Oy 5 MSR a .244 570 Pale 549 .500 500 .422 520
1S. epee eee .202 584 242 570 .507 496 .442 575
Average: 31 per cent right
Section D
iE oe 9 RS ae O41 674 .O51 664 .224 576 {22 580
eee ee 5 ware .006 751 .O13 723 .023 700
£1 ere me <1 ae eer 5 O21 704 .005 758
Pye arn eye ees I3I 612 aad 514 2364 743 .328 545
Wi os eee I19 618 3222 By 461 510 .391 527
Average: 17 per cent right
70 Measurements of Mechanical Ability
TABLE XV. SHOWING THE RAw ScorES (NUMBER RIGHT), T-SCORE EQuIv-
ALENTS, AND THE PERCENTILE RANKS FOR EACH AGE CORRESPONDING
EACH SCORE FOR TEST I
Total Number Cases, 1130
Percentile Rank Percentile Rank
Dae. 7 for each of five ages ely for each of five ages
Score | Score Score | Score
(Num-|Rquiv-| 11 12 13 14 rs ||{(Num-lequiv-| rr 12 13 14 15
ber | alent rs. rs. | yrs. rs. rs. ber | alent rs. rs. rs. rs. res
Right) ‘ss Wy uP a fe Richt) a YE uA Ne 4s
mos mos mos mos. } Mos mos mos mos mos. mos
I 15 51 64 94] 9f | 84] 74] 69
2 16 3 65 95 92 85 76 71
3 17 53 66 96 93 87 78 738
4 18 54 66 07 | 94] 88 0s
5 19 I 55 67 98 | 95 | 89] 80] 76
6 20 I I 56 68 98 95 90 82 78
7 OT 2 I ie? 69 99 96 QI 83 79
8 22 2 I 58 70 09 07 92 84 80
9 23 2 2 59 70 99 | 97 ] 83 85 81
ae) 24 2) 2 I 60 71 98 94 86 82
Il 25 3 2 I 61 71 98 94 87 83
12 26 4 3 I 62 72 08 95 88 84
13 25 4 3 2 63 72 98 | 95 89 | 85
14 28 5 3 2 I 64 73 99 96 90 86
15 29 6 4 3 2 I 65 73 99 | 96 OI 7
16 30 7 4 3 2 : 66 74 O7 fF 92°) 83
2 31 8 5 3 3 2 67 74 07 | 92] 89
18 32 9 5 4 3 2 68 75 08 93 90
19 33 10 6 5 4 3 69 75 98 93 90
20 34 II 7 6 5 4 70 76 99 | 94] 9I
21 35 13 8 7 6 5 75 76 94 OI
22 36 uty ide) 8 7 6 72 TA 95 92
23 Sr 7 12 9 8 a Ts 77 95 92
24 38 19 14 10 9 8 74. 78 96 93
25 39 22 16 12 II 9 75 78 96 94
26 40 25 17 14 ite) ame) 76 79 97 94
27 41 28 19 16 14 LI 77 79 97 95
28 42 32 21 18 16 re 78 80 98 95
29 43 36 24 20 18 I5 79 80 98 96
30 44 40 27 23 20 17 80 81 99 96
31 45 43 31 26 23 19 8I 8I 97
Bo 46 47 35 29 2K 21 82 82 97
33 47 50 40 32 27 23 83 82 97
34 48 5401 45\) 935.) 20 14025 84 83 98
35 49 59 50 38 31 27 85 83 08
36 50 OFS Sac Aa SA a PesCt Eee 84 98
37 51 oad WR oa WORE oid Niet ey hale 7 84 98
28 52 70 62 50 42 39 88 85 98
39 53 73 66 55 46 42 89 85 99
40 54 76 70 59 50} 45 90 86 99
41 55 79 73 61 53 47 OI 86 99
42 56 81 75 63 55 50 92 87 99
43 57 83 a7 65 57 52 93 87 99
44 58 85 79 67 59 55 94 88 99
45 59 87 81 70 61 57 95 88 99
46 60 88 83 73 64 60 Median
47 61 90 85 76 66 62 ||Number Right} 33 35 38] 40 42
4
49 62 92 89 80 70 65 Median
50 63 93 | 90] 82] 72] 67 T-Score AT) maou 952"| Sas be
A Description of the Tests 71
TABLE XVI. Raw Scores (NUMBER RIGHT), T-SCORE EQUIVALENTS, AND
PERCENTILE RANKS FOR EACH SCORE FOR EACH AGE FOR TEsT II
Total Number Cases, 1087
Percentile Rank Percentile Rank
for each of six ages for each of six ages
Raw T. Raw Ty
Score | Score Score | Score
Num- |kquiv-| ro | rz | r2 | 13 | 14 | x5 |{(Num-|equiv-] 10 | rz | 12 | 13 | 14 15
ber alent | yrs.} yrs.| yrs.| yrs. | yrs. | yrs. ber | alent | yrs.| yrs.| yrs. rs. | yrs.
Right) si “4 vp 4 ve Right) ee re 4 y vp yrs.
mos.}mos.|Mos.}/Mos.}Mos./MoOs, mos.;Mo0OSs.}MmoOs.}MoOs.}Mos.|mos,
I 20 I 43 6r j95 |90 |83 {78 |75 {71
2 22 I 44 62 |06 jor |85 |80 |77 {74
3 24 I 45 62 |97 193 |87 |82 |79 |77
4 20 I
5 28 2 I 46 63 |97 |94 |88 {84 |82 [790
A7 64 198 |95 |90 |86 |84 {81
6 29 3 2 I 48 64 |99 |96 j9o2 |88 |86 183
7 30 4 2 3 I 49 65 [99.4197 |94 |90 |88 {85
8 31 5 3 2 I 59 66 199.9198 |o5 |o2 |90 |87
9 32 6 4 2 I
10 33 8 5 3 2 I I 51 67 99 |96 |94 |02 |890
52 68 99.2197 |95 |93 |90
II Te 4a6 th Os Al I I 53 69 99.4;97 |96 |94 |92
12 35 12 8 3 2 2 I 54 70 99.6198 |97 |95 |94
13 36 15 9 6 4 3 2 55 71 99.9199 |98 |96 95
14 36 18 | II 7 4 3 2
15 37 21 | 13 8 5 Ae 56 72 909.2108 |97 |96
57 73 99.4199 |98 197
16 37 PM) ws Bas De Oa a ad | 58 74 99.6]99.3}98 {98
17 38 27 | 18 | «1 8 6 5 59 75 99.9199.6198 {98
18 39 30 | 20 | 13 9 7 6 60 76 99.9199 |98
19 40 Sa e217 05 4 ato 8 8
20 40 36.|"a2 [16 5|or2 } 20 9 61 ae 99.1198
62 78 99 .3}98
21 4I 39 |726,, 18! era) 12") to 63 79 99.5|99
22 42 AZ Ne20. 4820) (STOR TA. | 102 64 80 99.7199
23 43 AO | F399) (027 Nels 110 Pod 65 81 99.9|99.I
24 44 SO jse (ces ae2h Wiroulveo
25 45 54 SOh 25 [a 2s 2001 ro 66 82 99.2
67 83 99.3
26 46 | 58 | 42] 31 | 26] 23 | 21 68 84 99.5
27 47 620) FAG esa e255) ez 23 69 85 99.7
28 48 65 | 50 | 38 | 30 | 28 | 26 70 86 99.9
29 48 | 68 | 54 | 42 | 34 | 31 | 29
30 49 | 71 | 58 | 46 | 37 | 34 | 32 71 87
72 87
31 50 | 74 | 62} 50] 40 | 37 | 35 73 88
32 5I | 76 | 65 | 54} 44 | 40 | 37 74 88
33 52 78 | 68 | 57 | 47 | 43 | 40 75 89
77 90
36 55 84 | 77 | 67 | 58 | 54] 50 78 90
37 pag AM Re I Ee CEN eo IN oy SOAS, | ee eeeceerreceere sn reves: emir arvana 6 alg aa
38 57 88 | 8x | 73 |. 65 | Oz | 56 Median
41 59 93 7179) 73 | 70 | 65 Median
42 60 94 | 88 | 81 | 75 | 72 | 68 T-Score 44] 48 | 50 |] 53 | 54 |55
72 Measurements of Mechanical Abthty
Form of Distribution for Picture Tests I and II. In order to
convey an idea of the form of distribution for Picture Tests I
and II for Grades 6, 7, and 8, the following figures are included.
