You are on page 1of 15

Psychological Bulletin

1972, Vol. 77, No. 2, 129-143

THURSTONE'S PRIMARY MENTAL ABILITIES AND


STRUCTURE-OF-INTELLECT ABILITIES
J. P. GUILFORD i
University of Southern California

The major objective of the present study was to find in the factor-analytic
investigations of L. L. Thurstone some information concerning factorial in-
tellectual abilities that have unique places in the structure-of-intellect (SI)
model, and, more incidentally, to determine which SI abilities are represented
in Thurstone's published primary mental abilities (PMA) battery of tests.
Although a number of his factors can be cited as probable forecasts of SI
abilities, in only one or two instances was a Thurstone factor clearly repre-
sentative of a single SI ability, unconfounded with other abilities. This gen-
eral outcome was largely due to the fact that his analyzed batteries involved
too many different SI abilities for which the number and varieties of tests
were not adequate. In successive analyses, some of his factors tended to con-
verge in the direction of SI abilities, so that his published PMA tests represent
one SI ability each, with the exception of the Reasoning and Number tests.

The termination of a program of intensive been neglected in order to keep the study
research on intellectual abilities or functions within reasonable length.
in the Aptitudes Research Project (ARP) at
the University of Southern California seems COMPARISONS OF THE Two APPROACHES TO
to be an occasion for a reexamination of FACTOR ANALYSIS
L. L. Thurstone's "primary mental abilities," Preparatory to the comparisons that fol-
which have been well known in the literature low, it is important to have in mind certain
for the past 30 years. There is much in com- features of the methodology by which the
mon between his abilities and those issuing PMA and ARP abilities were investigated.
from the ARP program, but there are also There were basic similarities but also some sig-
numerous points of difference that it would nificant differences. Let us note the similarities
seem useful to point out. For those who em- first.
ploy the Thurstone primary mental abilities
(PMA) tests, either in research or in practical Some Similarities
contexts, the newer information arising from In both research programs, as much as
deeper probing into the nature of human in- possible, factor analysis was used as a hy-
telligence has much to offer by way of psy- pothesis-testing device. In planning a research
chological interpretations. study, considerable effort was devoted to a
Although a thorough reexamination of other rational consideration of a domain of abilities,
investigations of intellectual abilities might be with hypotheses proposed concerning abilities
in order, the historical importance of the to be expected in an area of functioning. The
Thurstone analyses gives them first call for supposed natures of those abilities suggested
consideration. In other words, this article is the kinds of tasks needed in order to test
not intended as a comprehensive review of those hypotheses. Being one of the pioneers
factor analyses of intellectual abilities. How- in such research, Thurstone had few prior
ever, it mentions some of the more relevant findings to guide him toward hypothetical
studies that have yielded significant informa- factor abilities. On the other hand, the ARP
tion that helps in the comparison of the had the findings of Thurstone and others,
Thurstone and ARP factors or that served particularly those of the Army Air Forces
as connecting links between them historically. (AAF) research program (Guilford & Lacey,
A few other possible connecting links have 1947). Furthermore, after some early analyses
1
Requests for reprints should be sent to J. P. Guil- the present writer developed the structure-of-
ford, P. 0. Box 1288, Beverly Hills, California 90213. intellect model, which became the ARP's
129
130 J. P. GUILFORD

fruitful source of hypotheses during most of combinations of analyzed test variable, due
its tenure. to incidental selections (Guilford & Zimmer-
Basically, the two methods of analysis were man, 1963). Degrees of obliqueness are prone
the same. The foundation for both was to lack invariance as batteries of tests change
Thurstone's multiple-factor theory, which and as each factor is represented by a different
calls for extracting from a matrix of inter- set of tests. Obliqueness is often to be at-
correlations among tests a number of or- tributed to failures to construct tests that
thogonal factors, and a rotation of axes in provide good experimental controls. With
order to achieve meaningful, psychological optimal controls, perhaps all factors should
variables. Thurstone used his own centroid approach mutual independence. Until we have
method of extracting factors. The ARP exerted all reasonable effort to control pre-
started its program with that method but cisely what examinees have to do in order
soon switched to the principal-axes technique, to make good scores in tests, we cannot re-
which computers made possible. The differ- ject the hypothesis of orthogonality. At any
ence is not an important one. Differences in rate, interpretations of the natures of fac-
rotational procedures were important; they torial abilities should not differ materially
are mentioned shortly. whether rotations are orthogonal or oblique.
Another important condition that was com- Experience seems to bear out this proposition.
mon to the two programs was the use of rela- A more critical difference in rotational
tively homogeneous groups of experimental methods pertains to the basis for location of
examinees or subjects. The homogeneity was axes in final solutions. Thurstone utilized
in terms of age and educational level, and graphic methods of rotation, aiming at the
sometimes, also, sex membership. The im- two objective criteria of positive manifold and
portance of homogeneity as an experimental simple structure. The former should apply
control cannot be overemphasized. As Kelley when dealing with aptitudes, which are uni-
(1928) pointed out, it is likely that many a polar concepts. The latter carries the hidden
Spearman g that had been found in early assumption that the tests that we create tend
factor analyses was due to lack of homoge- to cluster along the lines of fundamental psy-
neity of the tested population. Many tests chological variables of individual differences.
correlate with age, education, or sex, which Especially when tests are aimed at what are
makes them correlate with one another, thought to be relatively independent abilities,
whereas with homogeneous groups they might there is evidently enough truth in this assump-
correlate zero or near zero. If a genuine psy- tion to make the principle roughly applicable.
chological g is present in the data, the mul- It would certainly not be a safe assumption in
tiple-factor methods of analysis should bring all cases. Experience shows that it is easier
it out. It is impossible to find a g of uni- to construct factorially complex tests than
versal scope, as Spearman demanded, how- factorially univocal tests. Accumulating ex-
ever, where so many of the intercorrelations perience has led the present writer to dis-
among tests of intellectual qualities are zero. trust the simple-structure principle. Other
The present writer found that among several voices are being heard in the same vein
thousand such coefficients, as many as 18% (Butler, 1969).
could be regarded as zero (Guilford, 1964). In the ARP program, the overriding con-
cern was to achieve invariance of psycho-
Some Differences logical factors (see Guilford & Hoepfner,
The really important differences between 1971). No factor proposed as a psychological
Thurstone and ARP techniques were in the concept has much claim for attention unless
methods of rotation. To mention a less critical it can be replicated in places where it should
difference first, most of Thurstone's analyses be found. The graphic rotations performed in
entailed oblique rotations of axes, whereas all the early studies by the ARP observed psy-
the ARP rotations were orthogonal. The chological meaning as a third important cri-
choice of orthogonal rotations in the latter terion for rotations. When Cliff's method of
case was dictated by the belief that oblique rotation to congruency of factor matrices ap-
rotations are largely a function of particular peared on the scene (Cliff, 1966), it became
MENTAL AND STRVCTVRE-OF-INTELLECT ABILITIES 131

