. - . _ 7 ‘77- - :—_—_ __- —v fl“ “..‘1' "‘""H"‘-' ‘M' ‘--*‘.-, °'M“.~"‘-..“...'.‘.'.‘ 0- M1‘?\'.\\°"‘.’T"1'o‘.""'.'.'""'°' 'N...‘
‘ ‘.‘14.'0‘. cad” MM“'.\ m mm??? 'W“"“~'M' " '

l‘I- .

' PRACTICE EFFECTS on ~
fPATIAL ABILITY TEST SCORES   _ l‘ .

   
 

   
 

    

 

       

   
            
  

 
 

v
. I
. ‘
.. '
- -
‘ .
n
- l
. .
- §
. ~ .
. ‘
. . -
‘ I
. I
. O
. O
‘
Q " .
c L
‘ -
. I
A
o
-
g, .
' y
- v
w
u
u
. u
I . r f
0 _ ‘ I , .
. ,
I
. -
v k.
. . n o
0-- ‘
P. ' .- . ‘
‘ n
.
l - . ' o . . .
. . . . .. A ' n _
r . . _ . ._ _ r. ‘ _
--. _ r v .
7 ' . . I _ ' ' ’ ‘ ... .. ' .
i ' . I ‘ I " ' r c . 9-. .‘. ' ‘ ' ' . .
_ . . , . .. .— ._ '7. . a - . . . .
I on I . . a v ’ . - u ,- .
' ‘ c . . -' a o .'
. ‘ ' . ' I ' - I . . . . . ‘ ' ‘ r _ .
. o u , _ , . ". '," .
'U' . . ' _ . ‘ '0. ' 5'. f V 4 ‘
. “ . . . . . ‘ ' V o ' a. o r . —
.V‘ .. _ _ ‘ n 0 l ¢ 0 r a ‘ .‘. , ‘ ' ~
”V‘Q . . ‘. _ ' " ' '0 - ‘ .v v. . "
. " s g . . I ~ _ ‘ - - -,
no. ‘ - -. — v , " cal — . n. ’ ... .J. . 4 . ‘ ' ‘. g ‘
~ _ ' . . . . . , .. " ' a ' - .
-v . - . ‘ ‘.. ._ " . _'....- ‘ ... .
— . " a: . o. ‘ - .’t~ -a- . .
- - - . . .. . ' "' "" " ' -. -. -
.A ' ’ ’ ‘ ' - ,. ‘ . _’ .0. - a . . '
, _ . . o'— .‘ . ... , _
n. .. . -., u ,‘ ,.,..,,_ _. ._I. ’ . ‘.
. - ' 7' " '7 "..
‘ ' u l __ ' ' ‘ . ' ‘ ’ t— w'-O‘... ..._.r _ , . - .‘
l . D ' .n- - ' ."—’ -. . I . -
" 'Ov .-. ' ' . . . ' ' ' .0 0.1.0.7; ' 'O-r ' .
- o . . ' .. .. " ' ‘ " " ' ' -0v-~ ' ' "
- ~ . ., -- .' .'. ' ~ “ ‘ ode-rcl-.——- . " “ H" ' . ,.,..
. . . . .. . ', . .. ' ~' - .. ..-. . -rt-- - -,, ...-
I - o . a... .f - nlv o. '04. _"r-_’ I. ‘. ' ' _‘
' . ‘ I I '7 o I .. - I y ' O .." I' ' .0
. . . . ‘ ..‘. . "2 VI. ., I. . . .,... r g...¢/ .,, _" ' "' . . ' . ._ ._ "‘
I gnu, 1... . . .. ". v 'a'.",. .. r‘ " '11'.-‘. l'. ’ 4. .‘ . _ 1-. '
‘ ... _ ‘ . ' . . . ..3'_ . . ,. ,.”’ .vo.—-p...,._ .‘.<__'-’ _.’ r-o-. .-gv. .,. ...
A» . . - " . . .. 4 . "‘ . 0" . ~.‘.. 9!-‘lo- . O“.-’ §.'VI"-t-o up *—-(r’0"---o- .. ’I'O.o .I- r..'-’ 'f_-“a.'
.. I .- "‘ ' .‘D O I . _
. . . - . .,_ - . ',_" . '. ‘O ‘ _ . . ' "" _"‘" ’ " - .” . I ‘ '. 'v. a. ."-"’.o l-up‘. — '. .' - '000- on .>.o I,
Q . .. . . ... . '. .I.-- A o o r 7. . ,4..’.‘. - ‘ .,. p. ‘. ’0‘... ". " , - ' , , .. ._.r”,. .v".‘.
.7. . _ _ . .‘ ' ' . ‘ . A».I '-.--4 - .’ _ A.‘. «.0 I - ~70 .,,‘ . ,‘ ., ,'.."r' ’1’"! .' .-- ..-
.. 0. .— .~ v .,_. ,’ ,,‘- g-.,,.',. d
. . . t .. . . . . .... .- ‘ -u .- ,‘ l ' -’ _:.’ ‘ _ .‘ '\ I..'. - .IIJ’M’
""‘ o~.cog..’..,.._ - ... .‘g'\|.'.“"" O‘ ~ -v- -v‘ I. .o’.’((n’o utodi‘od' Q!’-. m ‘ ‘O 0 v- ”’0’ O’e-I‘OOV’. \ . v.
7 O

 

 

ABSTRACT

PRACTICE EFFECTS ON SPATIAL ABILITY TEST SCORES
BY
Fred Fallik

This research attempted to examine transfer of
learning effects as a result of practice on spatial ability
tests. Three aspects of practice effects were analyzed;
the amount of transfer incurred as a result of practice,
the amount of transfer incurred as a function of gender,
and the relative stability of transfer effects over periods
of latency.

Four research groups were established, each con-
taining 15 subjects; trained males, untrained males,
trained females, and untrained females. Training consisted
of four hours of practice on various types of spatial
ability tests conducted in two practice sessions. Prior to
training, all groups were administered the Revised
Minnesota Paper Form Board Test. At the conclusion of the
experiment, all subjects were administered an alternate
form of the Revised Minnesota Paper Form Board. Trained
subjects were administered two additional forms of the

Revised Minnesota Paper Form Board during the practice

Fred Fallik

sessions. Following the conclusion of practice, there was
a six day lapse at which time the post-test was adminis-
tered.

Analysis of error score differences based on
comparisons of different tests showed that practice
generally promoted transfer of learning; neither gender
appeared to benefit more than the other, and effects of
transfer appeared to remain relatively stable over time.

A serious drawback on interpretation of results was the
presence of ceiling effects for the trained males. It was
conjectured that this effect might have been a factor in

the absence of gender differences.

PRACTICE EFFECTS ON SPATIAL ABILITY TEST SCORES

BY

Fred Fallik

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology

1973

Dedication:

To Bev

ii

ACKNOWLEDGMENTS

I am indebted to Dr. Frank Schmidt for encourage-
ment, suggestions, and valuable criticisms during the
planning, carrying out, and analysis of this experiment,
and to Dr. John Wakeley for his valuable methodological
comments. Acknowledgment is also made to Dr. S. H.
Bartley for the initial motivation and inspiration which

provided the foundation of this research.

iii

TABLE OF CONTENTS

Page

LIST OF TABLES . . . . . . . . . . . . . Vi

LIST OF FIGURES O O O O O O O O O O O O O Vii
Chapter

I 0 INTRODUCTION 0 O C O O O O O O O O 1

II. REVIEW OF THE LITERATURE . . . . . . . 3

Historical Trends . . . . . . . . . 3

Factoring Spatial Ability . . . . . 6

Measures of Spatial Ability. . . . . . ll

Differential Test Results . . . . . . 14

Sex Differences . . . . . . . . 14

Improving Spatial Ability. . . . . . 15

Correlational Designs . . . . . . 15

Experimental Designs. . . . . . . 18

III. PROBLEM STATEMENT. . . . . . . . . . 25

Iv. METHODOLOGY 0 O O O O O O O O O O O 27

Subjects . . . . . . . . . . . . 27

Materials. . . . . . . . . . . . 27

Pretest . . . . . . . . . . . . 29

Training . . . . . . . . . . . . 29

Posttest 0 I O O O 0 O O O O O O 31

V. RESULTS 0 O O O O O O O O O O O O 33

VI. DISCUSSION . . . . . . . . . . . . 54

VII. CONCLUSIONS AND IMPLICATIONS . . . . . . 62

iv

Chapter Page
REFERENCES 0 O O 0‘ O O O O O O O O O O 64

APPENDIX. 0 O O O O O O O O O O O O O 69

10.

11.

12.

13.

LIST OF TABLES

Factors, Levels, Nature, Degrees of Freedom,
Mean Squares, and Error Terms . . . . .

Analysis of Variance Using Uncorrected
Difference Scores Based on Tests One and
Four . . . . . . . . . . . . .

Analysis of Variance Using Corrected Differ-
ence Scores Based on Tests One and Four. .

Analysis of Variance Comparing Practice and
Gender Effects for the Experimental Group .

Newman-Keuls Test for Measure Effects . . .

Comparison of Change Scores for the Male
Experimental and Control Groups . . . .

Comparison of Change Scores for the Female
Experimental and Control Groups . . . .

Comparison of Change Scores for the Control
Male and Female Groups . . . . . . .

Comparison of Change Scores for the Experi-
mental Male and Female Groups . . . . .

Average Error Scores, Standard Deviations, and
Correlations by Test and Group . . . . .

Average Change in Error Score From Test One
to Test Four . . . . . . . . . . .

2' Transformation of r14 and Standard Deviations
from rtt . . . . . . . . . . . .