It will be noted that all these distributions conform fairly
closely to the normal probability form. There is no reason to
suppose the irregularities are not due to chance.
ns 809
Range 0-66 right out
of a possible 77.
Median 26.44 (46 T}.
n = 667.
Range 0-95 right out
of a possible 78.
liedian 26.76, or (55 T).
Fic. 12. Picture Test I. Form of Fic. 13. Picture Test II. Form of
Distribution for Grades 6, 7 and 8 Distribution for Grades 6, 7 and
Combined. Combined.
A Description of the Tests 73
A Ty
(co)
Grade 6. n= 183. Grade 7. n = 214. Grade 8B. n = 246,
Range 0-52 Right Out Range 8-54 Right Out Range 6-64 Right Out
of a Possible 78, of a Possible 78. of a Possible 78.
Median 21.42 (43 17). Median 28,58 (56 T). Median 29.64 (57 T).
Fic. 14. Picture Test I. Form of Distribution for Grades 6, 7 and 8
Individually.
Grade 6. n= Se Grade 7. n= Grade 8. nF 312s
=eace : © 0-54 Bint Out Range 4-64 Rena Out Range 0-66 Right Out
of a Possible 77. of a Possible 77. of a Possible 77.
Median 26.48 (46 T). Median 26.47 (55 T). Median 36.23 (55 T).
Fic. 15. Picture Test II. Form of Distribution for Grades 6, 7 and 8
Individually.
4. RELIABILITY OF PICTURE TESTS
Asa measure of reliability of Test I, the first half was correlated
with the second half. For 103 cases in Grades 6, 7 and 8, r=.79.
For Test II, 200 unselected cases from Grades 6, 7 and 8 give
coefficients as follows: Between Exercise I and Exercise 2, r=.61.
Between Exercise 2 and Exercise 3, r=.68. These coefficients of
self-correlation are sufficiently high to be acceptable. In corre-
lating the scores in either of these tests with other scores, this
reliability measure must be considered. The effect of the un-
reliability is to reduce correlations, and also to increase the ap-
parent amount of overlapping of age or grade groups. The
reliability of these tests compares favorably with that of others.
74 Measurements of Mechanical Ability
5. CORRELATIONS WITH ASSEMBLING TESTS AND WITH
SHOP RANKS
The correlations of chief interest in the case of the. picture
tests are those with other criteria of mechanical ability. The
best of these is the score in the assembling tests. Those computed
are as follows:
Test I witH ASSEMBLING TEST [|
No. r
6th, 7th and 8th grade boys, Lincoln School. ........ a By, 85
8th grade boys, New York City public schools ..... 33 .59
8th grade boys, New York City public schools ..... 35 .88
6th grade boys, New York City public schools ..... 39 44
Test II with ASSEMBLING TEsT I
5th, 6th, 7th and 8th grade boys, Lincoln School... 50 Wy irs
7th grade boys, New York City public schools..... 69 .45
8th grade boys, New York City public schools... .. 30 .59
7th and 8th grade boys, Lincoln School........... 23 .82
The other criterion available is shop teachers’ ranks. The
coefficients found are:
Test I WITH SHoPp RANK .-
No. r
7th and 8th grade boys, Lincoln School........... 27 .83
Highrschooltboyadallivears) incl. ser oe) ie one ee 53
Othiwancirth Grade DOVer ii. il ce Gas ee eae 51
Oth rade DOVS ouch oe neon imaen Peo ge st 18 .59
DEN Prager Doves. 2 wees ieee sole eoe. ye hate tee eee hat ae hy .59
Test II witH SHop RANK
7th and 8th grade boys, Lincoln School........... 27 .84
6th and 7th grade boys, New York public schools .. 14 .43
6th grade boys, New York public schools. ......... ‘Np .65
The intercorrelations of Tests I and II are also of interest. The
coefficients found are:
No. r
7th and 8th grade boys, Lincoln School........... 25 .88-
5th, 6th, 7th, and 8th grade boys, New York City
public'schoolsi'2, wean. cae nee) te eta ene eae 220.) 41.00
It will be noted that the public schools’ ranks always correlate
lower than the private school ranks. This undoubtedly indicates
A Description of the Tests 75
that in the private schools where the classes are smaller their
abilities are better known. We may accept the highest correla-
tions as most nearly true, since all chance factors tend to reduce
the correlation.
6. SUMMARY OF PICFURE TESTS OF MECHANICAL APTITUDE
The foregoing facts indicate that in these tests we have two use-
ful instruments for detecting an ability which seems to be closely
correlated with the ability to score in the assembling tests, and
with qualities which lead shop teachers to rank pupils high or
low. Itis, therefore, entirely justifiable to assume in general that
a high score in the picture tests is an indication of general mechan-
ical aptitude. To obtain the best measure, both the assembling
tests and the picture tests, are advisable. For preliminary
classification, however, the picture tests alone may serve. The
most obvious query that occurs in comparing the assembling tests
and the picture tests is somewhat as follows:
‘“May a child not be expert with his fingers and be able to score
high in working with actual materials and still have but little
knowledge of the kind called for in the picture tests, or vice
versa?”’
The answer is of course to be found in our correlations.
These range as high as .88 between the Assembling Series and
the Picture Tests, which means that there is a very marked
tendency for these two traits to be found together. This is not
equivalent to saying that the two kinds of tests measure exactly
the same traits. The difference between the obtained correlation
and perfect correlation is a measure of the extent to which one
trait occurs without the other.
The ease with which these picture tests can be given and scored
will be the chief reason for substituting them for the assembling
series.
BAR
THE NEED FOR A BROADER DEFINITION OF
GENERAL INTELLIGENCE
SECTION XVIII
ILLUSTRIOUS SCHOOL FAILURES
Cases in which illustrious (not to include ‘merely successful’’)
men and women were, while in school, diagnosed as failures by
their teachers have been often cited. Many of the men and
women who later became world authorities in their fields, were
called at best but mediocre. Linnaeous’ gymnasium teacher told
his father that he was unfit for any profession. Yet this boy later
was to revolutionize the science of botany.) Charles Darwin says
in his autobiography that he ‘‘was considered by all his masters
and by his father as a very ordinary boy, rather below the com-
mon standard of intellect.’ Napoleon Bonaparte in the final
examination at his military school stood forty-second in his class.
We may well ask with Swift, ““Who were the forty-one above
him?’’ Robert Fulton was called a dullard because his mind
seemed filled with things outside of school. Priestly, the great
chemist, had ‘‘an exceedingly imperfect education.’’ Pasteur
“was not at all remarkable at school. Books and study had little
attraction for him.’’ M. Pierre Curie, late professor of physics at
the University of Paris, and co-discoverer with his wife of radium,
‘““was so stupid in school that his parents removed him and placed
him under a private tutor.”’ Such a list as this could, if space
permitted, be continued to great length. Many men who to-day
are national or world figures, but who had a poor school record,
could be cited.