the adopted procedure. All the earlier analyses CONTENTS


were then redone, using Cliff's method. In FIGURAL
applying the method, starting with the prin- SYMBOLIC
SEMANTIC
BEHAVIORAL
cipal-axes matrix, the investigators set up one
hypothetical factor matrix after another as PRODUCTS
targets toward which to aim rotations until
a good fit and good replications were achieved.
It is recognized that such constraints pro-
duce better fits to theory than are probably
justified. But the procedure faces the fact .TRANSFORMA-
TIONS
of life that the large degree of indeterminancy
•IMPLICATIONS
incident to factor-analytic processes does not
justify blind faith in results obtained by any OPERATIONS
known rigorous rotational operations. One EVALUATION
CONVERGENT
outcome has been of some interest. In general, PRODUCTION
DIVERGENT
when good fits to theory and good invariance MEMORY
PRODUCTION

are reached, simple structure of a kind is also COGNITION

achieved. The truth is that several different FIG. 1. The structure-of-intellect model.
solutions, all with what looks like simple struc-
ture, are possible. There have been attempts and that each ability would have extensive
to achieve mathematical definitions of simple generality, Thurstone assembled batteries for
structure, but rotations so as to reach any one analysis with entirely too much variety among
of them, such as in a varimax solution, may the tests. The result was that many an SI
lead the investigator seriously astray psycho- ability was represented by only one or two
logically (Guilford & Hoepfner, 1969). Even tests. In ARP experience, in an exploratory
in terms of mathematical definitions, the study it takes four or five tests aimed at a
simple-structure criterion is not to be fully hypothesized ability in order to ensure ade-
trusted. quate representation. In his first PMA analy-
Apart from the rotation problem, there was sis, Thurstone probably had 28 SI abilities
an incidental feature of the Thurstone analy- represented, as suggested by current knowl-
ses that endangered good solutions in some edge of components of such tests. Eleven were
places. This feature was his frequent use of represented by one test each, 8 by two tests
alternate forms of the same test in the same each, and 9 by three or more tests. Thurstone
analysis, for example, three reading tests, extracted and rotated 12 factors, of which he
three reasoning tests, or four numerical-opera- interpreted 9, and felt reasonably sure of his
tion tests. The outcome is harmless when such interpretations of less than that number. The
tests generate by themselves a factor that can fact that other psychological factors were
be easily spotted as being actually a specific. latent in his data was demonstrated later by
But the trouble is that they often carry other Fruchter (1948) and Zimmerman (1953) in
tests along with them, which probably means reanalyses of those data.
a confounding of the specific with something
else, and the other tests may be robbed of STRUCTURE OF INTELLECT
what should be variances from other common Since the ubiquitous frame of reference for
factors. The importance of this situation for what follows is the structure-of-intellect
number factors is pointed out later. model, for the benefit of readers less ac-
One historical handicap that Thurstone suf- quainted with it, this very brief exposition on
fered was that at his time no one dreamed of it is provided.
the very large number of intellectual abilities Each SI ability is represented by a par-
such as have been found by the ARP.2 Ex- ticular cell in a three-dimensional matrix (see
pecting only a few primary mental abilities Figure 1) and is defined by its unique con-
2
The number of intellectual abilities now demon- junction of a kind of operation with a kind of
strated by factor analysis is approximately 100, with content and a kind of product. Each ability
more predicted by the SI model. is designated by a trigram. For example, CMU
132 /. P. GUILFORD

stands for cognition of semantic units, which stone, 1938a), the factor that he character-
means awareness or understanding of ele- ized as "verbal relations" was with loadings
mentary constructs, such as objects thought of .38 or above for a variety of tests of types
about. CMU is most clearly assessed in in- whose dominant SI factors have been for
dividuals by means of a vocabulary test. An- abilities other than CMU (cognition of se-
other example is MFI, which stands for mem- mantic units), in SI terminology. Three tests
ory for figural implications, and is best as- should have represented CMC (cognition of
sessed by paired-associates memorizing of semantic classes).3 They were Reading I,
pairs of figures. The individual learns con- Reading II, and Word Grouping. Two tests
nections between members of the pairs so that are now known to represent DMR (divergent
one member comes to imply the other. EST production of semantic relations)—Controlled
stands for evaluation of symbolic transforma- Association and Inventive Opposites. The lat-
tions, which means rendering judgments con- ter also has some variance from NMR (con-
cerning changes in information that is com- vergent production of semantic relations) be-
monly in the form of letters and numbers. cause it restricts each item to only two re-
This ability is known to apply in judging sponses, each with a given initial letter. The
whether one algebraic expression is correctly divergent aspect probably conies from the
transformed into another, as by factoring, fact that alternative responses are to be
transposing, or substituting events. given. It was the presence of these two tests
It may have been noted that the first letter that enabled Fruchter (1948) to find a factor
of a trigram stands for the kind of operation, of associational fluency (DMR) in his re-
the second for the kind of content, and the analysis of Thurstone's data. It also enabled
third for the kind of product. In each case, Zimmerman (1953) to report a similar factor
the letter is the initial of the category name from the same data.
(D = divergent production, B = behavioral, Representing other SI abilities were Verbal
R = relation, and so on), except that N Analogies (cognition of semantic relations,
stands for convergent production (C was pre- CMR) and False Premises (evaluation of se-
empted by cognition), and M stands for se- mantic implications, EMI; and evaluation of
mantic (S was preempted by symbolic). semantic relations, EMR). Only the two vo-
Meanings of the IS categories of the model cabulary tests—-Vocabulary (Chicago) and
(5 operations, 4 contents, and 6 products) Vocabulary (Thorndike)—are of the kind
should become clear from the discussions, that commonly represents the best-known
which include descriptions of their empirical verbal ability CMU, but in Thurstone's first
referents, the psychological tests. Space is not analysis they had even stronger loadings on
taken to present formal definitions here. A other factors. In later analyses, however, he
detailed treatment may be found in the found his verbal factors narrowed more and
writer's book (Guilford, 1967). more to vocabulary and reading-comprehen-
sion tests, which others have found to repre-
THURSTONE FACTORS sent faithfully SI ability CMU.
We consider each of Thurstone's better-
known PMA factors that he reported during Space Factors
the interval from 1938 to 1951. They include Thurstone (1938a) first described his space
what he called verbal, space, number, percep- factor as a "facility in spatial and visual
tion, memory, induction, deduction, word imagery [p. 80]." This definition seems to fit
fluency, and his two closure factors Cl and the factor called "visualization" in the AAF
C2. We look at his verbal factors first. research (Guilford & Lacey, 1947) and finally
given the SI designation of CFT (cognition of
Verbal Factors figural transformations). The AAF program
The verbal factors found in most of Thur- 3
stone's analyses would now be regarded as It is possible to make statements regarding the
probable SI abilities represented by Thurstone's tests
verbal composites, each a confounding that because some of the same tests or their alternate
represented a number of semantic abilities. forms have been analyzed by the ARP. Others are
For example, in his first PMA analysis (Thur- similar to other ARP-analyzed tests.
MENTAL AND STRVCTVRE-OF-INTELLECT ABILITIES 133