Adjusted Correlation Between Initial and
Change Score . . . . . . . . . . .

vi

Page

40

40

41

44

44

46

46

47

48

49

50

51

52

LIST OF FIGURES

Figure Page
1. Analysis Design . . . . . . . . . . 35
2. Average Error Score by Test . . . . . . 37
3. Average Decrease in Error Score by Group. . 37
4. Comparison of Average Decrease in Error

Scores by Gender . . . . . . . . . 38
5. Comparison of Average Decrease in Error
Scores by Group . . . . . . . . . 39

vii

CHAPTER I

INTRODUCTION

Spatial ability, the ability to "visualize and
manipulate two or three dimensional objects in one’s mind"
(Schmidt, 1971), has long been recognized as having
theoretical importance. Most theories of intelligence
have, to a greater or lesser degree, incorporated spatial
ability as an aspect of this functioning. Spatial ability
was first considered as having only theoretical importance.
Angell (1910) reported on the practical application of
tests available at that time to the American Psychological
Association. He concluded that, "None of the (spatial
ability) tests possess any serious significance apart from
the intrOSpective evidence gained by it." This conclusion
has proved to be premature.

Tracing the history of spatial ability in terms of
its correlation with occupations, Smith (1964) pointed out
that spatial ability was considered of practical importance
in only a limited number of specialized fields such as
engineering, architecture, draftsman, etc. A contemporary

survey by the United States Employment Service (1957)

examined the extent of the practical importance of spatial
ability to areas not previously considered. The survey
indicated that there were at least 84 job titles in which
a high degree of spatial ability was required.

In light of the considerable practical importance
of spatial ability the determination of the nature of
spatial ability has been of some concern from both a
theoretical and experimental aspect.

Research concerning the nature of spatial ability
and its relationship to other abilities has been extensive.
Unlike other abilities, few attempts have been made to
examine the conditions and factors involved in improvement
of spatial ability. This area, the improvement of spatial

ability, thus becomes the focus of this research.

CHAPTER II
REVIEW OF THE LITERATURE

Historical Trends

Spatial ability, defined in various ways, has been
included in discussions of intellectual functioning since
scientific speculation on individual differences began.

Probably the first to empirically determine the
extent of intellectual individual differences was Binet and
Simon's (1905) classic attempt to predict scholastic
achievement. Although not specifically termed spatial
ability, Binet and Simon (1905) did measure visual memory;
an important component of spatial ability.

Spearman's (1927) contribution to knowledge of
spatial ability was also significant. Using testing and
rudimentary factor analysis Spearman (1927) derived what
has been called the "two factor" theory of intelligence.
All intellectual tasks could be described in terms of a
general ("9") intellectual component and a task specific
("s") component. Spearman recognized that many specific
factors were broad enough to encompass a wide range of

performances on related tasks. Spearman cited verbal

ability as the most important of these "group" factors.
Spatial ability was thus defined as being a component of
the mechanical-practical group factor. In general,
Spearman felt that the importance of group factors was
overshadowed by general and specific factors.

Burt (1949), following Spearman's lead in utilizing

the concept of a g or general factor, felt that group
factors deserved more emphasis than had previously been the
case. Burt, while keeping "g", discarded Spearman's two
factor theory. Instead Burt viewed intellectual functioning

as a series of dichotomous abilities arranged in a hierar-

chial form. The primary distinction proposed by Burt was

between and "practical" factors. This distinction

9
appears to be that between abstract and applied intellectual
functioning. Spatial ability, considered by Burt (1949)

as a practical ability (among other practical abilities),
was placed in a position of central importance in intel-
lectual functioning.

Vernon (1950), responding to Burt's (1949) attempt
at dichotomizing intelligence, replaced "9" at the apex of
the intellectual hierarchy. Two major components of "g"
were found; verbal-education ability (V:ED) and practical-
mechanical ability (KzM). VzED was further factored into
the minor group factors of verbal and numerical ability.
KzM, on the other hand, factored into three components;

spatial ability, manual ability, and mechanical information.

Third order factoring produced many specific factors which

Vernon considered trivial. Using factor analysis spatial
ability was found to load .25 with "g" (p. 7).

Departing from a hierarchial mode of intellectual
functioning, L. L. Thurstone (1938) placed primary emphasis
on group factors. De-emphasizing the role of a "general"
intellectual ability, Thurstone (1938) isolated seven main
factors one of which was spatial ability. While arguing
for group factors, Thurstone admitted that since scores on
the seven factors were correlated, a second order analysis
would produce a "g" or general factor.

Guilford's (1967) "structure of intellect" model
also places little stress on a hierarchial theory of
intelligence. As explained by Guilford, any type of
intellectual functioning can be described along three
fundamental dimensions; operation, content, and product.
In light of these dimensions, Guilford (1967, p. 227)
classified spatial ability as a cognitive (operation)
transformation (product) of figural material (content).
Guilford (1967) elaborated on two different processes which
may take this form; cognition of visual figural systems
(CPS-V) and cognition of figural transformations (CFT).
While empirically difficult to separate, CFS-V is related
to the ability to apprehend visually the spatial arrange-
ment of images. CFT, on the other hand, is related to
transformations in the image's quality or arrangement

(1967, p. 211).

Cattell (1971), unlike either Guilford (1967) or
Thurstone (1938) postulated two distinct kinds of intel-
lectual ability; crystallized and fluid. Crystallized
intelligence is fluid intelligence which has been "in-
vested" in some stable trait. Factor analysis by Cattell
showed spatial ability as a separate and distinct ability
(p. 29). In examining the relationship between spatial
ability and scores on the Cattell Culture Fair Intelligence
Test, correlations of +.73 and +.03 were found with fluid
and crystallized intelligence, respectively. Despite this
correlation, Cattell argued that spatial ability, as
defined by previous researchers, is part of a broader
second order factor called power of visualization (p. 107).
Thus, spatial ability can be grouped with spatial orien-
tation, form boards, Gestalt closure, etc., all of which
require some degree of visualization.

This brief overview of the history of spatial
ability as it relates to theories of human intelligence
attempted to contrast the different contexts and different
emphasis placed on its importance. The amount of theorizing
that has attempted to place it in perspective with other

human abilities only lends credence to its importance.

Factoring Spatial Ability
Aside from considering spatial ability in relation

to other human abilities, several attempts have been made

to shed light on the nature and processes of spatial
ability itself..

The existence of spatial ability, as separate and
distinct from other abilities, is no longer in question.
The problem that researchers have faced in studying spatial
ability centers on the nature of the ability and the com-
ponents that comprise it. Examination of the dynamics
involved in the spatial process have led many researchers
to conclude that the ability itself is composed of several
distinct sub-processes. The number, nature, and relative
importance of these processes remain, to an extent, a
matter of debate.

Burt (1949) described spatial ability as being
comprised of several sub-processes: "(Spatial ability is)
the ability to perceive, interpret, or mentally rearrange
objects as spatially related." Although spatial ability,
thus defined, is a sequential process, spatial relations
tasks could be of two types; static or kinetic. Both of
these tasks may be presented in two or three dimensions.
Spatial ability, according to Burt (1949), was the ability
to think concretely while verbal and numerical ability
involved more abstract functions. McFarlane (1925) also
makes this distinction between concrete and abstract
thought: "This type of performance test (i.e. spatial)
measures an ability whose uniqueness lies in the fact that
those possessing it in a high degree, analyze and judge

better about concrete spatial relations than do other

individuals who perhaps excel in dealing with more highly
abstract situations."

Another researcher who supports this position is El
Koussey (1935). Having reviewed the results of many
different tests of spatial ability, he concluded that the
tests measured what was essentially a unitary trait; that
of visual imagery. The consideration of spatial ability
as a unitary trait was probably heavily influenced by
having contrasted it with verbal and numerical ability
which, at that time, were also considered unitary traits.

Smith (1948), expanding on E1 Koussey's (1935)
definition of spatial ability, commented that it was an
ability to "form and retain an exact impression of shape
and pattern." According to this definition, spatial
ability was the result of two processes, one of image
formation and another of image retention.

Renshaw (1950) added a third sub-process; that of
rotation. Rotation involved the ability to mentally
manipulate images and to move them to alternative positions.
Both Spearman (1927) and Kelley (1928), prior to Renshaw
(1950), had argued for no more than two sub-processes.
Spearman presents an ambivalent argument, citing evidence
both for and against group factors. Kelley, like Smith
(1948), pointed out that there exists a process of mental
manipulation of spatial relations and another involving

sensing and retention of visual forms. This latter

ability has been redefined by Guilford (1947) as "visual
memory."

Thurstone's first analysis of primary mental
abilities (1938) found only one spatial ability factor.
Later studies done by the Army Air Force (Guilford, 1947)
found that spatial ability consisted of two factors, one
having to do with "awareness or appreciation of spatial
relations and the other with mental manipulation of objects
in space." During these researches a third, ill defined,
factor was sometimes found. In contrast with his earlier
work, later studies by Thurstone (1950, 1944), tended to
support these findings. French (1951), summarizing several
factor analysis studies, isolated three factors: S, SO,
and Vi. S was defined as an "ability to perceive spatial
relations accurately and to compare them with each other."
SO was an ability "to remain unconfused by varying orien-
tations in which a spatial pattern was presented." Vi
involved the "ability to comprehend imaginary movements in
three dimensional space or to manipulate the objects in the
imagination."

A later study by Michael, Guilford, and Zimmerman
(1957), also found three factors involved in spatial
ability. Isolated were factors called SR—o, Vz, and K.

Spatial relations and orientation (SR-O) refers to an
ability to comprehend the nature of an arrangement of
elements within a visual stimulus pattern . . ." Visuali-

zation (Vz) requires "mental manipulation of visual objects

10

involving a specified sequence of movements." French's
(1951) S and Vi factors appear to roughly correspond to
Michael's (1957) SR-O and V2 factors, respectively. A
third factor introduced by Michael (1957), and not accounted
for by French (1951), was K or Kinesthetic Imagery. This
factor represents, "merely a left-right discrimination with
respect to the location of the body.” This last factor K,
historically, is not included as a component of spatial
ability. The three factors identified by Michael, gt 31.
(1957), while factorially identifiable, were also found to
be highly correlated with each other.