Granting that these cases constitute but a minority, and grant-
ing also a certain tendency to exaggeration by biographers who
love contrasts, these cases are still too numerous and important
to be ignored. The fundamental fact remains that the abilities
1 Citations are from Swift: Mind in the Making, Chap. I.
76
Need for Broader Definition of General Intelligence a7
of many pupils are widely misjudged in school, and the abilities
displayed either unperceived or misunderstood because of ar-
rested development, poorly suited courses, stereotyped curricula,
and general lack of sufficiently broad means for estimating
ability.
No claim is here made that all so-called low I.Q.’s are misjudged
— only that many are.
0 el eg
DeNoe,
MN
—
SECTION XIX
Tue LARGE PERCENTAGE OF “Low INTELLIGENCE”
That a great majority of pupils who enter the first grade drop
out even before the end of first year high school is well known.
Strayer’s study of 318 cities, quoted by Terman, shows that of
those who enter the first grade, on the average only 37 per cent
enter first year high school, 25 per cent enter second year high
school, 17 per cent enter third year high school and 14 per cent
enter fourth year high school. Studies by Ayers and Thorndike
also show the same general tendency. Terman says, “‘It is not
uncommon for one-third to drop out without finishing the first
year of high school.’’ Retardation and elimination figures from
every city offer annually additional testimony of the same general
facts in elementary as well as high school. Terman believes that
‘not all of this elimination is traceable to inferior mental ability,
but that a large part is due to this cause there is no longer room
for doubt.’’ With this general statement all will of course agree.
The question, however, of just how much is due to actual lack of
intelligence in its broadest sense, we do not know. Terman pre-
sents much evidence to show that with the use of the general in-
telligence tests pupils who have low intelligence and who will drop
out can be largely discovered beforehand.
But a situation in which over 80 per cent of the pupil population
is eliminated before they reach their goal, is not greatly helped by
the statement that most of the pupils who thus are eliminated
haven’t the general intelligence to proceed further. Is it not
rather an indictment both of the curricula, and of the tests which
select largely on identical bases? Terman suggests the query,
“Are high school standards too high?’’ We might alse ask are
they too narrow? Or, in general, too far removed from the kinds
of mental capacities of pupils?
If such great numbers of the school population haven’t the kind ‘|
of ability we call general intelligence, why call it general?
Fortunately there now seems to be a tendency to scrutinize
more closely the nature of the courses offered as well as the
abilities of the pupils.
78
SECTION XX
Wuat Is GENERAL INTELLIGENCE?
Certain it is that the term general intelligence is sorely in need
of definition, for by the average person, and even a large number
of specialists in educational measurement, it is accepted at face
value to mean just what it says. But is it not a loose use of
terms that permits us to use the name ‘‘general”’ intelligence to
designate mental traits which are painstakingly limited to the
literary-academic tasks of our present intelligence tests? Are we
not misleading when we say that he, and (in effect) only he has
general intelligence, who with paper and pencil can effectively do
such things as, for example, solve simple problems in arithmetic,
state the opposites for each of a list of words, fill in a number of
deleted sentences, arrange words in certain logical relationships,
decide whether a given number or word is identical with another;
or write the seventh letter of the alphabet, arrange a jumble of
words to form meaningful sentences, make a cross that “‘shall be
in the circle but not in the triangle or square; state which day
comes before Sunday; or write whether a sentinel should be trust-
worthy, whether alliteration is a form of pentameter, whether
cessation of belligerency is ever desirable; or state “‘what one
should do if it is raining when we start to school,” or repeat ‘we
are having a fine time. We found a little mouse in the trap,”’ or
repeat ‘‘3-I-7-5-9,’’ or give the greatest possible number of words
in one minute which rhyme with ‘‘day,” or any combination of
such tasks that may occupy the 30 to 45 minutes, given to an
average present-day intelligence test?
What sort of mentality has the individual who makes a low
score in such tasks but who when he drops out of school has the
ability to organize a gang that is all but indissolvable? Or who
drops out of school and builds up a world-wide business on the
identical ground where “‘ brighter’’ men have failed? Or who can
wrest from a Robinson Crusoe situation a triumphant career? Or
even he who can start a balking automobile abandoned by
““superior’’ persons—men of higher I.Q.’s? Or what shall we say
79
80 Measurements of Mechanical Ability
for the lamented low intelligence of the New York boy who es-
caped from an institution for mental defectives and who before
the authorities recaptured him had obtained and was holding a
job paying him thirty-seven do'lars per week?
To say that there are but few such cases is untrue, for even
though the illustrious cases do constitute but a small minority,
who shall estimate how many more of that large percentage who
drop out of school, because it is unsuited to their needs, would
develop into careers of marked usefulness, if their real abilities
were discovered? |
To say that such persons as those cited (except, perhaps, such
' cases as the last mentioned) are not possessed of general intelli-
-gence is to quarrel with words. Though they may classify as
“low I.Q.’s”’ by present-day intelligence tests, surely we are on
_ uncertain ground if we take such results at face value and consider
their cases closed.
It is a question of what our tests measure, a question of what we
mean to include under the term general intelligence.
If we examine the type of criteria by which nearly all these tests
are justified, we find that these consist in the last analysis essen-
tially of teachers’ estimates of pupils’ ability in school, plus rec-
ords in other academic tests. But our major contention is pre-
cisely that for many children the teachers’ estimates and their
academic record is merely an estimate of success in bookish tasks,
and here it is that fallacies of intelligence ratings creep in.
It is submitted that these intelligence tests, at best, detect only
those academic qualities of pupils which are noted by teachers,
and which, it is freely granted, are of great importance for success
in ordinary school curricula, but which do not constitute the
whole of general intelligence. Of this our abler investigators! are
fully aware, but the average giver of tests is not aware of it,—or, if
so,—overlooks it.
npr:
1See Thorndike: ‘‘Tests of Intelligence, Reliability, Significance, etc.,’’ School
and Society, Vol. IX, Feb. 15, 1919, and Henmon, ‘‘ Measurement of Intelligence,”
tbid., Vol. XIII, Feb. 5, 1921.
SECTION XXI
OTHER KINDS OF INTELLIGENCE
As a matter of fact, it seems clear that intelligence may be
classified as of many kinds. Thus, for example, the campaign
manager exhibits a quality differing sharply from that of the
locomotive engineer; while the kind of intelligence required to lay
out the construction work of a Woolworth Building is not very like
that needed to write a forceful letter, and this in turn is not very
like that employed in painting a great picture, inventing a great
engine—or modern linotype.
While it may be true that a certain minimum body of “‘sense,”’
mental agility, and some general academic information underlies
all such activities, we know from at least a few correlations
obtained (one of which appears later) that the relationship is not
very close—though it is, to be sure, positive.
If we had trustworthy criteria of ability in social leadership and
in the various political and mechanical arts and sciences, it might
be possible to devise intelligence tests that would be more nearly
‘““general’’ than those we now have. This, however, is a more
difficult matter than to devise tests of academic ability. Again,
while to measure in this wide sense the present ability of our
school population represents a heavy task,—to prognosticate its
potential ability would truly be a Herculean undertaking. But
this is not equivalent to saying that it can’t be done. Much of
the same methodology and technique which we already have
would probably apply, and progress in this direction may be
locked for. Current literature is already sprinkled with dis-
cussions of the limitations of what our present so-called general
intelligence tests measure. While unfortunately much of the
criticism of intelligence tests emanates from self-appointed
critics, incompetent for the most part to pass scientifically upon
their merits or shortcomings, the best authorities, and many of
the authors of the tests themselves, are well aware that more
comprehensive and more valid instruments are urgently needed.
“Compared to what we should like to have they are very faulty.
Compared to what they replace, however, they may be notably
superior.”