found repeatedly a second space factor later, hension in the AAF research), and Paper
called "spatial orientation," which was de- Puzzles. The latter is like the Minnesota
nned as an ability to comprehend arrange- Form Board test, which has a legitimate claim
ments of objects in visual space. Of the 13 to represent CFT.
tests loading .39 and above in Thurstone's The appearance of two different factors
first analysis, 5 would seem to represent CFT. both based strongly on tests representing CFT
Surface Development, Block Counting, and requires explanation. Thurstone's distinction
Form Board have been found to do this in between the two was in terms of movement of
later analyses by the ARP and others the entire object versus movement of parts
(Michael, Zimmerman, & Guilford, 1950). within the object. Because at least three of
Lozenges A and Lozenges B appear to fit the the tests for Si were sufficiently similar to be
current conception of CFT, which is the actually alternate forms of the same test, there
ability to imagine changes in visual objects, is much suspicion that Si is primarily a
such as movements, rotations, reversals, or specific component. Such tests should be ex-
rearrangements. Two tests in Thurstone's list pected to have significant loadings on S2 as
seem to represent SI ability CFS (cognition well as on S3, but in the oblique solution they
of figural systems), which was denned above have loadings of essentially zero on 82. In an
with the name of spatial orientation-—his orthogonal solution, the loadings of these
Flags and Cubes. Both of these tests came tests on 82 might be significant. The correla-
out on a factor identifiable as CFS in the tion reported between Si and 82 was only .38,
analysis by Michael et al. (1950). however, which throws some doubt on this
Not until much later did Thurstone come suggestion. There may be some actual differ-
to recognize more than one space factor, as ence between abilities to imagine transforma-
seen in his 1950 report, in an analysis de- tion of wholes versus of parts, but this find-
voted primarily to visual-space tests.4 In that ing of Thurstone's stands unique and needs
study he found what he regarded as four verification.
space factors. Factor Si he defined as the Thurstone's factor S8 in his 1950 analysis
"ability to visualize a rigid configuration when was represented by only two tests, but they
it is moved into different positions." The tests can qualify as tests of the SI ability CFS or
were Figures and Cards, in both of which the AAF spatial orientation. Thurstone did
the examinee (the subject) is to say whether not interpret this factor except to put it in
two simple figures having the same shape and the space category. Its leading test Cubes
containing holes could be showing the same has a history of loading on factors that rather
side or whether one would have to be turned clearly represent ability CFS (Michael et al.,
over. They are essentially two forms of the 1950, 1951). Lozenges A could also well have
same test and both would qualify for SI some relation to CFS. But both tests may
ability CFT. The test Reversals and Rota- also have some relations to CFT.
tions was described as being similar to Figures The AAF research found a factor called Sa
and Cards except that the pictures were of (not to be confused with Thurstone's S 2 ),
real, familiar objects, such as bananas or which it interpreted as kinesthetic in nature.
keys. This test should also qualify for CFT. It was common to Thurstone's test called
Factor S2 was defined as the "ability to Hands and his test Flags. In Hands, a large
visualize a configuration in which there is number of sketches of the human hand are
movement or displacement of the several parts shown, each with the hand in a different
of the configuration." Four of the six tests position. The subject is to say whether the
on it present histories of having significant picture represents a right or a left hand. In
loadings on factors for CFT—Mechanical Flags, the subjects is to say whether the
same side of the United States flag is showing
Comprehension, Mechanical Movements, Sur- in two paired views, with the flag rotated at
face Development (called Pattern Compre- different angles. Examiners administering these
* L. L. Thurstone. Some primary abilities in tests have observed that subjects often make
visual thinking. (Report No. 59) Psychometric Lab- at least minimal hand movements in solving
oratory, University of Chicago, 19SO. the items.
134 J. P. GUILFORD

In his 1950 analysis, Thurstone found a Analogies, a figure-analogies test), another to


factor K, with Hands leading on it and with be primarily for CMR (Verbal Analogies),
Bolts as its helpmate. In each item of the and still another (Picture Recall) for MFU
latter, a bolt with a right-hand screw is par- (memory for figural units) or MMU (memory
tially screwed into a block of wood that is for semantic units) or both, depending upon
tilted in a certain position. The subject is to whether the subject remembers appearances
say in which of the two directions the bolt of the pictures or what they represent in the
must be turned in order to drive it further into way of real objects.
the block. Thurstone also reported the ob- The AAF psychologists had a decidedly
served involvement of the subjects' hands, better basis for calling "perceptual speed" a
which led him to conclude that the factor factor that coupled Thurstone's Identical
"represents kinesthetic imagery." Forms with their own tests, Spatial Orienta-
It can be added that Thurstone also ana- tion I, Spatial Orientation II, and Speed of
lyzed the same battery of space tests along Identification. The first of these three in-
with additional tests thought to involve me- volved comparing and judging pairs of aerial
chanical knowledge.5 So far as the space photographs for identity, the second involved
factors were concerned, the results were es- matching aerial photographs with places on
sentially the same as those in the 1950 analy- a map of the terrain, and the third involved
sis. matching identical airplanes. Although, as in
Thurstone's earlier analyses, some non-EFU
Factors Named "Perceptual" tests tended to go with the list on the AAF
In his first 1938 analysis, Thurstone factor, as other tests representing other abil-
(1938a) said of a factor that he called sim- ities in common with those "foreign" tests
ply "perceptual" that it was a "facility in were included in analyzed batteries, the re-
perceiving detail that is embedded in ir- striction of the perceptual-speed factor to
relevant materials [p. 81]." His interpreta- EFU-type tests became more definite. Inci-
tion apparently rested very heavily upon the dentally, this kind of outcome is an example
one leading test, Identical Forms, which in of how tests without support from other tests
recent history has always helped to mark of their stronger common factors in an analysis
factors for EFU (evaluation of figural units) go hither and yon, depending in part upon
and which, following AAF terminology, has chance-inflated correlation coefficients. Such
most often marked a factor called "perceptual tests of underrepresented factors often con-
speed." Identical Forms calls for judgments fuse the picture in an analysis, as they did in
of whether each of five similar figures is or is many of Thurstone's early studies.
not identical with a key figure, hence it should The AAF name for its factor, perceptual
logically represent EFU. But on the basis of speed, was in recognition of the fact that its
present knowledge, the other eight tests on tests comprised easy items that almost any-
the factor in Thurstone's first analysis repre- one could do perfectly if he has the time. In
sents anything but EFU. Three tests represent the ARP research, such a factor was fre-
the verbal factor CMU—Vocabulary, Comple- quently found, marked usually by Thurstone's
tion (a vocabulary test in completion form), Identical Forms and Part IV of the Guilford-
and Disarranged Sentences (given scrambled Zimmerman Aptitude Survey, entitled "Per-
words belonging to a sentence, the subject is ceptual Speed." In a recent investigation ° of
to read and comprehend the sentence). Two figural-evaluation abilities the question was
tests, verbal Classification and Word Group- raised whether speed is a natural attribute of
ing, have been found by the ARP to represent the ability. After all, there is nothing about
alike two other abilities, CMC and NMC an ability of EFU to suggest that speed is
(convergent production of semantic classes). important. To test this hypothesis, two new
One other test appears to be primarily for 0
CFR, cognition of figural relations (Pattern K. I. Hoffman, J. P. Guilford, R. Hoepfner, and
W. J. Doherty. A factor analysis of figural-cogni-
6
L. L. Thurstone. An analysis of mechanical apti- tion and figural-evaluation abilities. (Report No. 40)
tude. (Report No. 62) Psychological Laboratory, Uni- Psychological Laboratory, University of Southern
versity of Chicago, 1951. California, 1968.
MENTAL AND STRUCTURE-OF-INTELLECT ABILITIES 135