More recently, Horn (1965) concluded that visuali-
zation, as a factor, was much broader than spatial ability.
As defined by Pawlik (1966), visualization is "the ability
to imagine properly the movement or spatial displacement of
a configuration or some of its parts." Following this
definition, visualization is not a sub-process of spatial
ability but a higher order factor which is also involved in
such tasks as: envisioning gear movement relationships,
change of view when an object is rotated, and envisioning
what a paper will look like when it is cut in a folded form.
As defined by these operations, visualization has been
called by Cattell (1971, p. 34) a "provincial factor" since
the response area can be localized in the brain and the
factor covers a wide range of behavior. Other abilities

which could be subsumed under this redefined visualization

11

factor, in addition to spatial ability, are adaptive
flexibility (Guilford, 1967) and speed of closure
(Thurstone, 1950).

In summary then, spatial as a process, can gener-
ally be said to be comprised of several sub-processes.
While the nature and definition of these sub-processes
remains an area for debate, current research seems to
support the conclusion that there are three sub-processes.
The exact nature and effect of these sub-processes depends
upon the definition of spatial ability itself. Regardless
of their nature, they appear to be highly intercorrelated.
One of the sub-processes which has consistently been cited
as central to spatial ability, visualization, has recently
been postulated to subsume other cognitive abilities not
involving spatial ability. Visualization thus may be a

broader factor than spatial ability.

Measures of Spatial Ability

Generally speaking, tests involving spatial ability
may be divided into two classes; those requiring the
physical manipulation of objects into a specified pattern,
and paper and pencil adaptations of this process. Examples
of the first class are the McFarlane Object Manipulation
Task, the Crawford Spatial Relations Test, and the O'Connor
Wiggly Block Test. One of the most frequently used measures

of this class is the Minnesota Form Board Test. Generally

12

Speaking, this type of test has been found to be cumbersome
to administer, requiring close supervision and much time.

In attempting to reduce the drawbacks implicit in
solid object manipulation, there have been numerous
adaptations of object manipulations tasks to paper and
pencil form. The advantages of this method are numerous:
reduced cost of testing, mass testing, increased reliabi-
lity, etc. A representative test of this type is the
Revised Minnesota Paper Form Board.

Another trend has been the construction of spatial
tests in paper and pencil form which were not based
originally on physical manipulation tasks. Probably the
most extensively used test of this type is the Space
Relations part of the Differential Aptitude Test.

The revised Minnesota Paper Form Board Test, used
as the measure of spatial ability in this research, was
chosen for its wide and general acceptance as a valid and
reliable measure of spatial ability. As described by
Anastasi (1968, p. 361); "The results indicate that it is
one of the most valid instruments for measuring the ability
to visualize and manipulate objects in space."

This paper and pencil adaptation of the original
Minnesota Form Board Test has been found to have a signifi-
cant relation to tests using physical object manipulation.
Paterson, gt 2;. (1930), Super (1950), and Harrell (1939),
report correlations of .63, .59, and .65 respectively.

Four factor analytic studies (Division of Occupational

13

Analysis, 1945; Andrew, 1937; Harrell, 1939: Wittenborn,
1945) have found that the Paper Form Board had high
loadings on spatial visualization factors. Tinker (1944)
found that the Paper Form Board correlated .40 with general
intelligence. A similar correlation of .40 was obtained by
Murphy when he compared the results of the Paper Form Board
with a test of mechanical comprehension. Significant
validity coefficients for the Form Board have been reported
by Shellow (1926) relating to success of mechanical
inspectors, machine operators, job setters, tool maker
apprentices, dental students, sewing machine operators and
pressmen.

The test itself is available in two forms (AA and
BB). Alternate form reliabilities are reported in the
manual to be in the high .80's. The test is administered
in a twenty minute period and requires the subject to
visualize how two or more pieces fit together. The subject
is then required to select the correct completed figures
from several choices. It might be conjectured that the
task required by this test involves facility corresponding
to French's (1951), 8, SO, and Vi factors. Michael's
(1957) SR-O and Vz factors, which are highly similar to
French's factors, would also appear to be components

inherent in these spatial tasks.

14

Differential Test Results

Sex Differences

 

Summarizing a large body of literature dealing with
gender and its relation to tests of spatial ability,

Schmidt (1971) points out that, on the average, males score
significantly better on tests of spatial ability than do
females: "There are no reports of cultures in which females
are equal to males in this ability. It may be universally
true that men excel women in this ability" (p. 4).

Tyler (1965) makes the observation that this male-
female difference is not restricted to any particular
developmental period but rather is evidenced from preschool
levels to adulthood. Male superiority in spatial ability
tasks has not been limited to Western cultures and has been
shown to be cross-cultural (Witkin, 1962). Research results
have consistently demonstrated that, on the average, only
20 per cent of females do as well or better than the average
male (Schmidt, 1971). This percentage also has been shown
to be manifest regardless of social or economic class
(Havinghurst and Breese, 1947).

The pervading male superiority in spatial ability
has led Garai and Scheinfeld (1967, p. 202) to postulate
that part of the differences may be due to alternate
approaches to solving spatial problems:

The female seems to exhibit a more global approach in

the perception of spatial relations than the male
which makes her perceive the stimulus and its setting

15

or "peripheral field" as closely connected, whereas the
males tend to be more analytical in his perceptual
approach, concentrating his attention on the perceptual
stimuli itself, while showing a much greater ability
than the female to dissociate it from its visual
context or peripheral field.
Field dependence and field independence according to
Garai and Scheinfeld (1967) thus characterize the cognitive
styles of female and male approaches to solving spatial
problems.
Theorists are in opposition over the relative
importance of environmental and non-environmental factors
in the development of these cognitive styles and spatial

ability (Schmidt, 1971). This debate has stimulated

research into the effects of learning on spatial ability.
Improving Spatial Ability

Correlational Desigg_.--Several researchers have
attempted to examine the effects of learning on spatial
ability. These attempts have generally been of two kinds.
The first type involves the improvement of spatial ability
as a result of incidental learning situations. The
majority of the research conducted in this area has been of
this type. The typical design involves a pre and post test
of spatial ability with some form of intervening training.
The training involved is considered to have required a high
degree of spatial ability. Some forms that this training
have taken are: classes in architecture (Faubian, 33 31.,

1942) and engineering study (Churchill, $2.21., 1942);

16

Myer, 1958). The underlying assumption of these researches
is that the tasks which the subjects are required to
perform in their curriculum involve spatial ability to a
large degree. Since the subjects are required to make use
of this ability, they should become more facile at it.
Thus, when tested after a period of training, their spatial
ability test scores should have significantly improved from
their pre-training scores. In the research cited, no
increase in spatial ability scores was noted. Smith (1964,
p. 124), commenting on these investigations noted that,
"The general conclusion to be drawn from these studies is
that the gains in performance on a spatial test are likely
to be small if, in effect, they are measurable at all after
limited amounts of training."

Probably one of the most significant factors
affecting the lack of improvement can be attributed to the
relatively brief periods of training (usually a semester)
between pre and post test. As Schmidt (1971, p. 2) points
out: "An undeveloped ability . . . no matter what the
reasons for its lack of development . . . can not be used;
nor can it, in general, be quickly developed after years
of neglect."

An exception to the general finding that spatial
ability cannot be increased through incidental transfer of
learning is that of Blade and Watson (1955). Using five
groups of subjects (three experimental and two control)

Blade and Watson measured the change in spatial ability

17

over a period of four years of engineering study. All
entering freshman male engineering students at three
universities (104 Ss, 593 Ss, and 114 85) were administered
the CEEB Spatial Visualization test prior to registration.
Two control groups who received no training in engineering
(77 Ss and 124 83) were also pretested. A second adminis-
tration of the CEEB Spatial was conducted at the conclusion
of the freshman year. The third and last administration was
conducted after four years of engineering and non-
engineering study. Although experimental attrition lowered
the number of $5 the authors report no consistent relation-
ship between attrition and degree of spatial ability.

Analysis of test scores indicated that after one
year of college Ss who received engineering training showed
significant gains in spatial ability from their pretest
scores (.74 S.D., .80 S.D., and .75 at each of the three
colleges). The control groups did, on the average improve
from their pretest scores but the engineering group's
improvement was over three times as great.

Improvement in the engineering group was not
linear. After the first year of study, the engineering
group reached a plateau. After four years of study the
average improvement from the pretest scores did not differ
significantly from those attained after one year of study.

It would appear then that spatial ability is
amenable to improvement as the result of incidental

learning but this conclusion has serious reservations.

18

Several researchers, who employed short periods of training,
(Faubian, £5 21., 1942; Churchill, gt_gl., 1942; Myer,

1958) report no transfer effects. Blade and Watson, (1955)
on the other hand, do report significant increases. It
might be conjectured then that the deciding factor in

achieving transfer was the length of the training itself.

Experimental Designs.--Comprising a second group of
studies attempting to investigate the effects of learning
on spatial ability, experimental designs have been few.

Krumboltz and Christal (1960) noted that spatial
ability tests were particularly susceptible to practice
effects. Their research was designed to examine the
relationship between retest latency and retest form with
amount of transfer. Practice was defined as the adminis-
tration of a pretest of spatial ability. Retesting was
accomplished after either a ten minute or seven hour delay.
During this delay period, subjects completed a biographical
inventory. Retest measures were of three types. In the
first instance were subjects whose retest consisted of the
same test administered as the pretest. A second group of
subjects received as their retest a parallel form of the
pretest. The third type of retest was an alternate type
of spatial ability test. A fourth, control group received
only the pretest, Krumboltz and Christal (1960) conjectured
that the different types of retests would be differentially

susceptible to pretest practice effects and that retest

19

latency would affect the amount of practice effect. Five
hundred twelve male, Air Force, ROTC student officers were
used as subjects.