8I
SECTION XXII
GENERAL INTELLIGENCE AND MECHANICAL ABILITY
The tests of mechanical ability herein described may serve as an
example and case in point, showing a type of intelligence and also
emphasizing the need for clearer definition of just what we mean
when we say a child has but little general intelligence.
During 1919-20 several hundred boys in a New York City
public school (P. S. 64, Manhattan) were given a very exhaustive
intelligence rating by means of the combined results in the follow-
ing well known tests.!
I. THE INTELLIGENCE TESTS
The intelligence tests used in the study were:
1. National Intelligence Test A
National Intelligence Test B
Haggerty Intelligence Test Delta 2
Otis Intelligence Test
. Meyers Mental Measure
. Thorndike Visual Vocabulary Scale
AAP ws
The results of these six tests were pooled, giving equal weight
to each, and the final rating called the composite intelligence score.
These boys were next given a series of mechanical tests, consisting
of the following.
2. THE MECHANICAL TESTS
The mechanical tests used in the study were:
1. Assembling Series 1
2. Assembling Series 2
3. Picture Test I
4. Picture Test II
The detailed nature of each of these mechanical tests has been
previously given.
1 For full report see Stenquist (J. L.), ‘‘ Better Grading through Standard Mental
Tests,” Bureau of Reference, Research and Statistics Bulletin, 1921.
82
Need for Broader Definition of General Intelligence 83
If we now compare the results in the two types of examination
we may observe the following points for this group.
In the correlation between the Assembling Test, Series I, and
COMPOSITE SCORE IN 6 INTELLIGEKCE TESTS
bik TPT SD a ad Rd —fe{_[sol foe Tae
Foe ce UY YY A a | EO 0 1
GEC wal YE A eS fg
b2y 7a B E A
ts iG. 16 i A a be
ba The correlation between vi : fo ane |
General Intelligence and General
72) Mechanical Ability (2 Assembling BE| Oye | ied able: | ae
fal Tests and 2 Picture Tests) eer
~
Ei ole 7a ed Sd BS
bbs jl SM ese OB a AW ec
| fel [el [eo] e/*el%| | eal
[| ie) 20 ox cases | | ele [%) [11 A ot ctees go [|
eel TT TTT | felefofef el dele e] [oP ey TT
Ey OD a a WW
om NS Cee
~ ;
De A a
ott Stet fear eee
Pt tt fet fet ete) toe | [Pees avestice®| |
ot el | | lel el oles) ele! LT] oles) of [el | | |
| [| felt] | [Pelosi %o| oho! g%| [el%i ei |) |
F519 DE
oS
iam sei Pie rely spride fel Pal 1 (rere
(oi) 2 Sol i SG WF
oc pageeeeet SECCUGaans
ene oo
[TUE V OF ESE A iH
Rime amimatsletieli esha De ees
ee sae LI le ee |
(PS BR FS A AH
the composite intelligence score, r= .230 + .04 for 267 7th and 8th
grade boys (Fig. 17). Between Assembling Test, Series II, and
the composite intelligence score, r = .338 =.06 for 100 7th and 8th
grade boys. Between Picture Test I and the same intelligence
84 Measurements of Mechanical Abthty
rating, 7=.52+.07 for 50 6th, 7th and 8th grade boys. Between
Picture Test II and the same intelligence rating, 7=.64 + .06 for
520 6th, 7th and 8th grade boys. (See Fig. 18.)
ee ee are ee elo
FA) bMS
ole ke,
The correlation between
‘General Intelligence and Mech-
anical Assembling Test, Series s
i eR ee
BOHEME EEE EEE EEEE EERE
BP
Da SLL ABLETON
PE Oe:
If we now combine all of the four mechanical tests into one
average T-score, and correlate it with the same intelligence rating,
we find 7 drops to .21+.07 for 275 7th and 8th grade boys. (See
Fig. 16.)
The important inference to draw from these results is not with
Need for Broader Definition of General Intelligence 85
regard to the exact coefficients obtained, but with regard to the
general fact of low correlation between the two kinds of ability
here represented. Results obtained in the Army for over 14,000
COMPOSITE SCORE IN 6 INTELLIGENCE TESTS
Tee] Te] Jed Tool [so] [oe] [se] [56] fo] [aol [oe]
CARRE oe a
al a ad rte ey Ne
58 The aaa 8 between ied i i fall (D) 287% BA Sd)
F General Intelligence and Picture | § | | | | Kar)
c ptitude.
ae Seas ai iene
of (7 ame ath erate tors) TTT P| Tel |_ 1 [Palo
fr
TN
2 HH
aleetaa Cid) (SS SSS Cie apes
COSRS S088 0588 See
[mle lsaletiaisutehe ele td lecle lt lel | lal aia
lel 1 [tele leiele lel fafa lsat le le
CCR PRPerpaeceepeet
: Plet 11 Pele (|
= Spaten hace eaveeetccntaee eaueorare Average PS
= Vol | [Palle le ls Poclel%s4’ to ltl 1% 1%1 lelel 111
Lm 40)
Cie RL Rss Rere Rite le eC
a”
Risin lmmeieleis eleleh. lf ate
Wied feria [alate erations Petes lel ell Lael
et halel forsee leet tofelel Sad
leases! et te TT ee el
aallalePerelcle ls ele meet Pet Lay lobe
NSD ISIAS Gees see
SSCL 2OE Sat
FIP en latmapnhe Ps anbane [| pal fend marea
ean gees bat a fal zeto = (| Je fa] a
F170 joa NA nC Feral nw AAT lan. vam |
ie akan fan ola elon [alana dee Aida [ ol cdot Iaae
men bear out the same general fact.!
An individual’s position in General Intelligence is thus shown
to be largely independent of his position in General Mechanical
Ability and Aptitude.
1 See page 58.
86 Measurements of Mechanical Ability
Analysis of Total Distribution. Examination of Fig. 16, which
" for convenience has been divided into quadrants each lettered,
showing per cent of pupils included, shows that of the total
cases, all in Groups A and C are below average in general intel-
ligence, but all in Group C, or 20 per cent, are above average
ability in the mechanical tests. All the pupils in both Groups C
and D, or 46 per cent, are above average in mechanical ability.
Of these 26 per cent are also above in general intelligence. But
for the mechanical tests showing their marked ability in this
direction also, it is unlikely that many of Group D would be en-
couraged to look toward careers in mechanical fields, since they
have marked abstract intelligence. Conversely, those in Group
B would not be known to be deficient in mechanical ability,
though above average in intelligence. Considering mechanical
ability alone we may say that Groups C and D would likely
succeed in this direction, while Groups A and B would not be likely
to do so.
Again, if we were to rely merely on the intelligence tests all in
Group C would fail to be recognized as having ability, although 55
pupils, or 20 per cent, have ability of the other kind. Consider
next Group A, who are low in both tests: It is not without value
to have this double negative information. At least advice can be
given less blindly than without such information. Again, there
may be quite different types of abilities in which some of these
may excel. Having them segregated we can proceed more in-
telligently than otherwise, to say the least. Less progress should
be looked for, for one thing.
In short, the mechanical tests have given us important clues as
to abilities which would not be revealed by the abstract intelli-
gence tests alone. Though the correlation is positive it is so low
as to permit wide differences in deviation. These are the measure
of abilities untouched by so-called general intelligence tests.
The Trustworthiness of the Measurements. As regards the
reliability of our measure of general intelligence: Comprised as it
is of six excellent tests, say one of which would generally be ac-
cepted as a measure of general intelligence, constitutes an unim-
peachable estimate of that type of ability which we now call
general intelligence. In mechanical ability we have repeated
tests of each of two types of mechanical tasks,—the assembling
tests involving skill, and the picture tests involving mechanical
Need for Broader Definition of General Intelligence 87
information and reasoning, i.e., we have in fact four distinct
measurements of each pupil. The reliability of our measures
is, therefore, acceptable, and much better than is generally
obtainable.