tests were designed, one almost a power test Thurstone & Thurstone, 1941). The 1941
(Judgment of Size) and one a partially analysis brought four CSU (cognition of sym-
speeded test with items varying in difficulty bolic units) tests into the picture with two
(Judging Figural Combinations). In Judg- EFU tests and one EMU test on the same
ment of Size, a central geometric form is to be factor. The four CSU tests were Mirror Read-
compared with four surrounding figures of ing, Word Puzzles, Incomplete Words, and
the same shape, with the subject to say which Identical Numbers. Mirror Reading required
one is identical in size. In Judging Figural the subject to recognize a word as it would
Combinations, the subject is to say which of be seen in a mirror. Word Puzzles was an
five outline squares contains the same small anagrams test, the words being presented with
geometric figures as a standard set, where scrambled order of letters. The two EFU tests
there are variations of shapes, sizes, and num- were Identical Pictures, and Faces, both re-
bers of small figures. These tests came out quiring matching identical objects. Verbal
with Thurstone's Identical Forms on a factor, enumeration was described as an EMU test
thus linking the factor with history. Thurstone earlier.
was wrong in concluding that a feature of Bechtoldt (1947), a student of Thurstone,
the ability is "perceiving detail that is em- seems to have succeeded in separating the
bedded in irrelevant material [Thurstone, three SI evaluation-of-units factors in his
1938a, p. 81]," and the AAF psychologists analysis. His factor C, which he interpreted
were wrong in attributing a speed feature to as "speed of recognition of predetermined
the ability. symbols in contexts of discrete distractors
Historically, there has been another issue [French, 1951, p. 651," by its conjunction of
in progenitors for EFU. In a special analysis tests justifies identification with ESU. Note
of his perceptual ability, Thurstone (1938b) his reference to symbols in his statement. In
found a factor identified as "perceptual" but three tests, the subject was to cross out speci-
having no tests loaded significantly on it that fied letters of numbers within mixed sets of
would now be placed in the figural-content other letters and numbers.
category. There were three tests that should His factor A was described as "fluency of
represent the parallel ability ESU (evaluation associational recognition with perceptual ma-
of symbolic units)—Identical Numbers, Iden- terials [French, 1951, p. 65]." The qualifica-
tical Names, and Scattered Xs. All three are tion "perceptual" is misleading, for the con-
obviously in the symbolic category; matching tent of the tests is definitely semantic, not
things for identity is an evaluative task; and figural. The strongest test for the factor was
the products are units. Interestingly enough, Word Checking, which asked the subject to
two other tests on Thurstone's factor are of check words that belong to a specified class,
the kind that are now known to measure the for instance, things growing and smaller than
ability EMU (evaluation of semantic units), a football. Other tests were Unfinished House
another parallel variable. Concrete Associa- ("check listed words associated with an un-
tion asks the subject to mark words in a list finished house") and Verbal Enumeration,
that are clearly associated with a key word. which was previously described. Still other
Verbal Enumeration asks the subject to check pertinent tests were Opposites ("in a long list,
all words in a list that belong to a stated class. mark every pair of words that includes op-
Abstract Classification was of a similar nature; posites"). Size Comparison ("in pairs of
another EMU test. Thus, what Thurstone named objects, mark the larger of the two"),
identified as his perceptual factor in his sec- and Boys' Names ("in a list of words, check
ond analysis was apparently a confounding names of boys"). All these tests logically
of ESU and EMU. The operation and product qualify for ability EMU.
were the same as for EFU, but the contents Bechtoldt's factor Y was interpreted by him
were different. as "facility in organizing simultaneous visual
Similar confoundings occurred in subsequent configurations under distraction of continued
Thurstone analyses, once including tests for act [French, 1951, p. 65]." Only the "visual
all three abilities, EFU, ESU, and EMU, and configuration" component of this description
once for only two of them (Thurstone, 1940; really fits the picture of the factor. Thurstone's
136 J. P. GU1LFORD

Identical Forms helped to identify the factor, symbolic implications). It is apparent that
but with a barely significant loading. Stronger the Digit Symbol test involves newly learned
tests included Shape Constancy (involving connections between digits and symbols and
matching and equating pairs of objects for that the goodness of learning and memory is
shape), Picture Squares (involving finding a an important asset in gaining speed in that
pair of identical figures in a matrix of 16 test. Numerical-operations tests reflect not
figures), and Two-Hand Coordination, which recent and controlled learning but older and
seems to have no claim to being a representa- much less controlled learning. Confirmation
tive of EFU, in spite of its having the highest of the relation of numerical-operation tests to
loading of .48. MSI was forthcoming in a new analysis in a
study of mathematical aptitude (Guilford,
Numerical Factors Hoepfner, & Petersen, 1965). But in the same
In his extensive review of factor analyses analysis there was also a relation to NSI, thus
involving intellectual abilities, French (1951) also supporting the original placement of
remarked that the Thurstone factor of num- numerical-facility factors in the SI model.
ber or numerical facility was among the best There is further information that throws
established. There would have been much light on the original Thurstone number factor
consensus on this proposition. Placement of and most of such factors found later. That is
this factor in the SI model proved to be to the effect that they have been heavily con-
something of a problem, however, and the founded with a number-operation specific. It
reason eventually became apparent. The his- has usually happened that, as in Thurstone's
tory of number factors is unique. analysis, more than one numerical-operation
In Thurstone's (1938a) first analysis the test has been used. In the AAF analyses, two
list of tests representing his number factor number tests were included, and the loadings
looks very good. There were four tests repre- were near .80. In other words, they were
senting the four basic number operations specific-inflated. Tenopyr 7 made a special
leading the list: Multiplication, Addition, study of symbolic-memory factors, in which
Subtraction, and Division, with loadings of she introduced several new tests designed for
.81 to .62. Experience has shown that such MSI along with four numerical-operation
high loadings should be suspected. The other tests. The result was a very strong number-
four tests in the list all involved numerical specific factor featuring the number tests,
operations to some extent, and their loadings only one of which (Addition) had also a sig-
were .62 to .38. They were Number Code, nificant loading on the common factor MSI.
Numerical Judgment, Tabular Completion, In a reanalysis of the same battery,8 only one
and Arithmetic Reasoning, all of which in- numerical-operation score (a composite from
volved some numerical operations but also four number tests) was included, so as to
other processes that should have shown up in avoid involvement with a number-specific
terms of significant loadings on other factors. component. The score went on MSI, but with
The first placement of the numerical-fa- a loading of only .35. The natural conclusion
cility factor in the SI model by the writer is that the number factors of historical note
was for ability NSI (convergent production have been largely specific variables, unique to
of symbolic implications), for the reason that tests requiring operations with numbers, with
a certain operation applied to two numbers possible confoundings with certain factors
implies another number and no other, as each representing SI abilities, depending upon the
individual has learned and as he should pro- composition of the test battery. The common-
duce (Guilford, 1959). Two independent find- factor affiliations for numerical-operation tests
ings, however, called for reconsideration. Both
Davis (1956) and de Mille (1962) found that 7
M. L. Tenopyr. A factor analytic study of sym-
a number-operation test correlated in such a bolic-memory abilities. Unpublished doctoral disserta-
manner with the Digit Symbol test of the tion, University of Southern California, 1966.
8
M. L. Tenopyr, J. P. Guilford, and R. Hoepfner.
Wechsler-Bellevue Intelligence Scale, and A factor analysis of symbolic-memory abilities. (Re-
with others, as to emerge with it on a factor port No. 38) Psychological Laboratory, University of
that could be identified as MSI (memory for Southern California, 1966.
MENTAL AND STRUCTURE-OF-INTELLECT ABILITIES 137

