Lecturer's Précis - Olckers (1951)
"A Factorial Study of Arithmetical
Ability"
Copyright Notice: This material was
written and published in Wales by Derek J. Smith (Chartered Engineer). It forms
part of a multifile e-learning resource, and subject only to acknowledging
Derek J. Smith's rights under international copyright law to be identified as
author may be freely downloaded and printed off in single complete copies
solely for the purposes of private study and/or review. Commercial exploitation
rights are reserved. The remote hyperlinks have been selected for the academic
appropriacy of their contents; they were free of offensive and litigious
content when selected, and will be periodically checked to have remained so. Copyright
© 2010, High Tower Consultants Limited.
|
|
First published online 12:50 GMT 3rd March 2005, Copyright Derek J. Smith
(Chartered Engineer). This version [HT.1 - transfer of
copyright] dated 18:00 14th January 2010
1 - Introduction
Olckers
began by briefly reviewing the history of the notion of “mental factors” in psychological
explanation, explaining how there had been a flurry of interest in psychometric
testing in the early decades of the 20th century, and singling out the
pioneering work of Sir Cyril Burt [Research
Methods Glossary] and Charles Spearman [Research
Methods Glossary] for special mention.
ASIDE: Both Burt (1917)
and Spearman (1904, 1927) had resorted to mathematical procedures to help them
look for subtle explanatory patterns in the accumulated results of their
intelligence testing. At issue was whether there was really such a thing as
“general intelligence”. If there was, then a given person would be expected to
perform approximately equally on tests of mathematics, geography, and English,
say, provided (a) that they had put in equal amounts of study, and (b) that the
tests had been cross-matched for difficulty. The intelligent ones would do well
on all the tests, and the unintelligent ones would do poorly on all of them. On
the other hand, if there were several fundamentally different types of
intelligence, then people would perform well at (and presumably find it easier
to learn) the subjects Nature had best equipped them for, and less well at the
others. Both authors concluded that intelligence was a combination of both
specific and general abilities [known as s- and g-factors respectively], thus
ushering in a period of scientific controversy which continues in some respects
to this day. A good example of a modern theory of multiple specific
intelligences is provided by Howard Gardner’s Theory of Multiple Intelligences (Gardner, 1983).
As
for mathematical abilities in particular, Olckers quoted Spearman's (1927)
claim that "the inter-correlations of arithmetical abilities are traceable
to a single factor over and above g" (Olckers, p2). He also identified a
number of other pre-1951 studies, including .....
·
Thorndike (1922): Olckers cited this monograph by the
influential American learning theorist Edward
Lee.Thorndike (1874-1949) as typical of the s-factor school of thought.
Thorndike, he noted, “conceived the mind to consist of a large number of
particularized and independent abilities” (Olckers, p2). There were, for
example, “at least seven processes or minor functions involved in two-place
column addition, each of which is psychologically distinct and requires distinct
educational treatment” (Thorndike, 1922).
·
Alexander (1935): This study indicated two major
factors in mathematical abilities, namely "a general factor and a specific
factor common to all the arithmetical processes" (Olckers, p2).
ASIDE: In fact, Alexander
introduced the potentially valuable notion of the "functional
cluster" [alternatively "functional unit", "functional
ability", or "unitary trait"] to describe bundles [our term] of
abilities which "function almost exactly alike" (Alexander, 1935,
p116). This allows such terms as "verbal ability" to remain useful
and valid, despite the fact that more detailed investigation can find different
sub-skills within them. Moreover, when one of those sub-skills is common to
several abilities, it will naturally be reported as a common factor upon
statistical analysis; but this does not automatically equate it to Spearman's
g-factor. "The real problem," Alexander argued, was to find "the
smallest number of such factors which will account for all the variation"
(Ibid., p117), but this was not
the same as the number of variables measured in the test battery used.
·
Thurstone (1938): Louis
Leon Thurstone (1887-1955) used a battery of 56 various tests, and, like
Thorndike, identified "a number of independent primary factors and no
general common factor" (Olckers, p2). The primary factors included one for
"number ability", another for reasoning, and another for induction.
·
Cunningham and Price (1934): Olckers cited this nationwide
survey of arithmetical ability in Australia as having demonstrated that
arithmetical calculation and mathematical problem solving [presumably
Thurstone's "number ability" and "reasoning", respectively]
were "almost, though not quite, distinct from one another" (Olckers,
p2).