The Validity of the Measurements. The validity of a test deals
with the question of what it is that it measures,—i.e., with
correlations with criteria.
The question of what the intelligence tests measure has already
been dealt with in Section XX. As to what the mechanical tests
measure we may cite the correlations which have been found in
comparing mechanical test scores with pupils’ rank in shop
courses, or in general science courses, as given on pages 59 and 74.
These correlations are all subject to chance errors which reduce
them. The true correlations are therefore higher,—probably .7 or
higher.
Shop teachers’ ranks are of course no better than regular teach-
ers’ ranks which have been attacked in a previous section. But
there is every reason to believe them equally good. Were other
and better criteria available these would be excluded. In several
of the above instances, however, only the average rank given by
two shop teachers (intercorrelating .88 or better) were used.
The mechanical tests may, therefore, be judged from these
figures to detect to a marked degree the same qualities in pupils
that are considered by shop and science teachers in judging
pupils’ relative abilities.
The second way of deciding what these mechanical tests
measure is the very direct one of merely looking at the tests and
judging what type of task it is that has been set up. Thus we
may note at once that they represent an attempt (in all except
Picture Test II) to get away from words. They deal with con-
crete and real things, as against description of things. In the
case of the Assembling Test it gives opportunity to do with hands
and mind, rather than to perform with a pencil only, or to juggle
mental abstractions.
It may be thought, however, that the mixture of abilities
revealed by combining picture and assembling tests is less il-
luminating than would be either taken alone. To observe this
point the records in one assembling test were plotted separately.
These appear in Fig. 17. Strangely enough, the percentages in
each quadrant is practically identical, with the correlation co-
7
88 Measurements of Mechanical Ability
efficient .23 as compared with .21 in the former case. The form
of distribution is very similar. The same interpretations may,
therefore, be made whether we employ Fig. 16 or Fig. 17.
In the same way the results of Picture Test II were plotted in
Fig. 18. Here the higher correlation is apparent. The two tests
are measuring more nearly the same kind of ability.
SECTION XXIII
THE RELATIVE IMPORTANCE OF THESE Two KINDS OF ABILITY
Of the relative importance of each of these two types of ability
readers must form their own conclusions. But it should be kept
in mind that we are living in a world that is dominated on
every hand by every form of mechanical device and machine.
Every moment of present-day life is influenced directly or in-
directly by the products of mechanical skill and genius. Is it not
important that ability in this field should be discovered and
developed? Rather than merely to dismiss our apparently stupid
pupils as low in what we now call general intelligence, and to rele-
gate them to some convenient class, might not our time profitably
be spent in disclosing other kinds of intelligence of which they may
be possessed ?
The question of ‘‘what knowledge is of most worth”’ will
probably never be finally answered to the satisfaction of all. But
it seems certain that as life becomes more and more complex, the
world’s tasks become more varied, and group inter-dependence
increases, there is constant need for broader conceptions of what
constitutes worth-while mental ability. We should recall that
the history of the past century, as has often been said, could well
be written in terms of the achievement of applied science and ap-
plied mechanical genius. Inventions of hitherto undreamed of
significance, which have revolutionized or at least profoundly in-
fluenced the life of every nation on the globe, have sprung from
this field of knowledge. And while the attempts to measure the
mental abilities back of these forces, which are herein described,
represent but crude beginnings, the importance of the task ts
stoutly maintained. Indeed, to explore, measure and adequately
capitalize these capacities seems at least as important as doing the
same for the more abstract type intelligence required in academic
school subjects. The discovery of special abilities has a two-
fold significance and like the quality of mercy “‘is twice blessed”’:
It not only opens the door of new promise to pupils, many of
whom have been labelled as failures, but in doing so it leads
toward further contributions to society.
89
SECTION XXIV
FICTITIOUS STIGMAS
There is a more or less universal notion that a low score in such
tasks as have here been called intelligence tests constitutes a dis-
grace that must be shunned at all costs. To fail to receive a high
rating in intelligence is most deplorable—a great calamity. This
feeling has come about partly through the loose use of the term
general intelligence, and partly through distorted estimate of the
role of intelligence in human conduct. But, absurd as it may
seem, there is a brief, and a reasonable one, which can be held for
the pupil with an actual low I.Q. as well as for the one with a
supposed low I.Q. For just as in man we find enormous individ-
ual differences in intelligence, so (fortunately) in the work of the
world we find equally great variation in the character of the
various tasks. As a matter of fact, the outstanding industrial
tendency of the past decade has been to reduce the number of
skilled jobs and increase the number of unskilled ones. The
constant tendency of our modern machine age is in this direction,
be it right or wrong. Again, consider the hundreds of thousands
of menial tasks outside of industry that somebody in every society
must perform. Is it not clear that happiness, contentment and
efficiency in such jobs are far more apt to come with a low I.Q.
than with one that is high? Indeed, even when we consider the
world’s sweetest and most lovable characters, it is not always
their high general abstract intelligence that makes the strongest
appeal. Haven’t we in the academic atmosphere of our school
rooms come to value the intellectual side of human nature out of
proportion to its real significance in life? Surely far worse
calamities can befall the human animal than that being pro-
nounced as of low intelligence. Physical disease, a crippled body,
an insane or actually feeble mind, with the multitude of tragic
afflictions which this may imply—these and many other lament-
able conditions which may befall should be kept in the background
of our mind when we feel inclined to bemoan the lot of the stupid
individual.
go
SECTION XXV
SUMMARY OF Part II
Part II attempts to point out some of the fallacies that are
prevalent in the present-day considerations of mental tests. It
recalls the many cases of illustrious men who were called school
failures, and calls attention to the large percentages of pupils who
at present appear to lack sufficient mentality to carry on current
curricula, and suggests the query, ‘“‘Is it the curricula or the
mental ability of the population that is at fault?’’ It criticizes
present-day intelligence tests as narrow and academic in scope,
being based largely on school success, shows the loose use to
which the term “ general intelligence” is often put, and maintains
that there are in fact very likely many other kinds of intelligence
than that measured by the tests given that name. Asan illustra-
tion the results of a study of mechanical ability are offered. Here
it is shown that at least 40 per cent of the pupils from a typical
school, who are below average in general abstract intelligence, are
above average in the kind of ability required in four mechanical
tests, the detailed nature of which is described. It is submitted
that such ability may be of quite as general importance as that
required to score high in the abstract general intelligence tests, in
view of the fact that present environment is so largely permeated
with the fruits of mechanical genius and applied science. Finally,
it is maintained that there is a strong, but wrong tendency to at-
tach a stigma to pupils scoring low in these so-called general
intelligence tests. Even for those pupils whose true general
intelligence is found actually low,—after adequate tests (many
being only apparently low)—even for these ample ground exists
for hopes of a useful and happy life doing tasks for which they are,
in fact, better adapted than are individuals of high intelligence.
Attention is called to the fact that just as we find very great
individual differences in the abilities of human beings, so we
find (fortunately) very wide variation in the types of the world’s
work which is to be done; and that if the kind and degree of
abilities possessed by an individual are discovered and properly
capitalized, it should be possible to find appropriate opportu-
nities for every one.
APPENDIX
ASSEMBLING TESTS
1. DIRECTIONS FOR GIVING AND SCORING
1. (2) GENERAL MANAGEMENT:
Boxes are always handled in strap carriers; bundles of 8 or 10 can easily be
moved about. Caution pupils to be careful not to drop boxes or parts. Ifa
part should be lost from a box, place a protruding slip of paper in the compart-
ment from which it is missing. Such box can then be identified instantly, and
repaired later. Series I yellow tags; Series II green tags.