seem to be with MSI and NSI, although the subject memorizes his items of informa-
weakly in both instances. Numerical-opera- tion, more than to actual correlations among
tion tests therefore seem to measure mostly memory abilities.
specific number-operation skills. In his 19 SO analysis, Thurstone was con-
cerned mainly with spatial abilities, hence he
Memory Factors featured visual-figural tests. Two of them—
Thurstone's (1938a) first analysis reported Memory for Pictures and Memory for Geo-
a memory factor, which he apparently re- metric Designs—generated a common factor
garded as possibly only one of a family of that he regarded as being distinct from his
abilities, for he suggested possible factorial earlier memory factor. Both tests were in
differences to be expected between incidental recognition form and were designed so that
and intentional memory, rote and logical verbalizing should not help examinees very
memory, and memory within different sense much. The SI ability suggested is MFU. One
modalities. The three leading tests on his surprise was that a test called Visual Memory
memory factor were all of the paired-associ- did not also go on the factor. The name of
ates variety—Number-Number, Word-Num- that test is misleading, however. Two frames
ber, and Initials (associated with family for exposure, each with an irregular figure on
names). It will be recognized that the con- it, were flashed on a screen with a very short
tent is symbolic, and what is learned is in blank interval between them (time interval
the nature of implications. The dominant SI not stated), with the subject to say whether
ability is therefore MSI, an ability that was or not the second is identical with the first.
just discussed in connection with the number Despite the condition of successive exposure
factors. There were two other tests loaded on of the two figures to be compared, the test
the factor, Figure Recognition and Word probably represents EFU and should not be
Recognition, which probably represent abilities expected to go for MFU. The nature of the
MFU and MMU (the latter assuming that factor on which Visual Memory did go, with
word meanings were stored), but possibly one other test, cannot be interpreted, but it
also MSU (memory for symbolic units), if could have been for EFU. This is suggested
appearances of words in terms of letter con- by the need for comparison and judgment of
tent were stored. Paired-associates tests in two figures as to identity, although exposed
recall form, which he used, can also involve in quick succession rather than simultaneously
memory for units, so MSU could have been a as in the usual EFU tests.
confounding component in his memory factor.
From the 1940 analysis, a clear-cut factor, Induction Factors
interpretable as MSI, was reported, with two In different analyses, Thurstone found what
good tests on it. They were Initials and Word- he called an "induction" factor that rather
Number, two MSI tests from the first analysis. clearly corresponds to SI ability CSS (cogni-
In the 1941 analysis, the memory factor was tion of symbolic systems), although in each
represented by three tests for three different case one or two extra tests of other SI
SI memory abilities. Figure Recognition un- abilities went along, apparently for lack of
doubtedly stood for MFU, Digit Span for any better place to go. The clearly CSS tests
MSS (memory for symbolic systems), and were Number Series, Letter Series, Marks,
First Names (another paired-associates test) Number Patterns, Tabular Completion, Secret
for MSI. The cohesion of memory tests as a Writing (using a letter code), and Letter
distinct group is some evidence, but not very Grouping. All these tests involve seeing some
comforting evidence, for the uniqueness of the kind of systematic arrangement of symbols
memory operation category. Recent analyses in order to solve problems, and their relations
in the ARP program have also revealed that to CSS have been repeatedly demonstrated
differentiation of abilities by factor analysis by the ARP.
within the memory category is not always In the opinion of the writer, both the terms
easy. This difficulty is attributed to lack of "reasoning" and "induction" are much too
experimental control of what content and broad as labels for the ability CSS, which is
product are cognized and then stored when only one of a great number, all of which
138 J. P. GUILFORD

could be defended as reasoning abilities and The operation is evaluation rather than con-
many of them as induction abilities. It has vergent production because the subject need
been proposed elsewhere (Guilford, 1960) that not draw his own conclusions; they are given
the term "induction" could be applied to most to him. A conclusion is an implication from
of the cognition abilities, of which there are the premises, therefore we should readily ac-
24 in the present SI model. It could apply to cept the involvement of EMI. The involve-
all; an explanation follows. ment of EMR must arise from the fact that
If induction is regarded as abstracting or premises and conclusions often state rela-
extracting general information from particu- tions. Relations are obvious in statements
lars, the term can be defended as applying to such as "Jones is older than Brown," the rela-
all of the six products of information. Units tion being "older than." Relations are not so
have attributes by which they are identified, obvious in other kinds of propositions, but
and those attributes become known through subjects possibly process the statements as
repeated encounters with objects that possess such.
them. Classes, relations, systems, and trans- It will probably be agreed that "deduction"
formations are transposable entities. As such, should apply to the act of drawing conclu-
they are derived through encounters with dif- sions, not the act of judging them. It should
ferent individual cases that embody them. Im- take a test in completion form, requiring the
plications come about through replicated en- subject to produce his own conclusions, to
counters, to which the law of frequency in justify the label of "deduction." The ARP
learning applies. By such reasoning, the con- has used such tests, and they go mainly on
cept of induction can be replaced with the the factor for NMI, a convergent-production
operation category of cognition, with many ability (Merrifield, Guilford, & Christensen,
advantages. The disposal of "deduction" is 1962). There is sometimes a little EMI vari-
executed next. ance, enough to be significant, which should
mean that the subject's evaluations of his
Deduction Factors own conclusions offer some help in making a
On both occasions in which Thurstone rec- good score.
ognized deductive factors (1938a, 1940), two As a process, deduction should be assigned
syllogistic tests were significant markers, al- logically to convergent production not evalua-
beit along with some nonsyllogistic tests in tion. The writer has proposed elsewhere (Guil-
both cases. In the first case, the syllogistic ford, 1960) that the four convergent-produc-
tests were False Premises and Syllogisms. The tion-of-implications abilities, NFI, NSI, NMI,
two were perhaps sufficiently different to be and NBI, which are concerned with implica-
accepted as two different tests rather than tions, be regarded as deductive abilities. But
two forms of the same test. The main differ- he also proposed that the four abilities con-
ence was the use of nonsensical premises in cerned with convergent production of rela-
the first case, such as, "All truants are gold- tions—NFR, NSR, NMR, and NBR—also
fish," and realistic premises in the second, involve deducing conclusions, in thinking by
such as, "Jones is older than Brown." In the analogy. Further thinking about the matter
second analysis, the two syllogistic tests leads to the suggestion that all the convergent-
loaded on the deduction factor were known production abilities should be accepted as
as Reasoning II and Reasoning III. The dif- deductive, in which case that operation could
ference was said to be in the complexity of well replace completely "deduction" as a
the syllogisms in the two cases. The nature process. Then distinct varieties would be
of the complexity was not stated. Both were recognized, and usefully so. Both the concepts
true-false tests. of "induction" and "deduction" have hereto-
In ARP experience, true-false and multiple- fore been very ambiguous. The cognition and
choice syllogistic tests rather consistently convergent-production operations are not only
shared their common-factor variances between unambiguously defined, they also have em-
EMI (evaluation of semantic implications) pirical referents in the form of tasks that
and EMR (evaluation of semantic relations). involve them.
MENTAL AND STRUCTURE-OF-INTELLECT ABILITIES 139