Analyzing pre and posttest change scores by analy-
sis of variance techniques, Krumboltz and Christal (1960)
found that the practice groups did improve their spatial
ability scores significantly more than the control group.
Retest forms (identical, parallel, and alternate type) also
proved to be significantly different. Alternate type of
retest showed no significant transfer of learning from the
pretest while parallel and identical retesting did. In
general, the two different retesting latencies (ten minutes
and seven hours) were not found to significantly affect
transfer.

Brinkmann (1966) similarly attempted to investigate
the effects of training and practice on spatial ability.
The subjects in this study were 54 male and female junior
high school students. Two different methods of training
were employed; programmed instruction and tactile experi-
ence. The first method employed the use of 505 geometric
patterns. The task of the subject was to visualize the
pattern and determine its shape after manipulation. These
patterns were developed in the form of frames and placed
in a programmed instruction text. Each frame was followed
by its correct solution. They were arranged so that they
were progressively more difficult. The second method of

instruction involved solid object manipulation and paper

20

folding tasks. All subjects were provided with a "Solid
Object Manipulation Kit" which contained a series of cubes,
rectangular solids, pyramids, cones, etc. The pattern
folding task involved the cutting out and folding of
patterns which, when folded, resembled geometric figures.
All subjects could work at their own pace within the
allotted class time. Practice sessions were held over a
period of three weeks with one fifty minute period per day.
Total training time was thus twelve and one-half hours.

The control group received no special instruction but
attended their regularly assigned classes in their normal
program. Two tests of spatial ability were administered to
all subjects. Prior to training all subjects were given
the Space Relation part of the Differential Aptitude Test.
Both groups were then assigned and matched for age, sex,
and educational level. At the conclusion of the research,
all subjects were given a posttest of spatial ability which
was an alternate form of the Space Relations part of the
Differential Aptitude Test.

Analysis of the results of this research indicated
that the experimental training groups had significantly
improved their posttest scores of spatial ability. The
control group showed no improvement. Both groups were
found to be initially equal on the pretest of spatial
ability. At the conclusion of the research, the mean
difference score for the experimental group was 18.18 while

the control group pre and posttest differed by an average

21

of only 3.18 points. The experimental group also showed
more variance than the control group, the standard devi-
ations being 14.6 and 10.8 respectively. Using a T test,
the mean differences were found to be significant at the
.01 level. Interestingly, while pretest measures showed
initial sex differences in spatial ability, no sex differ-
ences were found on the posttest measures. The females

in the experimental group had, on the average, significantly
increased their score more than the average male. Whether
the solid object manipulation or the programmed instruction
proved more effective was unclear since no separation of
effect was possible within the design.

More recently, Junttonen, £5 21. (1972), noted that,
while the average male was superior in spatial ability, the
average female was superior to males in verbal fluency.
Their research attempted to counterbalance these two
abilities.

Using 201 subjects, four groups of students were
tested on the Concealed Figures Test of Spatial Ability.
Three experimental groups were given forms containing
verbal labels associated with the spatial problem. These
labels were of three types; correct, incorrect, and non-
sense. It was hypothesized that (Junttonen, 1972, p. l)

" . . . correct verbal labels (would) facilitate females'
performance more than males'; and that incorrect labels

(would) hinder females' performance more than males'.”

22

Results showed that this was not the case. Labeling
of any kind facilitated performance and in all cases for
men more than for women.

In contrast with attempts to improve Spatial ability
through incidental transfer, all three of the experimental
designs, Krumboltz and Christal (1960), Brinkmann (1966),
and Junttonen, 32 31. (1972) did achieve positive results.
The deciding factor between these two groups of studies may
well be the degree of similarity between training require—
ments and test requirements. Commenting on the general
nature of transfer of learning effects found among varied
abilities and subjects, Deese and Hulse (1967, p. 357)
point out that: "We can say that two tasks can positively
transfer to one another to the extent that there is some
commonality between the outcomes required by the tasks"

(p. 357).

The effect on transfer of training to test is not
limited to spatial ability but is apparent on many differ-
ent types of perceptual and cognitive tasks involving both
human and infra-human subjects (Epstein, 1967).

In light of Blade and Watson's (1955) results it
might be theorized that task-test dissimilarity can be
compensated for with extended periods of training.

In summary, several factors appear to be involved
in improving spatial ability by transfer of learning. In
general, short term incidental training does not appear to

effect transfer. On the other hand, incidental training of

23

a much more extensive nature has been shown to induce
significant transfer effects. It should be noted that no
research involving incidental transfer effects have included
women in their samples.

Using an experimental design approach, spatial
ability has been shown to be highly amenable to transfer of
learning effects. Two research designs (Brinkmann, 1966
and Junttonnen, 1972) have used both male and female
subjects. The performance effects, when related to gender
differences, appears to be an unsettled question since
these two researchers present conflicting results. The
third experimental design (Krumboltz and Christal, 1960)
used only a male population and while they also found
positive transfer effects with practice, made no attempt to
address themselves to gender differences. Only one of the
experimental designs reviewed attempted to ascertain the
relative stability of transfer effects over a period of
time. Using a seven hour delay in retesting, Krumboltz and
Christal (1960) found equal transfer, in general, for short
(ten minutes) and long (seven hours) delays in retesting.

The general conclusion that was drawn by researchers
whose attempt to achieve positive transfer was successful
was that spatial ability could be improved. The relation-
ship between the change in spatial ability test scores and
the consequent improvement in the underlying spatial
ability is of importance. Measures of spatial ability have

typically shown relatively high construct validity. Studies

24

by Shellow (1926), Paterson, 2; El' (1930), Super (1950),
and Harrell (1939) report consistently high correlations
between spatial ability scores and tasks or occupations
requiring spatial ability. The weight of evidence would
thus indicate that there is relatively high degree of
isomorphism between the improvement of spatial ability test
scores and the improvement of the underlying spatial
ability.

In light of this discussion, several tentative
conclusions can be drawn. Practice has, under certain
conditions, been shown to significantly improve scores on
spatial ability tests. This effect may occur, for males,
with exposure to a pretest. Several different techniques
of training have been shown to significantly improve
spatial ability. It would also appear that the greater the
similarity between training task and test task, the more
pronounced the transfer effect will be. While the effects
of gender have been examined in relation to improvement,

conflicting results have been reported.

 

III JI' {III ‘

CHAPTER III

PROBLEM STATEMENT

Research, which has attempted to measure the effect
of training or practice on spatial ability, has occurred in
two forms--experimental practice and incidental learning.
Generally speaking, transfer of learning effects have been
more evident using experimental designs. The greater the
similarity between practice and training, the more likely
are positive transfer effects. It is plausible that
practice—test dissimilarity may be compensated for with
increased duration of training. The effect of training, as
it relates to differential improvement by sex, has typically
not been an area of investigation.

In light of these results, several questions are
left unanswered. While short term practice effects were
noted by Krumboltz and Christal (1960) for parallel and
identical retest situations, no effects were found for
alternate type retesting. Would more extensive training on
alternate type tests produce transfer?

Brinkmann (1966) and Junttonen (1972) report con-

flicting results related to the degree of transfer relative

25

26

to sex. This issue needs clarification. Relatively
speaking, no research has attempted to measure the stability
of transfer effects over an extended time. Does the
learning, affected in practice sessions, decay rapidly or
maintain stability?

It thus becomes the purpose of this study to attempt
to clarify and extend previous research done on the effects
of practice on spatial ability. Specifically, tentative
answers to three questions will be sought:

1. Can transfer of training effects be elicited by
practice on alternate types of spatial ability
tests?

2. Do males or females benefit more from training?

3. Is there any decay in transfer effects with long

periods of latency?

CHAPTER IV

METHODOLOGY

Subjects

The subjects for this research were 60 volunteers,
recruited from undergraduate introductory psychology
classes at Michigan State University. They were reimbursed
for participation in this research by research credits in
their respective classes. Thirty of these subjects were
male and thirty female. The experimental group consisted
of 30 subjects (15 male and 15 female) randomly assigned.
The control group was similarly structured. No other

restrictions are made.

Materials

 

All materials for this research consisted of various
tests of spatial ability. The pretest and posttest con-
sisted of alternate forms of the Revised Minnesota Paper
Form Board Test. Consisting of 64 items, each form (AA and
BB) was subsequently divided into two shorter tests of 32
items each. Odd-even items were assigned as parallel
forms. Thus, four parallel tests of spatial ability were

devised. (1) Form AA odd, (2) Form AA even, (3) Form BB

27

28

odd, and (4) Form BB even. Normal test time of 20 minutes
was halved for these forms to 10 minutes per test. All
other test procedures, as described in the manual were
adhered to. This test was chosen on the basis of large
number of items and its noted reliability and validity in
the literature. Tests used as training material were
chosen on the basis of their availability and relative
similarity in format. Although differing in the nature of
the task, all tests required the subjects to mentally re-
arrange a series of objects and identify the completed
figure from similar figures. These tests were:
1. Survey of Object Visualization. Copyright 1945,
California Test Bureau.
2. Survey of Space Relations Ability. Copyright 1944,
California Test Bureau.
3. Spatial Relations part of the Differential Aptitude
Test. Form M, Copyright 1962, The Psychological
Corp.
4. Multi-Aptitude Tests, Factor IV, Spatial Visuali-
zation. California Test Bureau, 1959.
5. The Dailey Vocational Test, Spatial Visualization
Test. Copyright 1964. Houghton Mifflin Co.
6. The Differential Aptitude Test, Space Relations

Part. Form L, Copyright 1962, The Psychological

Corp.

29

Pretest

All subjects (N = 60) convened at the same time as
a group. At this time they were all administered Form AA,
odd questions, of the Revised Minnesota Paper Form Board
Test. During the administration of the test, the subjects
were randomly assigned into two groups; the Experimental
(N = 30) and the Control (N = 30). Each group was composed
of fifteen males and fifteen females. At the conclusion of
the pretest situation, those assigned to the experimental
situation were asked to remain in their seats while the
control group was asked to report one week hence at a

specific time and place.