So the problem facing Olckers in
1951 was how to improve upon the factor analytic conclusions already reached.
He carried out three separate studies of his own, one a preliminary screening
study, and two more precisely targeted follow-ups, as now summarised .....
2 - The Research
STUDY #1 - PRELIMINARY: In the first study, he subjected 112 secondary school pupils (46 girls,
66 boys, average age 14;4 years) to a battery of 15 different types of
arithmetical test, as follows .....
Test 1 - Addition: This was a 4 minute test of columnar addition.
Test 2 - Subtraction: This was a 4 minute test of
columnar subtraction.
Test 3 - Multiplication: This was a 4 minute test of
multiplication, starting with single digits and increasing in difficulty to
five digits by two digits.
Test 4 - Division: This was a 4 minute test of division, starting
with single digits and increasing in difficulty to five digits by two digits.
Test 5 - Convert Roman Numerals: This was a 4 minute 25-item test of
converting Roman numerals into everyday notation [e.g. LVI into 56]. [Perhaps not
surprisingly, this test correlated poorly with the four arithmetical tests
above - see the Row 5 correlation coefficients in the table below.]
Test 6 - Dot Counting: This was a 3 minute test to count
the number of dots in an irregular 20-row array.
Test 7 - Repetition of Digits: This was a 15-item test of the
ability to repeat orally presented digit sequences.
Test 8 - Fractions: This was a 7 minute 24-item test "of the
four rules applied to ordinary and decimal fractions" (Olckers, p3).
Test 9 - Problem Solving: This was a 10 minute 28-item test
of [unspecified] problem solving.
The
remaining six tests were taken from the South African Group Test of
Intelligence. The tests were named as follows, but no details were given .....
Test 10 - Classification, Test 11 - Analysis,
Test 12 - Number Series, Test 13 - Letter Test, Test 14 - Same-Opposite Test,
Test 15 - Figure Test.
Initial
analysis of the accumulated test scores involved correlating all possible pairs
of test scores. Thus the scores on Test 1 were correlated with Tests 2 to 15,
the scores on Test 2 were correlated with Tests 3 to 15, the scores on Test 3
were correlated with Tests 4 to 15, and so on until every possible pairing had
been exhausted. Here are the correlation coefficients which were reported. To
find any particular pairing simply locate the first test as a column and the
second test as a row, and read off the value in the cell of intersection of
column and row .....
|
|
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
13 |
14 |
|
1 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
2 |
.49 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
3 |
.47 |
.57 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
4 |
.40 |
.59 |
.55 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
5 |
.09 |
.08 |
-.04 |
.05 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
6 |
.34 |
.32 |
.39 |
.24 |
-.04 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
7 |
.11 |
.03 |
.07 |
.24 |
.09 |
.07 |
- |
- |
- |
- |
- |
- |
- |
- |
|
8 |
.14 |
.40 |
.29 |
.39 |
.34 |
.00 |
.20 |
- |
- |
- |
- |
- |
- |
- |
|
9 |
.18 |
.27 |
.12 |
.36 |
.35 |
.02 |
.32 |
.47 |
- |
- |
- |
- |
- |
- |
|
10 |
.21 |
.20 |
.13 |
.11 |
.28 |
-.04 |
.18 |
.28 |
.20 |
- |
- |
- |
- |
- |
|
11 |
.05 |
.12 |
.08 |
.13 |
.41 |
.07 |
.26 |
.27 |
.24 |
.28 |
- |
- |
- |
- |
|
12 |
.27 |
.21 |
.25 |
.17 |
.20 |
.11 |
.05 |
.47 |
.51 |
.36 |
.28 |
- |
- |
- |
|
13 |
.13 |
.04 |
.14 |
.16 |
.22 |
.10 |
.20 |
.28 |
.17 |
.51 |
.33 |
.25 |
- |
- |
|
14 |
.02 |
-.02 |
-.02 |
.06 |
.22 |
-.20 |
.15 |
.31 |
.43 |
.31 |
.28 |
.11 |
.20 |
- |
|
15 |
.15 |
.22 |
.18 |
.31 |
.19 |
.12 |
.04 |
.42 |
.35 |
.36 |
.18 |
.37 |
.32 |
.18 |
|
NB: Note the colour-coding we have added to Olcker's
black-and-white original. Where a correlation is high [we have bold green highlighted
all correlations of .40 and above; there were none of -.40 or below], the
argument is that both tests are, to that extent, measuring the same thing.