Factors Leading to General Reasoning either word fluency or SI ability DSU (di-
Although not recognizing it as a reasoning vergent production of symbolic units), which
ability, Thurstone (1938a) reported a factor most word-fluency factors of the past have
that was a possible forerunner of what later approached. The test in question was First
became known as "general reasoning" in the and Last Letters, in which the subject is to
AAF program of analyses. The list of tests list words beginning with specified first and
for it was headed by Arithmetic Reasoning, last letters. It had the smallest significant
with a loading of .58. Thurstone character- loading (.39) among the tests reported for
ized the psychological trait as the ability to the factor. Leading on the factor were three
solve problems when there is "restriction in tests of types that are now known to represent
solution." The other six tests on the factor ability CSU, cognition of symbolic units, not
were heterogeneous with regard to probable divergent production of the same. Two of the
SI content, however. There were three vo- three were anagram tests (Anagrams and Dis-
cabulary tests (although another factor had arranged Words), and the third was Spelling.
been tagged as the verbal factor), a CFT test The type of spelling test used included words
(Mechanical Movements), a CSU test (Spell- some of which were misspelled, with the
ing), and a numerical test (Numerical Judg- subject to say which ones. Twice in ARP
ment). analyses such a test went on the factor for
In AAF research, it became customary to CSU along with anagrams tests and others.
conclude that any factor for which an arith- In Thurstone's (1940) analysis, the factor
metical-reasoning test had a high loading that he identified as Word Fluency now ap-
represented "general reasoning." The qualifi- pears to have been a confounding of CSU
cation "general" was in recognition of the and CSS. The CSU tests were Mirror Read-
fact that quite a variety of tests often loaded ing and Disarranged Words, which have been
significantly on the factor. described before, and the CSS tests were Letter
A series of analyses on reasoning abilities Series and Number Series, which had mini-
by the ARP ended with one directed espe- mally significant loadings on the factor.
cially at the nature of the general-reasoning In the Thurstone and Thurstone (1941)
factor. Alternative hypotheses as to its nature analysis, 5 good DSU tests were among the
were generated and investigated, with the strongest of 10 tests on the factor identified
outcome supporting best the idea that it is as word fluency. They were Prefixes, First
the ability to see the structure of problems Letter, Suffixes, Rhyming Words, and First
(Kettner, Guilford, & Christensen, 1956). and Last Letters, all of which call for listing
Later, the factor for general reasoning was of real words fulfilling specified class prop-
placed in the cell for CMS (cognition of erties that their names indicate. That this
semantic systems) of the SI model. A ver- factor was confounded with CSU again is
bally stated arithmetical problem conveys to indicated by three good-looking CSU tests in
the problem solver a somewhat complex con- the list: Anagrams, Four-Letter Words, and
ception of a structure or system, conceived Word Puzzles (an anagrams test). In Four-
semantically. There are, of course, many other Letter Words, such words are hidden in con-
kinds of semantic systems. tinuous rows of otherwise pied type. It is a
matter of recognizing familiar words under
difficult conditions, as is true of most other
Word-Fluency Factors CSU tests.
From his first analysis, Thurstone reported
a factor whose psychological character he Closure Factors
described as "fluency dealing with words" Thurstone's two closure factors, Cl and
(Thurstone, 1938a). He recognized that the C2, have occupied the attention of a number
meanings of the words in this connection were of investigators since he reported them in
not of any importance; only orthographic 1944. The study was regarded as exploratory,
features were relevant. His list of tests for as were most of the earlier analyses. Al-
the factor, however, contained only one that though the tests were restricted to visual in-
would today satisfy the identification with put, the question was raised as to whether
140 J. P. GUILFORD