All those assigned to the experimental group after
the pretest remained seated while the control group was
dismissed. At this time they were read Protocol One
(Appendix) which informed them that they were to take part
in an experiment whose purpose was to improve their spatial
ability. To accomplish this end they were asked to partici-
pate in a series of practice tests of spatial ability.

This practice was conducted in two sessions; that evening
and the following night. One week from that date they took
another test of spatial ability to measure their improve-
ment over time. The practice tests were not subject to the
time constraints imposed by the test manual. Recommended

testing time was doubled. Thus a test whose administration

30

time was 30 minutes was actually administered for 60
minutes. Subjects were also provided with scratch paper
to assist them in solving the problems. At the end of each
practice test the correct answers were given. Subjects
were given a brief period to review, either by themselves
or with others, problems which presented difficulty.

All subjects in the experimental group then
proceeded to take the following spatial ability tests:

1. Survey of Space Relations Ability (40 minutes
practice time).

2. Differential Aptitude Test Form M (40 minutes
practice time).

3. Dailey Vocational Test (40 minutes practice time).

Total practice time for training session one was
2 hours. At the completion of the practice session subjects
were given form BB Even of the Revised Minnesota Paper
Form Board Test.

Practice Session--Night Two:

At the appointed time and place the following
evening the experimental group reconvened. The procedure
for the second practice session was essentially identical
to that of the first session. Protocol Two (Appendix) was
read to the subjects and the Form AA Even of the Revised
Minnesota Paper Form Board Test was administered prior to
practice. Practice tests' administration time was again
double the recommended administration time. The following

tests were then presented for practice:

31

1. Differential Aptitude Test Form L (40 minutes
practice time).

2. Multi-Aptitude Test--Factor IV (40 minutes practice
time).

3. Survey of Object Visualization (40 minutes practice
time).

It should be noted that, while actual practice
lasted a total of four hours, total training session time
was six hours. The two hour difference was filled by
protocol presentation, feedback sessions, material distri-
bution and collection in addition to frequent (3 ten minute)

breaks each session.

Posttest

All subjects reconvened at the appointed time and
place for the posttest. No subjects who received training
failed to return. The lack of experimental attrition might
be explained by the withholding of class credit assignation
until the completion of the entire research. The control
groups, on the other hand, who received much less class
credit for the research did show some attrition. Of the
39 control subjects who took the pretest, 33 also were
present at the posttest. Three control subjects and four
experimental subjects were randomly dropped from the experi-
mental to equate subject size for each experimental

condition.

32

At the conclusion of the posttest (Revised Minne-
sota Paper Form Board Form BB Odd) all protocols were
collected, subject cards signed and subjects dismissed.
Feedback, if requested, was presented after preliminary

data analysis, one week after the posttest.

CHAPTER V

RESULTS

The data collected for this research consisted of
error scores derived from four administrations of parallel
forms of the Revised Minnesota Paper Form Board. Inter-
spersed among these administrations were various amounts of
training and latency. As illustrated in Figure 1, Analysis
Design, test one was administered to all subjects prior to
any training. Test two was administered to the experimental
group immediately after 2 hours of practice. Test three,
also administered to the experimental group, was conducted
immediately prior to the beginning of the second training
session. Test four, administered to all subjects, was
conducted one week following the initial pretest (Test one).
In this manner, differences between groups were obtained
between:

1. Test one and test two (2 hours of practice).
2. Test one and test three (2 hours of practice and

21 hours latency).

3. Test one and test four (4 hours of practice, one

week latency).

33

34

4. Test two and test three (21 hours latency).

5. Test two and test four (21 hours latency, 2 hours
training, 6 days latency).

6. Test three and test four (2 hours training, 6 days
latency).

In all cases, change scores were derived by sub-
tracting each individual's post score from his pre score.
Three types of statistical techniques were used to analyze
the data resultant from the comparisons listed above;
analysis of variance, student t tests and correlational
analysis. The factors, levels, nature, degrees of freedom,
mean squares, and error terms for the analysis of variance
are presented in Table l. The nature of the data collected,
and the manner in which it was analyzed, were directed at
answering those questions raised in the problem section and
determining the veracity of the hypothesis presented in
Chapter IV.

Figure l portrays the temporal sequence of events
as they occurred during the experiment. To recapitulate,
all subjects were pretested together (Test one) and then
randomly assigned to either the experimental or control
groups. At this point, the control group was dismissed
and instructed as to the date, time, and place of the post-
test. After the dismissal of the control group, the
experimental group received two hours of practice followed
immediately by Test two. These subjects were then

released. Twenty-one hours after the initial training

35

”:50:

go.—

 

High.

 

 

__-——— _— ——~ —'~D — _ ~— —

.cmwmmo mammasc¢ .H seamen

H455:

 

36

session, the experimental group reconvenied and were
administered Test Three. The second training session then
ensued after which the subjects were released. Six days
following this second training session, all subjects in
both the experimental and control groups were administered
Test Four, the posttest.

Figure 2 depicts the average number of errors made
by each group on each test administered to that group. The
data upon which this figure was based can be found in
Table 10. It would appear from inspection of this figure
that groups differed in average number of errors when
related to test. The relative stability of errors by group
also appears to show some intergroup differences.

Figure 3 portrays the average decrease in error
score by group. Inspection of the figure reveals that, at
least in terms of ordinality, there are differences between
groups in terms of the degree with which each group
decreased their error scores from the pretest to the post-
test situation; the experimental males showing the greatest
reduction in errors followed by the experimental females,
the control males, and the control females.

Figure 4 shows how, on the average, change scores
were a function of gender. The degree of parallelism
between the experimental and control groups, both in this
and Figure 5, reflect the absence of interaction effects
evidenced in statistical tests. It can be seen from

inspection of this figure that the experimental and control

 

lllllill|

AVERAGE ERRORS

AVERAGE DECREASE IN ERRORS

P‘P‘

HHNNwwl-‘J-‘UIU‘ O‘O‘VNCDCDWOH

0
CU:-
p

l QIOIU‘OWOWLOLUW ‘O1UWOU7Q WOLUTP

 

37

 

 

 

Figure 2.

.méwo

.,..».°

OU‘

”.3191 ALL

0

O OH HNN DOW-{>111 0“-

DUI.

I
|-‘ I

| I 1
Test One Test Two Test Three

Average Error Score by Test.

\

_‘
Test Four

 

Figure 3.

I F I
EXPERIMENTAL EXPERIMENTAL CONTROL
MALES FEMALES MALES

Average Decrease in Error Score by Group.

1.
CONTROL
FEMALES

 

 

6.0-

5.5-

5.0-1
i”.
o 4.5 -‘
E
m 4.0 '“ EXPERIMENTAL
.5. 3.5 '-
0 ~
g 3.0
g 2.5"
D 2L0 1
0 A
m 1.5
‘3.
m 1.0"
>
S 0.5 -

0.0 - CONTROL

- 0.5 d
l r
MALES FEMALES

Figure 4. Comparison of Average Decrease in Errors by Gender.

groups differ in respect to their average decrease in
error scores from the pre to posttest situations.

Figure 5 similarly reflects group differences in
that the males, as a group, appear to have a differential
reduction in errors from pre to posttest when compared with
females as a group.

Comparing Figures 4 and 5, the degree of separation
between the lines appears to be more pronounced for group
effects than it is for gender effects. The relative
magnitude of this separation is reflected in the data
analysis; training effects were consistently significant
whereas gender effects were significant only when corrected

for composite error effects.

6.0-1
5.5“
5.0-
4.5 e
U)
'5 4.0..
ii
w 3.5 -
-5 3.0-
§ 2.5-
3 2.0.. MALES
U
8 1.5—4
Q)
g 1.0 1
H
E 0.5-
0.0- FEMALES
0.5

 

 

I I

EXPER IMEN'I'AL CONTROL

Figure 5. Comparison of Average Decrease in Errors by Group.

Table 1 is an expression of the significant aspects
and characteristics of the experimental design. Table l

was developed through the use of SAAVED (Frankmann, 1968).

Table 2 is an analysis of variance design based on
change scores from Test One to Test Four. Each subject's
posttest error score was subtracted from his pretest error
score to derive a difference or change score, the dependent
variable. Of the two main factors, it can be seen that
group (training vs. no training) achieved significance
while gender effects (male vs. female) did not. Interaction
effects, similarly, were also not significant. Eta squared
shows that group factor variance accounted for over 15 per

cent of the total variance while gender effects accounted

40

 

 

 

 

 

Table 1. Factors, Levels, Nature, Degrees of Freedom, Mean
Squares, and Error Terms.
Factors Levels Nature
A Group 2 Fixed
B Gender 2 Fixed
S Subjects 15 Random
Source df E(MS) MSerror df(MS)e
A 1 B0 02 S/AB 56
S
2
B 1 As US S/AB 56
2
S/AB 56 OS . . .
2 2
AB 1 CA S §_/AB 56
Total 59

 

Table 2.--Ana1ysis of Variance Using Uncorrected Differ-
ence Scores Based on Tests One and Four.

 

 

Source SS df MS F P Eta2
Group 198.02 1 198.02 10.74 .002 .154
Gender 50.41 1 50.41 2.73 .104 .039
Group X 2.82 1 2.82 .15 .697 .002
Gender
18.45 .80

Error 1032.93 56

Total 1284.18 59

 

41

for only 4 per cent of the total variance. Interaction
effects accounted for only .2 per cent of the total vari-
ance. It might be concluded from this table that, on the
average, the effect of being trained was significantly
different from the effect of not being trained.

Table 3, similar to Table 2, is based upon the
difference or change scores from Test One to Test Four. It
was recognized that this dependent variable, change score,
was essentially a composite measure based upon two adminis-
trations of parallel forms of the Revised Minnesota Paper
Form Board Test. A composite score, as such, is subject
to the combined error variance associated with each
administration of the pre and posttest.

Table 3.--Analysis of Variance Using Corrected Difference
Scores Based on Tests One and Four.