Thus people who score highly on Test 1 (Addition) also score highly on Test 2
(Subtraction) [the correlation being a healthy (but far from absolute) .49].
Where a correlation is low [we have bold red highlighted all correlations
between -.10 and .10], the argument is that both tests are, to that extent,
NOT measuring the same thing. Thus people who score highly on Test 3
(Multiplication) are not automatically scoring highly on Test 14
(Same-Opposite) [the correlation being -.02]. One of the tests which
correlates poorly with its neighbours is Test 7 (Repetition of Digits), a
popular test of short-term memory, presumably because that type of resource
was not heavily tapped by the other tests. |
||||||||||||||
The
correlation matrix was then factored down to five major factors, using
Thurstone’s (1934/2005
online) “centroid method with guessed communalities” (Olckers, p4). Here
are the five suggested factors …..
Factor I - "Concentration and Retentive
Power": This
is the "still obscure" (Olckers, p11) ability for a combination of
retentiveness, alertness, and sustained attention to the problem at hand.
Factor II - Reasoning: This is the ability "to select
essentials and to see relationships" (Olckers, p6).
Factor III - Speed of Performance: This is the ability to associate
ideas quickly in "performing simple well-practised tasks of a continuous
nature" (Olckers, p6).
Factor IV - Directed Relational Thinking: This is the ability to manipulate
symbols "according to well-known rules" (Olckers, p6).
Factor V - Residual: This factor was left undefined
pending further investigation.
STUDY #2 - FIRST FOLLOW-UP: In the second study, Olckers subjected a new sample of schoolchildren to
a different test battery. This time the battery contained 13 different types of
arithmetic test, as follows …..
Test 1 - Four Rules A: This was a 4 minute test of
"alternate rows of two-digit addition, subtraction, multiplication, and
division sums with whole numbers, involving not more than one step each [which]
would define the nature of the speed factor involved in computation" (Olckers,
p7).
Test 2 - Simple Opposites: This was a 2˝ minute 80-item test
in which subjects had to write down the first letter of the opposite of each
given word [e.g. given "hot" answer "c" (for
"cold")].
Test 3 - Alphabet: This was a 2˝ minute 80-item test in which subjects
had to write down the letter following a given letter pair [e.g. given
"MN" answer "O"].
Test 4 - Four Rules B: As Test 1, but with more complex
four-step questions and two extra minutes.
Test 5 - Cancellation A: This was a 3 minute test of simple
digit cancellation. Presented with a block of randomly sequenced digits,
subjects had to make a line through every 9 they could find.
Test 6 - Four Rules C: As Test 4, but using a different
question set.
Test 7 - Fractions: This was a 4 minute 20-item test of addition,
subtraction, multiplication, and division of ordinary fractions.
Test 8 - Change: This was a 2 minute 20-item test of giving
change.
Test 9 - Cancellation B: This was a 4 minute test of
rule-driven digit cancellation, similar to Test 5, but more complicated. The
specific task was to cancel Os, but only if preceded by 3s, UNLESS followed by
a 7, in which case the 7 had to be cancelled!! The test was preceded by a
number of untimed practice trials to ensure that subjects understood what was required
of them.
Test 10 - Problems: This was a 5 minute 20-item test
"covering the main principles involved in arithmetic" (Olckers, p8).
Test 11 - Number Series: This was a 5 minute 25-item test of
series completion [e.g. given "1, 4, 9" the subject should detect
that the rule was increasing squares, and continue "16, 25"].
Test 12 - Definitions: This was a 3 minute 30-item test in
which subjects had to select the first letter of the word being defined from an
array of five alternative possibilities [e.g. given "used for washing
with" the subject should answer "s" (for "soap")].
Test 13 - Repetition of Digits: This was a 30-item test of the
ability to repeat orally presented digit sequences.