such factors as were to be expected would of .54 was Shape Constancy, which was re-
transcend sensory modalities. Here we con- ported in connection with that ability earlier.
centrate attention on three of the new factors, Two other factors in the analysis seem to
two of which were identified as closure abil- have better claims to interpretations as CFU
ities, where closure means the integration of and NFT, however, which leaves factor A
sensory input so as to differentiate figure from something of a mystery.
ground. Much attention was given to the Thurstone called his factor F "speed of
question of whether it makes a difference if perception," which should not be confused
the closure takes place with and without dis- with earlier factors going by a similar name.
tractions or detracting influences. Here the two rather faithful markers for CFU
In the results from the Thurstone (1944) —Street Gestalt Completion and Multilated
investigation were the bases for what became Words—had higher loadings than for factor
known as Closure 1, or the "speed and A (.53 versus .35 and .44 versus .34, respec-
strength of closure," and in SI placement, tively). New tests, with even stronger load-
ability CFU-V (where CFU is the cognition ings on factor F were Peripheral Span and
of figural units and the added "V" indicates Dark Adaptation. Both involved perception of
that the ability pertains to visual input). This single letters (hence figural content) under
qualification is sometimes needed to remind us difficult viewing conditions, flashed in the
that there is another factorial ability recog- periphery of the subject's visual field. There
nized as CFU-A, for auditory input, from a were only two discordant facts against accept-
factor reported by Fleishman, Roberts, and ing factor F as CFU. These facts were the
Friedman (19S8). The latter deals with such absence of Dotted Outlines and Hidden
tasks as recognizing radio-code signals and Digits from the list of significant tests. In a
sets of dot sounds. In what follows, it should recent analysis (see Footnote 6), the ARP
be understood that CFU refers to CFU-V, did find that a test Hidden Print (like Hidden
for we are not concerned further with CFU-A. Digits except that letters were also used)
It was stated earlier that Thurstone had was one of the strongest tests for CFU.
actually found a factor such as CFU without The same ARP analysis threw much doubt
realizing what it was. In his 1944 analysis, upon the assumed need for distracting or de-
curiously, two factors looked like CFU. He tracting material working against closure and
declined to name his factor A, which he de- even need for closure at all. A new test,
scribed as the ability to form a perceptual Close-Ups, composed of close-up photographs
closure against some distraction and to hold of segments of familiar objects, such as a
that closure against detracting influences. The key or a pineapple, was quite strongly loaded
four CFU tests (as shown by later ARP ex- on the factor. There was no need to fuse un-
perience) were Hidden Digits, Street Gestalt organized elements into objects, and there
Completion, Dotted Outlines, and Mutilated were no distracting elements. The same could
Words. All of these tests except Dotted Out- be said regarding Thurstone's two peripheral-
lines were described earlier. The dotted out- vision tests, in which the objects were com-
lines were of 6-inch letters shown on a screen, plete, isolated letters.
with widely spaced dots to form the lines. Closure 2 is the identification for Thur-
Other tests had distinctly higher loadings on stone's 1944 factor E. Three NFT tests were
factor A, for example, two CFT tests (PMA conspicuous in the list, namely, Hidden Pic-
Space and Kohs Blocks) and two NFT (con- tures (with human and animal forms con-
vergent production of figural transformations) cealed in a landscape) and the two Gott-
tests (Gottschaldt Figures A and Gottschaldt schaldt Figures tests, mentioned above. The
Figures B). The last two involved seeing sim- loadings for the latter were just about equal
ple figures concealed in complex figures. From to their loadings on factor A. The nature of
present hindsight, factor A in the 1944 analy- the leading test, Two-Hand Coordination, is
sis should probably be regarded as a con- difficult to hypothesize in SI terms. There is
founding of CFU and NFT. There may have little in the description of it to suggest NFT.
been some confounding also with the factor The principle of the test was to reflect the
for EFU, for the leading test with a loading difficulty in tapping with the two hands simul-
MENTAL AND STRUCTVRE-OF-INTELLECT ABILITIES 141
taneously as compared with tapping with each Concerning his "perceptual" factor there
hand separately, in a task from which one was much confusion. Although out of them
should expect a large interference effect. In came the AAF ability "perceptual speed,"
interpreting his factor E, Thurstone used ex- later to be identified with EFU, Thurstone's
pressions such as "ability to shake off one analyses were perpetually plagued with fail-
set in order to take a new one," "freedom ures to discriminate EFU, ESU, and EMU,
from Gestaltbindung," and "flexibility in which differ only in terms of informational
manipulating several more or less irrelevant content. Two or more of the three were
or conflicting gestalts." Flexibility seems ob- usually confounded in his results.
viously involved in many of the tests. The The number factor, with Thurstone as with
kind of flexibility would seem to be well de- others, has proved to have been largely a
scribed by SI ability NFT, convergent pro- specific affair, a matter of number-operation
duction of figural transformations. This con- skills. When only one numerical-operation
cept is more precisely defined and is perhaps test is included in an analysis, to avoid the
the core of Thurstone's 1944 factor E. possibility of confounding with specific vari-
ance, such a test shows some small common-
RECAPITULATION factor components identifiable as MSI and
NSI—memory and convergent production
In summarization, we may consider what
with symbolic implications, respectively.
structure-of-intellect abilities had their germs Rather consistently and unusually clearly,
in the findings of Thurstone's factor analyses
Thurstone brought out a factor identifiable
and in his insights regarding outcomes. It is
as MSI, because in each analysis he used two
also of interest to see what SI abilities are
or three different tests of the paired-associates
represented in his published PMA tests.
type. The units of information to be associ-
ated were numbers, letters, or names, hence
Forerunners of SI Abilities
they were symbolic information. Implication
Thurstone's analyses cannot be justifiably is essentially a new concept replacing as-
cited as supporting evidence of the validity of sociation. In a late analysis, Thurstone found
SI theory or the abilities involved, because signs of another memory ability that was
rarely was any ability, now recognized as probably MFU, for it involved recognition
unique, found unconfounded in his results. of pictorial information.
But his insightful descriptions of his hypo- One of the clearest SI abilities found by
thetical psychological variables were very sug- Thurstone was for CSS (cognition of sym-
gestive, and the later use of many of his tests bolic systems). He identified it as "induction,"
in following out implications from his results but in earlier discussions it was argued that
was often fruitful. Thurstone was free to admit many or all of the abilities in the cognition
that there were historical antecedents for category of the SI model could be regarded
some of his factors, for instance, verbal, space, as instances of induction.
number, and memory. Others had strong He thought that he had demonstrated a
claims of being novel findings. "deduction" ability primarily by the use of
The verbal factors in Thurstone's early multiple-choice syllogistic tests, but results
analyses showed relations to numerous tests, in ARP research have shown that such tests
almost all verbal, but representing a number are related to both SI abilities EMI and
of semantic abilities. As time went on, those EMR. In other words, Thurstone's deduction
factors converged toward a variable inter- factor was a composite of two SI abilities,
pretable as CMU. and they have to do with evaluation, not with
His space factor was undoubtedly closer drawing conclusions. Drawing conclusions is a
to what was later called visualization than matter of convergent production, and it was
to the spatial orientation from AAF research. recommended that we substitute the latter
In other words, it led to ability CFT rather operation for the concept of deduction.
than CFS. He found some indication of the Without realizing it, Thurstone found a
latter in his last reported analysis of space factor that could be cited as a forerunner of
tests. the AAF "general reasoning" and the SI
142 J. P. GUILFORD