 

 

Source SS df MS F P Eta2
Group 206.94 1 206.94 21.47 .0005 .260
Gender 46.34 1 46.34 4.81 .033 .058
Group X 1.93 l 1.93 .20 .66 .002
Gender

.679
Error 539.86 56 9.64

Total 795.08 59

 

42

To correct for the increased error incurred by
combining pre and post measures into a composite score,
each subject's estimated true change score was obtained by
the formula:

C = rdd(c) + C
Where:
C = an individual's estimated true change score.
rdd = the reliability of the difference score measures.
c = the individual's obtained score in mean deviate
form.
C = the mean change score for the individual's group.
The reliability of the difference score measures,

rdd' was estimated by the formula (after McNemar, 1958):

 

2 2
r = r1101 + r220 — 2r120102
dd 05 + o5 - 2 o o
1 2 r1212

Subscripts 1 and 2 refer to pre and posttests
respectively. The internal reliability of these measures
(rll and r22) was estimated from the Kuder—Richardson
Formula #20. K-R #20 for the split-half Revised Minnesota
Paper Form Board was found to be fairly high, .84. The
Pearson product moment correlation between Tests One and
Four for each group was computed (see Table 13). These
correlations, the estimated internal reliability and the
respective variances and standard deviations for each group

were substituted in the above formula. In this manner,

43

the reliability of the differences scores for each group
was determined and each subject's obtained change score
corrected to obtain an estimate of his true change score.
These adjusted change scores were then analyzed and the
results presented in Table 3.

Examination of Table 3 reveals that both group and
gender effects are significant. Since Table 3, like Table
2, is based upon a comparison of Test One and Test Four
scores, the emergence of gender as a significant effect
might be attributed to this adjustment.

Table 4 presents an analysis of variance design
relating to the overall effect of training and gender for
the experimental group. As was the case in the previous
analyses, gender has two levels; males and females. The
variable measure, which refers to test administration, has
four levels; Test One, Test Two, Test Three, and Test Four.
The analysis thus is a 2 x 4 design. Inspection of the
table reveals that neither gender nor interaction were
significant. Measure effects, in this repeated measures
design, were significant; the tests scores were not
homogeneous. These measures or test differences are
portrayed in Figure 2.

Table 5 is based upon the Newman-Keuls Test for
repeated measures of simple main effects. The technique
and procedure, as described by Winer (1962, p. 309) are
applicable to a repeated measures design for post-hoc

comparisons of simple main effects. Since no prediction

44

Table 4.--Analysis of Variance Comparing Practice and
Gender Effects.

 

 

Source SS df MS F P
Between

Subjects 1288.34 29 . . . .
Gender 75.21 1 75.21 1.74 n.s.
Errorb 1213.13 28 43.33 . . . .
Within

Subjects 3979.25 90 . . . . .
Measure 351.69 3 117.23 2.73 .05
G X M 13.77 3 4.59 .11 n.s.
Errorw 3613.79 84 43.02 . . . .
Total 5267.59 119 . . . . . .

 

Table S.--Newman-Keuls Test for Measure Effects.

 

 

Level TA - TB r q P
T3 - T4 1.367 2 3.372 n.s.
T2 - T4 2.233 3 4.056 n.s.
T1 - T4 4.70 4 4.452 .05
T2 - T3 .866 2 3.372 n.s.
Tl - T3 3.333 3 4.056 n.s.
T - T 2.467 2 3.372 n.s.

 

45

had been made prior to research relating to the effects of
each of the two training sessions, a post-hoc comparison
was conducted. Inspection of the table reveals that none
of the test mean comparison, with the exception of Test One
verses Test Four, is significant. The general formula used
in this comparison is:

MS
error

TA - TB = q(r,f) n x 2

Where:
TA and TB = the obtained mean difference between any
two given tests.
q = the obtained studentized range statistic
for r steps and f degree of freedom.
n = the number of subjects in each group.
MS = the within groups error variance divided

error

by the appropriate degrees of freedom.
Tables 6, 7, 8, and 9 are similar in that they

consist of student t tests based upon independent group
means among the four groups. Table 6 compares the change
scores (Test One minus Test Four) for the experimental and
control males. The results, based upon significance levels
for a one-tail t test, show these two groups to be signifi-
cantly different from each other. It could conceivably be
concluded from this table that the experimental difference

between the experimental and control males had a significant

46

Table 6.--Comparison of Change Scores for the Male Experi-
mental and Control Groups.

 

 

Group N R o t df P
Experimental 15 5.4 1.13
1.78 28 .05
Control 15 2.2 .46

 

effect; training significantly effected spatial ability
test scores as opposed to not training.

Table 7, similarly, compares the change scores
between experimental and control groups but in terms of
females. The mean difference in change scores between
these two groups was found to be significant for one-tail
probability levels. The implication of this findings, as
was the case for the males, is that training, compared with
the lack of training, could have had an effect on change
scores for the two groups.

Table 7.--Comparison of Change Scores for the Female
Experimental and Control Groups.

 

 

Group N R o t df P
Experimental 15 4.0 .84
2.78 28 .005
Control 15 0.27 .02

 

Comparing the results of Table 9 and 10 with the
analysis of variance presented in Table 2, it can be seen

that they are compatible; training can be postulated as

47

contributing significantly to variation in change scores.
Since both males and females appear to have benefited from
training (Tables 6 and 7), gender effects tended to be
minimized.

Tables 8 and 9 which examine the effects of gender
as it relates to training, show this lack of gender effects.
Table 8, which compares the average change score for the
male and female control groups, was related to significance
levels for a two tail test. Unlike Tables 6, 7, and 9,
directionality was not predicted for the control group
since previous research had indicated an absence of change
in either direction for control groups. Table 8 shows that
the average change scores for males and females, who were
not formally trained, were not significantly different from
each other. Being male or female evidently had little
effect on change scores from the pre to posttest.

Table 8.--Comparison of Change Scores for the Control Male
and Female Groups.

 

 

Groups N R o t df p
Males 15 2.2 .519

1.80 28 n.s.
Females 15 -.07 -.016

 

Table 9 similarly compares males and females but
under the experimental or trained condition. The results

of this comparison, like those found under untrained

48

Table 9.-—Comparison of Change Scores for the Experimental
Male and Female Groups.

 

 

Groups N X o t df P
Males 15 5.4 1.11

.280 28 n.s.
Females 15 4.0 .726

 

conditions, show a lack of significant differences by
gender.

In light of the results evidenced in Tables 6, 7,
8, and 9, it might be stated that:

1. Training appears to have had a significant effect
on both males and females when contrasted to un-
trained males and females.

2. The effects of gender were not significant either
under the training or no training conditions; males
and females tended to benefit equally.

These results are consistent with Table 2 which
show that training had a significant effect on change
scores as compared with no training. Gender effects in
Table 2 approached but did not attain significance. Table
3, compared with these findings in Tables 9 through 12
shows that gender effects present may attain significance
when the effects of test unreliability are partialed out.

Table 10 presents a summary of some statistics

relevant to and incorporated in the analysis presented.

49

 

 

mm. so.m no.6 v6.m 04.0 oe.q mm.s ms.¢ mm.m moumum>e
.mm. m.m N.s m.v me.» mmemsmm
Hoauaoo

mm. m.q om.o m.¢ me.m moan:
Houucoo

om. sm.m am.m Hm.m v.n va.m sm.s m.m o~.oe mmflmsom
HmucmEHmexm

me. o.m m.m om.m v.m mq.v m.o mm.v om.m mmemz
Hmucmefluomxm

e a H mummy v umme v umme m umme m umme m umme m umme H umme H umoe
cmo3umm u o m o m o m o m

 

 

 

 

 

 

.msouo was umme ma mcoflumamnuou

pas .mcoHumH>mo pumpcsum .mmnoom Houum mmmum>4lu.oa manna

50

Statistics are presented marginally for each of the four
groups and for each test administered to that group.

Table 11 exhibits the average change scores from
pre to posttest for all groups. Average change scores are
presented in both unit change (number of errors) and
standard deviation units of change. Standard deviations of
change were derived for each group by dividing the average
unit change score with the standard deviation of that

group's pretest:

Table ll.--Average Change in Error Score From Test One to
Test Four.

 

 

 

 

Males Females
R o R o
i = 4.70
Experimental 5.40 1.11 3.99 .726
o = .907
SE = 1.06
Control 2.23 .519 -.07 -.016
o = .241
3.8 .833 1.96 .380

 

Table 12 was designed to test the significance of
the pretest-posttest correlation (r14). Initially all
obtained pre—post correlations were transformed to Fisher's

Z'. Assuming no intervening effects between pre and

51

 

 

 

 

Table 12.—-Z' Transformations of r14 and Standard Deviations
From r .
tt
Males Females
Z' 2 Z' 2
Z' = .388
Experimental .192 3.57 .549 2.33
z = 4.31
2' = .848
Control .590 2.19 1.38 .54
z = 1.93
.400 4.25 .793 2.21

 

posttest, r can be estimated by r The reliability of

14

the Revised Minnesota Paper Form Board, estimated from the

tt°

pretest was .84 by K-R #20. Converted to Fisher's Z', an
rtt of .84 was found to be 1.221. All transformed corre-
lations (2') were then converted to standard score form and
tested for significance of differences (Following Edwards,
1960, p. 81). Results showed that all pre-post correlations
were significantly different from that which would be
expected if there were no experimental intervening effects
with the exception of the control group as a whole and the
control female group. The implications of this finding are
that the obtained pre-post correlations were significantly
different from chance fluctuations of the expected pre-post
correlations. It would appear that the relationships

between subjects on the pretest changed significantly from

those on the posttest with the exceptions noted above.

52

Table 13 shows the adjusted correlation between the
pretest and change scores for each of the four groups.
Initial-change correlation analysis was included to
ascertain treatment by subject effect. The initial-change
correlation provides a description of the relationship
between pretest spatial ability level and the amount of
relative change or improvement as measured by decreases in

error SCOI‘B.