The
test scores were then correlated as before .....
|
|
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
|
1 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
2 |
.34 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
3 |
.30 |
.36 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
4 |
.67 |
.21 |
.31 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
5 |
.49 |
.40 |
.39 |
.30 |
- |
- |
- |
- |
- |
- |
- |
- |
|
6 |
.58 |
.07 |
.18 |
.64 |
.08 |
- |
- |
- |
- |
- |
- |
- |
|
7 |
.45 |
.05 |
-.01 |
.42 |
.08 |
.51 |
- |
- |
- |
- |
- |
- |
|
8 |
.27 |
.17 |
-.06 |
.38 |
.02 |
.31 |
.44 |
- |
- |
- |
- |
- |
|
9 |
.05 |
.12 |
.11 |
.15 |
.46 |
.08 |
.09 |
.11 |
- |
- |
- |
- |
|
10 |
.05 |
.06 |
.08 |
.13 |
-.01 |
.07 |
.33 |
.50 |
.02 |
- |
- |
- |
|
11 |
.20 |
.19 |
.00 |
.28 |
.05 |
.29 |
.53 |
.56 |
.24 |
.49 |
- |
- |
|
12 |
.27 |
.25 |
.03 |
.18 |
.16 |
.17 |
.42 |
.21 |
.08 |
.37 |
.21 |
- |
|
13 |
-.05 |
.12 |
.15 |
.10 |
.03 |
.14 |
.07 |
.10 |
.00 |
.16 |
.16 |
.16 |
|
NB: Colour-coding as for Study #1. |
||||||||||||
The correlation matrix was then factored down to four major factors, as
follows …..
Factor I - "Concentration and Retentive
Power": Same
as Study #1, Factor I.
Factor II - Speed of Performance: Same as Study #1, Factor III.
Factor III - Manipulation: Same as Study #1, Factor IV.
Factor IV - Reasoning and Residual: Same as Study #1, Factor II, plus
residual.
STUDY #3 - SECOND FOLLOW-UP: In the third study, Olckers subjected a new sample of schoolchildren to
a different test battery. This time the battery contained 21 different types of
arithmetic and other test, as follows …..
Test 1 - Speed of Movement: This was
a test of how many short lines a subject could draw relatively accurately on a
piece of paper in 15 seconds, averaged over three attempts.
Test 2 - Repetition of Digits: Same as
Study #2.
Test 3 - Four Rules A: Same as
Study #2.
Test 4 - Dividing Geometrical Figures: This was
a 4 minute 20-item test of spatial recognition and analysis. In each item, a complex
geometric figure was positioned alongside two component figures, and the
subject had to trace the cutting line on the original needed to produce the two
components.
Test 5 - Definitions: Same as
Study #2.
Test 6 - Fractions: Same as
Study #2.
Test 7 - Opposites: Same as
Study #2.
Test 8 - Four Rules B: Same as
Study #2.
Test 9 - Change: Same as Study #2.
Test 10 - Number Series: Same as
Study #2.
Test 11 - Alphabet: Same as
Study #2.
Test 12 - Three Digit Addition: This was
a 4 minute test of 3-digit addition.
Test 13 - Maze: This was a 3 minute
non-verbal reasoning test requiring subjects to trace their exit route on a
series of 30 mazes of increasing difficulty.
Test 14 - Problems: Same as
Study #2.
Test 15 - Reading: This was a 3 minute
test of silent reading in which comprehension was periodically assessed by
multiple-choice Cloze questions.
Test 16 - Reasoning: This was
an 8 minute 15-item test "which could only be solved by eliminating the
wrong answers", and which was intended to test "a different type of
reasoning from that involved in arithmetical problems" (Olckers, p13) [no further details given].
Test 17 - Part-Whole: This was
a two part test, with 2 minutes allowed for each part. Firstly subjects had to
select from lists of five options the names of parts of named objects, and
secondly they had to select the objects which matched named parts.
Test 18 - Cancellation of 9s: Same as
Study #2.
Test 19 - Compound Words: This was
a 5 minute 90-item test in which given words had to be extended to form
compound words [e.g. given "table" add "cloth" to make
"tablecloth"].
Test 20 - Spelling: A spelling test
taken from the Cape Departmental Standard VI Examination, with words either
presented alone or in sentence context.
Test 21 - Arithmetic: A total of
2 hours arithmetic testing taken from the Cape Departmental Standard VI
Examination.