(b) To Give TEstT:
Use regular classroom, and single desks, if possible. With pupils seated, and
40 to 50 boxes, and also score sheets, near the examiner’s desk, proceed as
follows:
1. Distribute score sheets, one for each pupil. (Make sure you have the
right ones.) Each pupil fills out score sheet blanks—name, age, etc.—and
leaves blank on his desk to be enclosed in the box when he finishes. (If he fails to
enclose it there is no way of identifying his box.)
2. Appoint one boy for each row to distribute boxes to each row. Do not
permit the boxes to be opened until all begin.
3. When each pupil has his box instruct as follows: ‘‘We will now read the
directions; you read them, but not aloud. (Examiner now takes one box and
reads the directions on box aloud, while the pupils read silently.) As soon as
examiner has finished, and all understand, he says, ‘‘ You have 30 minutes; all
ready, begin.”’
Note that boxes open backward. See that all get started right, beginning
with Model A, B, etc. After about 3 minutes say again, ‘‘Do not spend more
than about 3 minutes on any one model.’”’ Examiner should write down the
time of beginning, being careful to allow just 30 minutes.
4. When time is up, each pupil-hands in his box (with record sheet inside).
Stack the boxes immediately beside the scorer’s desk if they are to be scored at
once,
(c) FINISHING BEFORE 30 MINUTES ARE UP:
A few extra-skilled pupils will finish before 30 minutes have elapsed. Have
them mark the time spent on their record sheet, and allow each such record
one-half point for each minute remaining up to 30—e.g., 22 minutes spent plus
8/2, that is, 4 would be added to the score.
92
Appendix 93
(d) SCORING:
Select two or three pupils, who appeared to be doing the best in the test, as
assistants. With boxes conveniently stacked beside his desk examiner-scorer
proceeds as follows:
1. Sit down at desk. Take one box, open (cover toward you). Unfold
Record Sheet. Now inspect Model A, and write score on Record Sheet.
Inspect Model B, and write score on Record Sheet. Do the same for all models.
When you have entered a score for each model, pass the box to your first as-
sistant, who takes each model apart, being very careful that no parts are miss-
ing, and that no model is overlooked. (The examiner will need to instruct his
assistants once or twice for each model, after which they can disassemble models
quite as wellashecan. But the examiner must continually emphasize the im-
portance of extremely accurate inspectton—to see that all parts and all models are
O. K.)
2. Proceed in the same way with all the boxes. After a little practice this
process can be done at high speed, so that a whole class can be scored in a few
minutes.
To save lost motion the assistants stack the boxes directly on the strap
carriers, when they finish disassembling. Thus one bundle (of 10) after another
is finished, and strapped up ready for use again.
Note: After the boy assistants have become very expert, it is permissible to
train a very few of them to do the actual inspecting, that is, to actually enter the
scores, on the record sheets, as official scorers. This must, however, be closely
controlled by the teacher in charge, who will be responsible.
2. DETAILS OF GIVING PARTIAL SCORES
In the standard score sheets for each of Series I, II, and III, the partial score
values for various degrees of perfection in each model are listed as plus or minus
values, which are simply points above o (every model is graded 0, 1, 2, 3, 4, 5,
6, 7, 8, 9 or 10) or below 10. Minus values are used because it is often more
clear to ‘‘deduct”’ for a certain mistake than to ‘‘credit”’ for the partial solution.
A sample record is shown on page 98.
While these partial score values appear troublesome at first glance, they are
quickly memorized, and after practice with a class or two, it may not be
necessary even to consult the list of values. Occasionally new combinations of
parts of models appear, which are not listed. These need give the scorer no
great concern. He should assign what seems (in terms of the other partial
values) a reasonable score value. The justification for this is that these small
variations in partial scores affect but slightly the final score, because of the
method of scoring.
When each model has been given an individual raw score add these up, look
up the equivalent T-Scale score in the proper table! and enter this T-Scale
score in proper place under ‘‘Final Score.”’ This can all be done very rapidly
with a little practice and with assistance as suggested under “‘ Directions for
Giving and Scoring,”’ above.
1 Pages 95 or 98.
94. Measurements of Mechanical A bility
3. THE SHORT FORM METHOD—SCORING NUMBER RIGHT ONLY
For many purposes it will be found entirely adequate to disregard partial
scores and to count only the models solved perfectly. A large number of corre-
lations between the two methods of scoring results in an average coefficient of
between .8 and .9. A good plan when practicable is to give both Series I and
Series II, when scoring by the number right method. This gives a more reli-
able sampling, and minimizes the work of scoring. In utilizing this method of
scoring all values of 8 and 9 as well as 10 are counted as right.
4. RAW SCORES AND FINAL T-SCORES
The total number right (including the total of all partial score values, if the
partial score method of scoring which is the more reliable, is used) plus any time
credit which may be due, isthe raw score. For each raw score the final T-Scale
_ score appears in the table. This should be entered as pointed out above as the
final score. The T-Scale scores are the mean square Deviation Equivalents for
the distribution of 12-year-old boys, as has been explained on page 43. Tables I
and II not only give the T-Score values, but also the age distributions for several
ages, making possible an adequate definition of what a certain T-Score means,
5. NORMS
The median scores for each age constitute the Norms, for the maximum of
scores available at time of this publication (February 1921).!
6. FURTHER DETAILS OF SCORING, AND HOW TO INTERPRET
WHAT THE SCORES MEAN
On the opposite page appears a sample Standard Score Sheet for Series I.
Each pupil to be tested first fills in the heading on one such blank, and when he
has completed his work with the box, the score sheet is folded lengthwise and
placed inside the box for identification. When scoring the examiner then
writes 10 under ‘‘individual raw score”’ for each model properly assembled, and
whatever partial score (from 1 to 9) for models only partially assembled.
INTERPRETATION OF A SAMPLE SCORE
Suppose the record for John Brown, who is 12 years old, to be as follows:
Raw Score
Model'A; (perect) sige 2-2 te eum eect ee ee aL
Model Bist pertect) une june cas Ge eee eet
Model C, oR topes ote Ma coins MT ae eae
Model D, freriect) af eta He ag ty ee 10
Models ch ee ears, th Se aia Te Te 4
Modelub Soni adit hvac enn oy tenes eBay )
Model fy uiie yates Maieih oO seed i Unene mi seas: )
Model pees a oneen ct er kun area O
Mole] Feito afc chau ona ata meme teea ve cout O
Models] 3 a0 A ih Og ces ba tig pra ene re iti a i )
Ota aa) as Coe ae AL a ven At etal ag 42
Tame: Bonuses eee aces eet er vee ce )
‘Total Rawipcone eis dy ce een dee eee 42
1 See pages 45-46.
Appendix 95
By consulting the table! the T-Score is found to be 56. Referring now to
Table IX we find the following facts: A T-Score of 56 is equalled or exceeded by
only 20.8 per cent of 12-year-old boys; by only 43.2 per cent of 13-year old boys;
by 46.8 per cent of 14-year-olds, and by 75 per cent of adult men in the Army.
We may also note the medians for each age at the foot of Table IX. These
show that our score of 56 is exactly equal to the 14-year-old median. This
gives us a well-defined notion of what it means. It shows just where John
Brown stands in relation to boys of his own age as well as to those much older
than he is.
The same interpretation may of course be made for any score in any of the
Series. Any standard of performance can also be set up for any special purpose.
Thus for example it may be desired to select for certain reasons all pupils who
score higher than 75 per cent of 12-year-olds in general,”’ or ‘‘all who score lower
than 50 per cent of 13-year-olds in general,”’ etc.