Table 13.--Adjusted Correlation Between Initial and Change

 

 

Score.
Male Female
Experimental -.883 -.863 -.865
Control -.344 -.389 -.336
-.636 0.735

 

As McNemar (1969, p. 177) points out, the general
formula for the correlation of change with initial scores
does not take into account the attenuating effects of
measurement reliability. The general formula for the
correlation of change with initial score can be adjusted
for measurement reliability by use of the formula:

I. _ rifsf ' rxxsi
2 2 2

S + rxxsi - 2rifrxxsisf

 

53

Where:

r. = the adjusted initial-change correlation.

rxx = calculated instrument reliability used as an
estimate of rif under the condition of no
experimental effects.

Using this adjusted formula, the correlation between
initial and change scores for each of the four groups was
derived. Examination of the table reveals that while all
groups show a negative correlation between initial and
change score, there appears to be a great deal on group
differences with the experimental group as a whole showing
a higher negative correlation than the control group and

the females as a whole showing a slightly higher negative

correlation than the males.

CHAPTER VI

DISCUSSION

This experiment was based upon previous research
relating change of spatial ability test scores to practice
or training involving spatial ability tasks, Reiterating
the discussions in Section II, research has shown that
transfer of training involving spatial ability can occur
with: (1) long term (1 year) incidental training, or (2)
short term (test-retest) practice on parallel or alternate
forms of spatial ability tests. Previous research has
failed to show (or shown mixed results) that; (1) spatial
ability test scores can be improved by short term practice
on alternate types of spatial ability tests, and (2) changes
in spatial ability are relatively stable over short periods
of time, or (3) the effects of practice as it relates to
gender.

This research then, attempted to resolve these
three problems, namely; can positive transfer occur with
practice on alternate types of spatial ability tests? How

stable are transfer results over a period of non-training?

54

55

Is gender a significant factor in determining the amount of
positive transfer?

The goal of this research was then to supply ten-
tative answers to these problems. In the form of analysis
of variance tables, t tests, and correlational analysis,
the two independent variables, gender and training were
related to the dependent variable, test score. In several
instances, the dependent variable was derived by sub—
tracting scores on one test from scores on another test
(i.e., "change" scores). In other instances, the effects
of gender and practice were derived by evaluating the change
in error score on one test compared with another test.

Analysis of results pertaining to the first area of
interest, that of the relative effects of training, reveals
that, in general, training in the form of practice on
alternate types of spatial ability tests, does promote
positive transfer of learning. The lack of training
apparently had little overall effect.

As Table 11 portrays, experimental (trained) and
control (not trained) groups vary greatly in the average
decrease in numbers of errors from pre to posttest. The
experimental group improved almost one full standard
deviation (.907) from their pretest scores. The control
group improved .241 s.d. from their pretest scores with the
majority of the change coming from the control males.

The conclusion that training significantly affects

spatial ability is born out by results found in Tables 2,

56

3, and 4. Table 5 is addressed to the question of training
components effect on total transfer. Tables 2, 3, and 4,
show that training appears to have produced positive changes
in test scores. These tables differ in the groups compared
and correction effects.

Examination of Table 5 shows that, while the over-
all effect of training was significant, the components and
combinations of training varied as to their effect. The
table attests that, by themselves, neither specific training
sessions nor latency periods were significant. The overall
cumulative effects of these sessions, on the other hand,

did reach a significant level. It might be conjectured that
the two training sessions had an interactive effect. The
first latency period of 12 hours showed no significant
effects. Although no test was administered immediately
after the second training session, it might be feasible to
assume that the second latency period, which lasted six
days, would show little, if any transfer effects.

The second general question posed, that of differ-
ential transfer effects of gender, may not be answered as
definitively as the first area of interest. Under no
conditions, with the exception of Table 3, as described in
Chapter V, was based upon change scores corrected for
increased error variance manifest in deriving a composite
measure.

Table 3 shows that gender effects were significant;

males, on the average, had had a greater decrease in error

57

scores from the pre to posttest compared with females.
Contrasted with the conclusion derived from Table 3,

Table 9 shows that there was no statistically significant
difference between the change scores for the experimental
males and females; both benefited from training (Tables 6
and 7) and they tended to benefit equally (Table 9). A
similar effect was noted for the control group. While the
control group as a whole tended to show little pre-post
change, this lack of change was consistent for both the
untrained males and untrained females (Table 8).

The apparent contradiction between Table 3, which
shows gender effects, and tables which fail to show gender
effects, may be more of an artifact than a conflict of
results. Examination of Table 10 shows that the average
posttest error score for the experimental males was only
3.8 errors out of a maximum 32 errors. The standard
deviation of this group on the posttest was also the
smallest for any of the four groups, 2.0. Contrasted with
the experimental females, the control males, and the control
females, it might well be that ceiling effects resulted in
restriction of range for the experimental males. The effect
of this restriction might have limited trained males in the
sense that their "true" potential for improvement was not
realized during the course of the training. Thus, the
between groups variance for gender effects would have been
artifically low. Similar results of this restriction in

range could be evidenced in the low (.19) pre-post

58

correlation for the experimental males, inflated signifi-
cance levels for z scores found in Table 12, and a high
initial-change correlations in Table 13.

The conclusions to be drawn in describing gender
differences in improvement are therefore tentative at best.
Data analysis would indicate that the sex of the subject
had little effect on the transfer of learning incurred.

The effects of restriction in range for the experimental
males may well have resulted in the lack of gender differ-
ences. This lack of gender effects might then be con-
sidered more of an artifact than "true" homogeneity.

The third area of investigation, that of the
relative stability of transfer effects, might also be
subject to inconclusive results.

The only strict test of latency effects is re-
flected in Table 5. A section of Table 5 compares the
scores on Test Two with the test scores on Test Three.
Intervening between these two administrations was a 21 hour
latency period. Neither gender nor measure effects were
noted. The effects of the second period of latency, that
occurring between the conclusion of the second training
period and the posttest was not distinctly measured.

Table 5 also shows the effect of the experimental conditions
and gender effects between Test Three and Test Four. Since
Test Three occurred just prior to the second training
period, the effect of the six day latency between the

termination of the second training session and the posttest

59

is confounded with the effect of the second training
session. Since the first 12 hour latency period showed no
effects, it is probable that the second, 6 day latency,
would also show little effect. It might be concluded
therefore, that the effects of training have not been shown
to diminish with short latency periods, and in all probabi-
lity, would not have been shown to diminish over a longer
period. A fourth series of analyses, about which no pre-
dictions were made, examined the effect of treatment by
subjects. The ceiling effects, previously mentioned in
relation to the general absence of gender differences, also
placed severe limitations on the definition of training
effects on subjects within groups.

The pre-post product moment correlation (Table 10)
indicated that groups varied as to their respective pretest
and posttest relationships. The experimental males
evidenced the lowest correlation of all groups, r = +.l9.
This low correlation may have been a result of the re-
striction in range for that group. The control females,
whose mean error score remained stable from pre to posttest,
had the highest pre-post correlation, r = +.88. The
relationships between the eXperimental females changes
considerably from pre to posttest (r = +.50) as did those
for the control males (r = +.53).

Testing the significance of these changes by the
process described in Chapter V, it was found that, with the

exception of the control group as a whole and the female

60

control group, all groups had significantly different pre-
post correlations than would have been expected by chance
factors alone. Contamination of these results again may be
the product of ceiling effects for the experimental males.
It may therefore be concluded that only the experimental
females and the control males were significantly affected
by the research conditions in terms of their relationships
with the members of their groups. A further confounding
element to interpretation of these results may be found in
the regression phenomena which could have conceivably
accounted for much of the evidenced correlation differ-
ences. Interpretation of these treatment by subjects
effects is therefore suspended in light of the many con-
taminating elements.

In addition to examining the effects on the
relationships among subjects from pretest to posttest,
Table 13 portrays the adjusted correlation between initial
error score and change score. Ceiling effects, similar to
those in the previous discussion, are also pertinent in
description of this table. The initial-change correlation
for the experimental males conceivably was unduly inflated
by the restriction in range incurred by ceiling effects on
the posttest.

In summary, several statements can be made in
relation to the areas of conjecture presented in Chapter
III. Training on alternate types of spatial ability tests

appears to induce significant transfer of learning. While,

61

overall, the effects of training were significant, the
individual sessions were not. The effects of training thus
appear to be cumulative. Considering the effects of gender
in relation to training, serious consideration must be
given to the role of ceiling effects and subsequent
restriction of range as a limiting factor in the emergence
of gender differences. Within the confines of this design,
strict determination of stability of transfer effects was
not possible but, at least for short periods, transfer

effects appear to remain stable.

CHAPTER VII

CONCLUSIONS AND IMPLICATIONS

In the broadest sense, this research has attempted
to ascertain the degree of transfer of learning as it
relates to training and gender. The effects of latency
were also, to a degree, examined. In assessing the results
obtained in this research, several problems may be pointed
out.

As was mentioned in the previous section, the
discussion of improvement by gender and by subject, was
hampered to the extent that ceiling effects contaminated
many of the analyses. Thus no definitive determination can
be made in several important areas. The possibility of
artifacts in data may be countered by using a spatial
ability test requiring more power than speed ability.

One of the more severe limitations placed upon
interpretation of the results was an inherent deficiency
within the design itself. The absence of a test after the
second training period precluded a discussion of the
differential effects of that training session in comparison

with the previous session and restricted the discussion of

62

63

long latency effects to conjecture. The use of a fifth
"marker" test immediately following the second training
session would provide a more conclusive statement.

A third area, about which no determination can be
made, might be the effects of fatigue on subject responses.
A frequent complaint of the subjects during the practice
sessions, was that fatigue was a limiting factor in their
powers of concentration. The effects of this, if any, on
transfer are unknown. Although frequent breaks were given
subjects it might be advantageous to spread out training
sessions so that practice would be more distributed than
massed.