ability CMS. At least the fatcor was led cognition of symbolic systems, but items from
twice by an arithmetical-reasoning test, which the original Letter Grouping test should mea-
has consistently marked that factor. Before sure CSC (cognition of symbolic classes) as
it was given the CMS identification, it was well as CSS. In the latter test, the experi-
recognized as an ability to comprehend prob- menter has to see symbolic classes after seeing
lems. Arithmetical problems, and others, are the principle of each letter set.
conceived by the problem solver as systems. The PMA Number test is even more com-
Thurstone's word-fluency factor, based most plex factorially. Its common factors are MSI
often on anagrams tests, was undoubtedly a and NSI, for remembering and producing
much better candidate for CSU than for DSU. symbolic implications, but there is a strong
After the first analyses, some added tests that specific component of number-operation skills.
clearly represent DSU also came out on the The importance of number skills in education
factor, indicating that it was a confounding and in other connections in our society, of
of CSU and DSU. Both Fruchter (1948) and course, cannot be denied, so that numerical-
Zimmerman (1953) showed that another operation tests should continue to be useful.
fluency factor was latent in Thurstone's But for measures of the generalized skills
earliest analysis, represented by tests for SI known as MSI and NSI, there are much
ability DMR. better tests, stronger and more univocal.
Thurstone's analysis of perceptual tests led The PMA Perceptual test included at the
to two closure factors, Cl and C2, and lower age levels should be for ability EFU,
eventually to SI abilities CPU and NFT. Al- evaluation of figural units. The Memory test
though he regarded Cl as a matter of strength at the highest age level is for MSI, memory
of closure against detracting material, the for symbolic implications, or what has been
latest ARP analysis in which CPU appeared known historically as associative memory.
throws serious doubt upon the distraction There are no PMA tests representing the
feature and even on the closure feature. A "deduction" factor in the PMA battery, at
more accurate description of the trait is to any age. Neither of the closure factors is
call it the ability to recognize familiar visual represented, perhaps because they came after
objects under conditions of limited input. the battery had been adopted.
The essence of NFT is an ability to produce Taking a more general look at the PMA
required revisions of visual objects. It often battery, to note what kind of coverage of SI
involves tearing down old objects in order to abilities was accomplished, we see that there
form new ones. It does not seem to be a skill are three cognition abilities represented—•
for handling conflicting gestalts, as Thurstone CFT, CSS, and CMU. Three other operation
supposed. categories have one test each—MSI, DSU,
and EFU. In terms of content categories,
SI Abilities in Published PMA Tests
there are two figural abilities, three symbolic,
The SI abilities that are featured in the and only one semantic. In terms of product
PMA tests may vary somewhat from one age categories, three tests pertain to units, one to
level to another, but above the 5-7-year age systems, one to transformations, and one to
level the dominant abilities seem to be rather implications. Thus, although the sampling of
clear. The vocabulary tests for the "verbal" categories has some breadth, there is still
ability should measure status in CMU. The severe limitation in coverage of the great
Space test appears to be primarily for ability domain of human intelligence.
CFT or visualization, where movements or
other changes are cognized. The test Percep- REFERENCES
tion should measure EFU, the evaluation of BECHTOLDT, H. P. Factorial investigation of the per-
visual-figural units. Word Fluency is very ap- ceptual speed factor. American Psychologist, 1947,
propriate as a measure of DSU, divergent pro- 2, 304-305.
duction of symbolic units. BUTLEH, J. M. Simple structure reconsidered: Dis-
tinguishability and invariance in factor analysis.
Where the Reasoning test has two kinds of Multivariate Behavioral Research, 1969, 4, 5-28.
items in it, two different SI abilities are fea- CLIFF, N. Orthogonal rotation to congruence. Psy-
tured. Letter-series items assess ability CSS, chometrika, 1966, 31, 33-42.
MENTAL AND STRUCTURE-OF-INTELLECT ABILITIES 143
DAVIS, P. C. A factor analysis of the Wechsler-Belle- gram Research Reports, No. 5) Washington, D. C,:
vue Scale. Educational and Psychological Measure- United States Government Printing Office, 1947.
ment, 19S6, 16, 127-146. GUILFORD, J. P., & ZIMMERMAN, W. S. Some variable-
DE MILLE, R. Intellect after lobotomy in schizo- sampling problems in the rotation of axes in factor
phrenia. Psychological Monological Monographs, analysis. Psychological Bulletin, 1963, 60, 289-301.
1962, 76 (16, Whole No. 535). KELLEY, T. L. Crossroads in the mind of man: A
FLEISHMAN, E. A., ROBERTS, M. M., & FRIEDMAN, study of differentiate mental abilities. Stanford,
M. P. A factor analysis of aptitude and proficiency Calif.: Stanford University Press, 1928.
measures in radiotelegraphy. Journal of Applied KETTNER, N. W., GUILFORD, J. P., & CHRISTENSEN,
Psychology, 1958, 42, 129-137. P. R. A factor-analytical study of the factor called
FRENCH, J. W. The description of aptitude and general reasoning. Educational and Psychological
achievement tests in terms of rotated factors. Psy- Measurement, 1956, 16, 438-453.
chometric Monographs, No. 5, 1951. MERRIFIELD, P. R., GUILFORD, J. P., CHRISTENSEN,
FRUCHTER, B. The nature of verbal fluency. Educa- P. R., & FRICK, J. W. The role of intellectual
tional and Psychological Measurement, 1948, 8, factors in problem solving. Psychological Mono-
33-47. graphs, 1962, 76 (10, Whole No. 529).
GUILFORD, J. P. Personality. New York: McGraw- MICHAEL, W. B., ZIMMERMAN, W. S., & GUILFORD,
Hill, 19S9. J. P. An investigation of two hypotheses regarding
GUILFORD, J. P. Basic conceptual problems in the the nature of the spatial-relations and visualiza-
psychology of thinking. In E. Harms (Ed.), Funda- tion factors. Educational and Psychological Mea-
mentals of psychology: The psychology of think- surement, 1950, 10, 187-213.
ing. Annals of the New York Academy of Sciences, MICHAEL, W. B., ZIMMERMAN, W. S., & GUILFORD,
1960, 91, 6-21. J. P. An investigation of the nature of spatial-re-
GUILFORD, J. P. Zero intercorrelations among tests lations and visualization factors in two high school
of intellectual abilities. Psychological Bulletin, populations. Educational and Psychological Mea-
1964, 61, 401-404. surement, 1951, 11, 562-577.
GUILFORD, J. P. The nature of human intelligence. THURSTONE, L. L. Primary mental abilities. Psy-
New York: McGraw-Hill, 1967. chometric Monographs, No. 1, 1938. (a)
GUILFORD, J. P., & HOEPFNER, R. Comparisons of THURSTONE, L. L. The perceptual factor. Psycho-
varimax rotations with rotations to theoretical metrika, 1938, 3, 1-17. (b)
targets. Educational and Psychological Measure- THURSTONE, L. L. An experimental study of simple
ment, 1969, 29, 3-22. structure. Psychometrika, 1940, 5, 153-168.
GUILFORD, J. P., & HOEPFNER, R. The analysis of THURSTONE, L. L. A factorial study of perception.
intelligence. New York: McGraw-Hill, 1971. Psychometric Monographs, No. 4, 1944.
THURSTONE, L. L., & THURSTONE, T. G. Factorial
GUILFORD, J. P., HOEPFNER, R., & PETERSEN, H. Pre- studies of intelligence. Psychometric Monographs,
dicting achievement in ninth-grade mathematics
No. 2, 1941.
from measures of intellectual-aptitude factors, Edu- ZIMMERMAN, W. S. A revised orthogonal solution for
cational and Psychological Measurement, 1965, 25, Thurstone's original primary mental abilities test
659-682. battery. Psychometrika, 1953, 18, 77-93.
GUILFORD, J. P., & LACEY, J. I. Printed Classification
Tests. (Army Air Forces Aviation Psychology Pro- (Received June 22, 1970)

Manuscripts Accepted for Publication in the


Psychological Bulletin

African Infant Precocity. Neil Warren (School of African and Asian Studies, University of Sussex, Brighton,
BN1 9QY, England).
Amsel Frustration Effect: Interpretations and Research. John W. Scull (Laboratory of Experimental Psychology,
University of Sussex, Brighton, BN1 9QY, England).
Imprinting in Precocial Birds: Interpretation, Evidence, and Evaluation. D. W. Rajecki (Institute for Social
Research, Research Center for Group Dynamics, University of Michigan, Ann Arbor, Michigan 48106).
On the Function of Structural Paraphrase: The View from the Passive. Moshe Anisfeld (Department of Psychology,
Ferkauf Graduate School, Yeshiva University, 55 Fifth Avenue, New York, New York 10003) and Irene
Klenbort.
Comment on Misuse of Covariance. D. A. Sprott (Department of Statistics, Faculty of Mathematics, University
of Waterloo, Waterloo, Ontario, Canada).
Reply to Sprott. Selby H. Evans (Institute for the Study of Cognitive Systems, Texas Christian University
Research Foundation, Fort Worth, Texas 76129).

You might also like