7. DO. BESUSED AS: A) GROUP: TEST
Each series is designed to be used as a class test, it being more practicable to
test an entire class than a single pupil. In order to facilitate this the outfits
have all been made up in one standard size. The uniform boxes are easily
handled by means of special strap carriers, eight or ten such outfits when
strapped up being not materially more difficult to handle than an ordinary
suitcase. The outfits can of course be used over and over again.
Full directions are printed on each box. These are read aloud by the exam-
iner and silently by the pupils.
At first thought it may appear that the expense involved is too great to use
these tests as group tests. But it should be remembered the expense of time
necessary in individual testing zs far greater than the cost of apparatus, not to
mention the general impracticability of the method, in public schools. It
should also be kept in mind that sets of 40 to 50 of these tests, for testing entire
classes can be used continuously, and should be considered as permanent
equipment. If the mental measurement of children is worth obtaining it is
worth providing the necessary materials, for great pains have been taken in
devising these tests to make them essentially group tests. In the Army entire
companies were tested at once.
1 At margin of standard score sheet.
96 Measurements of Mechanical Ability
SAMPLE SCORE SHEET
SERIES I
a (a TRA ae Soe Abe res ol (Ae i ore oles as ah, alah ie ah mead Saye ny AE NSE Oak ea se Oats eter
(Nearest Birthday)
RaTAC@ sc cause ee tem ia cee tee School: Uae ross soee wichle meine
ae STANDARD SCORE SHEET
AL
EST
T-SCORE STENQUIST ASSEMBLING TES
| | SERIES I
Individual NOTE: Do not fail to place this record inside box when you
sro. have finished the test. FOLD LENGTHWISE.
. Raw « ah ”
. Model A Score Score
Cupboard Catch. Spring wrong=—5; Knob wrong=—5; Bolt oto I 24
wrong = —5. 2to 3 30
Sait LOM Smet
Model B 6 tO7 “a
Clothes Pin. Spring properly placed on 1 stick =+2. Spring placed OO Olas
at end of one or both sticks = +2. Io to Ir 38
¥2/to.13) 40
Model C sr tO 15) "42
Paper Clip (Hunt). 1 lever properly in slot, but reversed =+2. Both +8 i .
levers properly in slot, but reversed =+8. Both levers backward in 20 to 21 45
slot =-+2. All other combinations =o. 22 to 23 46
24 to 25 47
Chain. For each pair of links properly joined, +2; any number of 28to29 49
links only half (singly) joined, +2; All other combinations =o. 30 to 3I 50
32 to 33. ‘51
| Model E Sei tosdg as
Bicycle Bell. Thumb lever on pin, reversed = +1; Correct = +2. ae to 37 53
Gear on pin, reversed = +1; correct =+2. Knocker on pin, inverted Aa ay ae ze
=-+1; correct=+2. Spring hooked properly = +4. 42 to 43 56
44t0 45 57
Model F 46 to 47 §8
Rubber Hose Shut-Off. Thumb lever above spring backward=+8; 48to49 59
Thumb lever inserted under spring, any position = +2. 50 to 51 60
Model G §4,tO S50 Gr
Wire Bottle Stopper. Rubber stop in place=+1. The two heavy 56to 57 62
wires properly connected =+4. Small wire properly connected=+5. 58to59 62
on
N
s
°
mn
w
a
°
60 to 61 63
Model H 62 to 63 64
é : 3 64 to 65 64
Push Button. Button right=-+1. Button disk upside down, all else 66 to 67 «65
O. K.=+4. All O. K. except not snapped = +6. 68 to 69 666
Model I icin
Lock. Lug in place=+4; Bolt in place=+1. Spring in place=+4. 74to75 68
Cover in place with screw = +1. 76to 77 69
78 to 79 70
Model J 80to 8I 72
Mouse Trap. All right except one spring = +7; Both springs wrong, 82 to s a4
otherwise right = +4; Only loop-lever, pin, and bait-trigger right =+2; 84t085 74
Only 1 et i i = fe 86 to 87. 75
nly loop-lever and pin right = +1 88 to 89 75
90 to9I 79
TIMER BONUSE ae ee NOTE TO SCORER: Score all perfectly 92 to 93 «80
LODAL assembled models 10. ‘‘—’’ means deduct 94 to 95 80
‘ ” 96 to 97 81
RAW ies ORE Aparna a se from 10. ‘‘+’’ means add to zero. 98 to99 81
100 to 82
A ppendix 97
SAMPLE SCORE SHEET
SeErigs II
hb t.e 6 Seo Oe 8 44 8 Be 66'S sh ED DS
Ce ble a. 8 he 8 48 © «Ole w Ce 6 a eS
STANDARD SCORE SHEET
STENQUIST ASSEMBLING TEST
WARIS SERIES II
Tt ' SCORE:
| | NOTE: Do not fail to place this record inside box when you
have finished the test. FOLD LENGTHWISE.
Individual
Raw
Score
Raw lly tA
Model A. Pistol. POT ag
Two sides properly joined with screw =—1; Hammer in place = +2; 2to 3 29
Spring in proper position = +7. Alton Saas
OORT er o>
Model B. Elbow Catch. aS) tO On a7
Catch in place=+3. Spring in place=+3. Pin in place = +3. 10 toIrl 39
I2to13 AI
: I4toiI5 42
Model C. Rope Coupling. 16to17 44
Sere properly joined with screws=-+1; Center stud properly in ygto19 45
place = +5.
Model D. Expansion Nut. 24to25 48
Rings in place and sides O. K. = +4; Nut reversed or bolt reversed =—6. 261027 49
Model E. Sash Fastener. 30 to 31 51
Top and Bottom in place, with screw in place, nut down=+3; Same, 34to35 54
with nut up = +2, I spring in place=+4; Both springs in place=+5. 36to37 55
38 to 39 «55
Model F. Expansion Rubber Stopper. 40to 41 56
Rubber properly on cone—+6. Bolt upside down = —4; nut wrong = a : 33
Te: 46 to 47 58
48 to 49 59
Model G. Calipers. 50 to 51 60
Spring in place on both arms with adjusting screw in place of eye =+5; 52 e 3 ye:
Pivot in place=+2. Sleeve in place =-+1. Be to 57 62
, 58 to 59 63
Model H. Paper Clip. 60 to 61 64
Spring in place on jaws = +2; Pin inserted properly = +6; Pin inserted 62 t063 65
improperly = +1. 64 to 65 66
66 to 67 67
Model I. Double Acting Hinge. 68 oo se
For each pin in proper place = +1. us iS a a
741075 71
Model J. Lock No. 2. rp enite oe
Bolt in place=-+1. Lugin place=+1. Both in place=+4; Spring 78t079 74
in place = +6; Cover in place = +1. 80 to 81 74
82 to 83. 77
84 to 85 77
IEEME BONUS saree aes e NOTE TO SCORER: Score all perfectly 86to87 78
TOTAL assembled models 10. ‘‘—’’ means deduct 88 to 89 78
RAW SCORE een from 10. ‘‘+’’ means ‘add to zero.”’ 90 togr 80
92 to 93 «82
98 Measurements of Mechanical Ability
SAMPLE SCORE SHEET
SERIES III (Tentative)
(Nearest Birthday)
Grade: x.iaiincs be eas ae SCHOO seo ts Fee rete Datevor (Birth eee cee «eee ie
SCORE SHEET
STENQUIST ASSEMBLING TEST
EBLE Sa ei vistiacore ates tts SERIES III
(If less than standard) (For Grades 2, 3, 4, 5 and 6)
NOTE: Place this sheet inside the box when you have finished. Fold lengthwise.