A question related to transfer but left unanswered
in this design, concerns the stability of transfer effects
over extended periods of time. Although it was found that
the effects were stable over a 21 hour latency and it was
conjectured that they would also be stable over a 6 day
latency, no prediction could be evidenced for longer
periods of time. In light of the considerable practical
importance of spatial ability, this question might well be
worth investigating.

As a final implication for future research, differ—
ent types of training procedures might be attempted to
ascertain the most effective and efficient transfer
methodology. Considering the practical importance of
spatial ability, this could result in more than passing

benefit.

REFERENCES

REFERENCES

Anastasi, A. Psychological Testing. New York: Macmillan,
1968.

Andrew, D. M. An analysis of the Minnesota Vocational Test
for clerical workers. Journal of Applied Psycho-
logy, 1937, 21, pp. 139-172.

Angell, J. R., 22 El‘ Report of the committee of the
American Psychological Assocation on the standardi-
zing of procedure in experimental tests. Psycho-
lggical Monographs, 1910, $1! No. 1.

Army Air Force. In F. S. Freeman. Theory and Practice of
P3 chological Testing. New York: Holt 5 Co.,
1950

Binet, A., and Simon T. In A. J. Edwards. Individual

Mental Testin Part I. Scranton, Pa.: Intext
Educational PEEIisEers, 1971, pp. 9-51.

Blade, M. F., and Watson, W. S. Increase in spatial
visualization test scores during engineering study.
Psychological Monographs, 1955, 62, No. 397.

Brinkman, E. H. Programmed instruction as a technique for
improving spatial visualization. Journal of
Applied Psychology, 1966, 52, 2, pp. 179-184.

Burt, C. The Factors of the Mind. London: U. of London

 

Press, I940.
Burt, C. The structure of the mind. British Journal of
Educational Psychology, 1949, __J pp. - .

 

Churchill, B. D., Curtis, J. M., Coomb,s C. H., and
Hassell, T. W. Effect of engineering school
training on the surface development test. Edu-
cational and Psycholggical Measurements, 19127‘2,

557'2795750.

64

65

Cattell, R. B. Abilities: Their Structure Growth and

Action. Boston: Houghton MIffIin, Inc., 1971.

Curtis, J. W. Cross Reference Test. Chicago: Psycho-
metric Affiliates, 1959.

Deese, J., and Hulse, S. H. The P5 cholo of Learning.
New York: McGraw-Hill, 1967, p. 357.
Division of Occupational Analysis Staff. Factor analysis of

occupational aptitude tests. Educational and Ps -
chological Measurements, 1945, 5, pp.Rl47-I55.

 

 

Edwards, A. L. Experimental Design in Psychological
Research. New YorRE Holt, RineHart, Winston,

p. 81.

 

Edwards, A. J. Individual Mental Testing: Part I History
and Theories. Scranton, Pa.: Intext Edficational
Publishers, 1971, p. 115.

 

E1 Koussey, A. A. The visual perception of space. British
Journal of Psychology, Mono. supp. #20, 1935.

Epstein, W. Varieties of Perceptual Learnin . New York:
McGraw-HiII, 1967, pp. 111-112.

Faubian, R. W., Cleveland, E. A., and Hassell, T. W. The
influence of training on mechanical aptitude test
scores. Educational and Psychological Measurements,
1942, 3, pp. 91-94.

Fernald, M. R. The diagnosis of mental imagery. Psycho-
lggical Monographs, 1912, 11, pp. 1-169.

Frances, R. The improvement of perceptual discrimination.
In Epstein, W. E., gp. cit., 1963, pp. 104-105.

French, J. W. The description of aptitude and achievement
tests in terms of rotated factors. Psychometric

Monographs, No. 5, 1951.

Fruchter, B. Measurement of spatial abilities. Edu-
cational and Psychological Measurements, I954,'11,

2' pp. 557-5950

Garai, J. E., and Schinfeld, A. Sex differences in mental
and behavioral traits. Genetic Psychologngono-

graphs, 1968, 11, pp. 169:599.

66

Guilford, J. P. The Nature of Human Intelligence. New
York: McGraw-Hill, 1967.

Guilford, J. P. Printed Classification TestngA F Aviation
Psycholog1ca1 Pro ram. ResearcHRRept. #5, Washing-
ton: Government Printing Office, 1947.

Hanawalt, N. G. The effect of practice upon the perception
of simple designs masked by more complex figures.
Journal of Experimental Psychology, 1942, 31,
pp. 134—148.

Harrell, T. W. A. A factor analysis of mechanical ability
tests. Psychological Bulletin, 1939, 39, p. 52.

Havinghurst, R. J., and Breese, F. F. Relations between
abilility and Social Studies in a midwestern com-
munity: III, Primary Mental Ability. Journal of
Educational Psychology, 1947, 38, pp. 2 - .

Horn, J. L. Fluid and crystallized intelligence: A factor
analytic study of the structure among primary
mental abilities. Unpublished Doctoral Disserta-
tion. U. of Illinois, 1965.

Junttonen, R., Wagner, D., and Fischer, K. The effects of
sex and Verbal labels on spatial test performance.
Unpublished manuscript. Michigan State University,
1972.

Kelley, T. L. Crossroads in the Mind of Man. Standford:
Stanford U. Press, 1928.

Krumboltz, J. D., and Christal, R. E. Short term practice
effects in tests of spatial aptitude. Personnel
and Guidance Journal, 1960, 2Q, pp. 385- .

MacFarlane, M. A study of practical ability. British
Journal of Psychology, Mono. supp. #8, .

Magnusson, D. Test Theory. Mass.: Addison-Wesley, 1966.

McNemar, Q. On growth measurement. Educational and P3 —
chological Measurements, 1958, IE, N5.'I, pp. -55.

McNemar, Q. Psychological Statistics. New York: Wiley &
Sons, , pp. - .

Michael, W. B., Guilford, J. P., and Zimmerman, W. S. The
description of spatial—visualization abilities.
Journal of Educational and Psychological Measure-
ments, I957, 22, pp. 185-199.

67

Murphy, L. W. The relation between mechanical ability
tests and verbal and non-verbal intelligence tests.
Journal of Psychology, 1936, 2, pp. 353-366.

Myers, C. T. Some observations of problem solving in
spatial relations tests. Research Bulletin. New
Jersey: Educational Testing SerGice, 195?: pp.
58-61.

Paterson, D. G., 3E.2l° The Minnesota Mechanical Abilities
Tests. Minneapolié: U. of Minnesota Press, 1930.

Pawlik, K. Concepts and calculations in human cognitive
abilities. In R. B. Cattel (ed.) Handbook of
Multivariate Experimeptal Psycholo . Chicago:
Raid McNaIly, 1966, Ch. 18, p. 3.

Ranshaw, T. A. Factorial study of two and three dimensional
space tests. Doctoral thesis, U. of Edinburgh,
1950.

Robinson, J. S. The effects of learning veiled lables for
stimuli on their later discrimination. Journal of
Experimental Psycholquj 1955, 19, pp. 112:114.

 

Schmidt, F. L. Spatial ability: The feminine achilles'
heel. Unpublished manuscript, Michigan State Uni-
versity, 1971.

Shellow, S. M. An intelligence test for stenographers.
Journal of Personnel Research, 1926, 5, pp. 306-308.

Smith, I. M. Spatial Abilit : Its Educational and Social
Significance. San D1ego: R. R. Knapp: 1964.

Smith, I. M. Measuring spatial ability in school pupils.
Occupational Psychology, 1948, 22, 3, pp. 150-159.

 

 

Spearman, C. The Abilities of Man. London: MacMillan,
1927.

Super, D. E. Appraising Vocational Fitness. New York:
Harper an Row, .

Thurstone, L. L. Prima Mental Abilities. Chicago:
Chicago U. Press, 1938.

Thurstone, L. L. Some primary abilities in visual thinking.
Psychometric Lab Re ort, No. 59. Chicago: U. of
Chicago Press, 1950.

 

 

68

Tinker, M. A. Speed, power and level in the revised
Minnesota Paper Formboard Test. Journal of Genetic
Psychology, 1944, 63, pp. 93-97.

Tyler, L. E. The Psychology of Human Differences. New
York: Appleton-Century-Crofts, 1965, p. 245.

 

United States Employment Service. Estimates of Worker Trait
Requirements for 4000 Jobs. Washington, D.C.:
Government Printing Office, 1957.

 

Vernon, P. E. The Structure of Human Abilities. New York:
Wiley, 1950.

Winer, B. J. Statistical Principles in Experimental Design.
New YorE: McGraw—Hill, 19622

 

Witkin, A. A., E£,2$° Ps chological Differentiation. New
York: W1ley, 9 .

Wittenborn, J. R. Mechanical ability: Its nature and
measurement. Educational and Psyghological Measure-
ments, 1945, 2, pp. 2414260.

APPENDIX

APPENDIX

1. Protocol for Training Session One

"Thank you all for staying. You have been chosen
to participate in an experiment designed to improve your
spatial ability. The test you have just taken was a measure
of that ability. Although previous research has shown that
spatial ability can be improved, it is the purpose of this
research to attempt a different approach to improving it.
I will try to improve your spatial ability by intensive
practice both tonight and tomorrow by allowing you to
practice on several types of spatial ability tests. As you
practice on these tests, try to develop an approach to the
problem which will enable you to envision the total figure.
At the end of each practice test, I will give you the
correct answers and you can check your own score. If you
have any questions about a problem you can ask it then. On
each test you will be allowed double the recommended test
time so there is no need to hurry. Are there any questions?
(Pause.) Let's begin the first practice test."

2. Protocol for Training Session Two

"I'm glad to see that you're all back. Tonight's
practice will be essentially the same as the first night.
You should feel that you are getting better at doing the
problems. If there are no questions then we will begin.
(Pause.) The next time we will meet will be next (date,
time, place). O.K.? Let's begin."

69

"I7'11111’11'111’l111(S