The
test scores were then correlated as before .....
|
|
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
13 |
14 |
15 |
16 |
17 |
18 |
19 |
20 |
|
1 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
2 |
.22 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
3 |
.32 |
.20 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
4 |
.09 |
.11 |
.19 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
5 |
.11 |
.08 |
.24 |
.20 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
6 |
.13 |
.13 |
.35 |
.20 |
.29 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
7 |
.19 |
.19 |
.31 |
.23 |
.25 |
.28 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
8 |
.11 |
.09 |
.50 |
.16 |
.23 |
.45 |
.30 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
9 |
.05 |
.12 |
.41 |
.20 |
.32 |
.48 |
.18 |
.44 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
10 |
.02 |
.13 |
.07 |
.40 |
.31 |
.41 |
.20 |
.26 |
.37 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
11 |
.12 |
.24 |
.24 |
.26 |
.18 |
.29 |
.45 |
.23 |
.20 |
.23 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
12 |
.04 |
.08 |
.53 |
.10 |
.15 |
.36 |
.28 |
.68 |
.44 |
.15 |
.25 |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
13 |
.06 |
.05 |
.07 |
.45 |
.14 |
.11 |
.15 |
.13 |
.16 |
.32 |
.20 |
.14 |
- |
- |
- |
- |
- |
- |
- |
- |
|
14 |
.15 |
.16 |
.19 |
.26 |
.40 |
.38 |
.16 |
.29 |
.55 |
.49 |
.23 |
.18 |
.18 |
- |
- |
- |
- |
- |
- |
- |
|
15 |
.10 |
.19 |
.20 |
.15 |
.23 |
.25 |
.38 |
.21 |
.17 |
.18 |
.27 |
.24 |
.23 |
.16 |
- |
- |
- |
- |
- |
- |
|
16 |
.05 |
.17 |
-.12 |
.14 |
.23 |
.21 |
.20 |
.15 |
.28 |
.29 |
.19 |
.05 |
.18 |
.31 |
.21 |
- |
- |
- |
- |
- |
|
17 |
-.04 |
.09 |
.18 |
.25 |
.41 |
.20 |
.43 |
.23 |
.22 |
.28 |
.30 |
.15 |
.32 |
.12 |
.45 |
.30 |
- |
- |
- |
- |
|
18 |
.16 |
-.01 |
.31 |
.23 |
.01 |
.13 |
.13 |
.28 |
.01 |
.00 |
.11 |
.24 |
.14 |
.19 |
.06 |
.01 |
.15 |
- |
- |
- |
|
19 |
.09 |
.09 |
.22 |
.21 |
.30 |
.19 |
.35 |
.20 |
.18 |
.19 |
.24 |
.12 |
.24 |
.15 |
.32 |
.24 |
.45 |
.14 |
- |
- |
|
20 |
.21 |
.23 |
.22 |
.04 |
.24 |
.25 |
.25 |
.28 |
.21 |
.18 |
.32 |
.26 |
.00 |
.26 |
.31 |
.11 |
.13 |
-.09 |
.16 |
- |
|
21 |
.08 |
.18 |
.26 |
.19 |
.37 |
.52 |
.14 |
.34 |
.51 |
.46 |
.20 |
.27 |
.21 |
.62 |
.24 |
.22 |
.16 |
-.14 |
.10 |
.37 |
|
NB: Colour-coding as for Study #1. |
||||||||||||||||||||
As
before, the correlation matrix was then factored down, and on this occasion
suggested the following six factors .....
Factor I - "Concentration and Retentive
Power": As in
Studies #1 and #2.
Factor II - Speed of Performance: As in Studies #1 and #2.
Factor III - Manipulation: As in Study #2, Factor III.
Factor IV - Reasoning: As in Study #2, Factor IV.
Factor V - "Fluency of Verbal Association
Under Restriction": This is the ability for precise use of verbal material such as reading,
vocabulary, definitions and opposites.
Factor VI - Visual Imagery and Residual: This is the ability to see the
spatial characteristics of the problem at hand, and to use visuo-spatial
intelligence to manipulate them.
3 - The Author's Conclusions
The
first conclusion was that "similar factor patterns were revealed in all
three [of] the analyses" (Olckers, p19). The four most consistently seen
factors were Concentration, Speed,
Manipulation, and Reasoning. Olckers' second conclusion was that
"there is no evidence of a general common factor in Spearman's sense"
(Olckers, p20). In other words, there was no need to postulate a g-factor for
mathematics, because the s-factors listed above account for the majority of the
variance [although Olckers allowed for the possibility that this may not be the
case in slightly younger children].
4 - References
See
the Master References List
[Home]