SCORES: Model A. Plain Bolt and Nut
No partial score. Right or wrong. Not necessary to screw nut up tight. Score:
0 or IO
Model B.—Bolt and Wing Nut. (Perfect Score =10.)
Nut reversed =plus 2 only.
Model C.—Plain Hinge. (Perfect Score =1o.)
Two halves joined, but one part inverted: plus 2. Pin inserted in one part
only =o. Score: 0, 2 or 10.
Model D.—Key and Ring. (Perfect Score =10.)
Key only half on ring =plus 2. No attempt =o.
Model E.—Turn Buckle. (Perfect Score =10.)
Screw eyes properly in one end only =plus 2. Not necessary to screw up tight.
Model F.—Drawer Pull. (Perfect Score =10.)
Washer wrong in any way, subtract 5. Finished surface on wrong side, subtract
4.
Model G.—Trunk Caster. (Perfect Score =10.)
For failure to push pin clear through, subtract 8.
Model H.—Plain Push Button. (Perfect Score =10.)
For button out of place, subtract 6. Parts merely laid together (not screwed up)
score I only.
Model I.—Belt and Buckle. (Perfect Score =20.)
Permanent end properly fastened, score 10. Loose end properly buckled, score 5.
Strap not reversed (right side out) credit 5. Subtract same amounts for each step
wrong.
Model J.—Nail Clip. (Perfect Score =20.)
Jaws and pin properly in place, score 10. Spring properly in place, 10. Spring
reversed, 5.
TOTAL
SCORE:
MECHANICAL APTITUDE TESTS
INSTRUCTIONS FOR GIVING TEST I
Pupils must be seated so as to prevent copying.
Desks are cleared, pencils provided, and monitors pass out booklets, one to
each pupil.
Examiner instructs all pupils to fill in properly the heading on the blanks,
being especially careful to obtain the correct age—by last birthday.
Examiner says: ‘‘Lay pencils down! Before you begin I will show you ex-
actly what you are todo. Let us read the directions.’’ Examiner then reads
aloud the instructions on the front page, while the pupils read silently. Ex-
aminer then asks if all understand. If some do not understand, repeat as much
as is necessary.
Examiner now says: ‘‘Open your booklets to Exercise 1, and turn the op-
posite page under like this.” (Demonstrate. The pictures of Exercise 6 which
appear upside down on page opposite Exercise I are then out of sight.) ‘‘You
see that there are 3 problems in Exercise I all like the sample test on the front
cover which we have just looked at; do them all in the same way. When you
have finished Exercise 1, turn the page over and do Exercise 2, then Exercise 3,
then Exercise 4, and so on until you have tried them all. If you don’t know the
right answers, guess. Write one letter in each square.”
Repeat privately any instructions necessary. Each child must understand
what he is asked to do. No child is expected to answer al] the questions cor-
rectly, but he should try them all. Examiner must see that answers are being
plainly written in the proper place; that is, in the blank spaces provided in the
margins.
Time: Allow 45 minutes if necessary. Booklets are handed in as soon as
finished, but examiner should be careful not to imply by word or manner that
this is a speed test. The intention is-to give all the time desired by 95 per cent
of pupils.
INSTRUCTIONS FOR GIVING TEST II
Pupils must be seated so as to prevent copying.
Desks are cleared and monitors pass out booklets, one to each pupil.
Examiner instructs all pupils to fill in properly the heading blanks, being
particularly careful to obtain correct age—by last birthday.
DIRECTIONS FOR EXERCISE I
Examiner says: ‘‘Lay pencils down. Before you begin I will show you ex-
actly what you are to do. Turn to Exercise 1. Let us read the directions.”
Examiner reads aloud, and pupils silently, the directions for Exercise I printed
99
100 Measurements of Mechanical Ability
in test booklet. Examiner must read slowly and point out ‘‘picture T’’ and
“‘picture H”’ while holding booklet up before class. Examiner must also point
out where letters T and H are written in the space for the answers. As soon as
all the pupils understand what they are to do, say: ‘‘ Ready—begin.”’ At the
end of 10 minutes, or when all have finished, say: ‘‘Stop. Lay pencils down.”
DIRECTIONS FOR EXERCISE 2
‘‘Turn to Exercise 2. Let us read the directions: ‘Look at Figure 1 on op-
posite page, and answer as many of the questions below as you can. Answer
each question with a single letter. If you don’t know, guess.’ When you have
finished Figure 1, do the same for Figure 2, Figure 3, and Figure 4. If you
don’t know what to do, raise your hand.’’ As before, instructions are repeated,
if necessary, until all understand what is wanted. When all understand,
examiner says: ‘‘Ready—begin.’”’ Allow 18 minutes. At the end of this time,
or when all have finished,! examiner says: ‘‘Stop. Turn to Exercise 3.”
DIRECTIONS FOR EXERCISE 3
Section A. ‘‘Look at the machine parts on the page opposite Exercise 3;
now look at Figure 1 and Figure 2 in Exercise 3. Find where each machine
part belongs in Figure 1 or in Figure 2. For example: part A belongs at I in
Figure 1 or in Figure 2; so A is written beside 1 in the space for the answers.”’
(Point to pulley A and to the pulleys numbered 1 in the two figures so that all
may see the correspondence.) ‘‘ Part W belongs at 2 in Figure 1 or in Figure 2;
so W is written beside 2 in the space for the answers.”’ (Point to pulley W and
to pulleys 2.) ‘‘In the same way find which of the machine parts belong at
3, 4, 5, etc., in Figure 1 or in Figure 2, and write the letters opposite these
numbers.”” Allow 10 minutes.
Section B. ‘‘ Now read all the questions in Section B and answer as many
of them as youcan. If youare not sure, guess. When you have finished, hand
in your booklet.’”” Allow 12 minutes.
As the nature of this test is somewhat unusual, the examiner must make sure
that the pupils understand what is required of them, and for this reason direc-
tions may be repeated, or given privately to any pupil who does not understand.
The examiner must not, of course, indicate or suggest what is the correct an-
swer in any case, when repeating instructions. Examiner should see that
answers are being written in the proper place.
DIRECTIONS FOR SCORING
These tests have been carefully planned to permit of rapid and accurate
scoring. All answers are designedly placed at the extreme right-hand margin
for each exercise, to facilitate easy checking of answers.
All answers are either right or wrong.
To find the number of correct answers, place the closed test booklet face up
on the cardboard key, allowing the latter to project at the right-hand edge
sufficiently to expose list of correct answers for Exercise 1; now open booklet to
Exercise 1 and check off, with ink or blue pencil, each right answer, counting as
1If they finish before time is up.
fs
9
Appendix 101
they are checked. Write the number of correct responses at the foot of the
column. Then turn to Exercise 2 without removing booklet, pulling the book-
let slightly over to the left on the key to expose list of correct answers for
Exercise 2, and continue checking and counting the right answers as before.
Do the same for all the exercises. Then copy the exercise scores on to the front
page and add to find the Total Score. Then fill in the corresponding T-Score
from table. In the case of Test I the booklet is reversed to correct Exercises
4,5,and6. The scoring can be done very rapidly and accurately by any teacher
or competent clerk.
i\\ Na Serene
>) Ne 7, ” i ’
ee ee
Kea we Brien | salted
si ‘gh la a Pe ey es A hs . iy
ya Tha ait “igi he Las bbe an rhe: aire on MH |
‘jeas Ne, \? wee ni Ye "ad Wein er ite tare balay
penn bk) " a ‘at ie
Mag Laie Site ely a 5
ia ae E Da Py M4. ess.
esi
Seis
a
| A
madres ec Om
esi
oS
yates
‘27%
"ee As
> =
"4 eat ane ian
ee ;
al