INTELLIGENCE, THINKING AND PERSONALITY

1
INTELLIGENCE, THINKING AND PERSONALITY

• Factor Analytic Theories of Intelligence

2
BASIS OF THEORIES

• Based on Factor-analytic treatment of
  psychometric tests
• How do scores on individual test items or on
  subtests correlate with one another?
• Do some scores correlate more than others?
• How should these correlations be accounted for?

3
PURPOSE OF THEORIES

• To explain individual differences in intelligence
• NOT (primarily) to explain how intelligent
  behaviour comes about
• i.e. not concerned with the cognitive apparatus
  underlying intelligence

4
SPEARMAN

• Invented the first form of factor analysis
• Proposed a two-factor theory of intelligence
• Every activity involves a general factor a
  specific factor (g s)
• g general intelligence and is innate
• the specific factors need not be innate

5
SPEARMAN - cont

• Spearman had noted that results of different
  intelligence tests (or subtests) almost always
  correlated positively.
• His method of tetrads or the later principal
  components version of factor analysis finds a
  factor (g) that accounts for these correlations.

6
SPEARMAN - cont

• However, this technique can miss clusters of
  strong correlations (e.g. verbal test results
  might go together in one cluster, and
  mathematical tests in another).
• Gould gives a detailed presentation in Chapter 6
  of The Mismeasure of Man

7
DIAGRAMMATIC REPRESENTATION OF TEST RESULTS AND
FACTORS

• Each test is represented as a line
• Angle between lines (actually cosine of angle)
  represents correlation
• Cosine of 0o is 1, so perfect correlation if
  lines are in same direction
• Cosine of 90o is 0, so no correlation if lines
  are at right angles

8
PRINCIPAL COMPONENTSANALYSIS

• Note - in general it is not possible to represent
  all correlations in two dimensions

g
maths tests
Verbal tests
2nd principal component
9
PRINCIPAL COMPONENTS ANALYSIS - cont

• The first principal component is the best single
  summary of the results.
• If the subtests are positively correlated, which
  is what Spearman observed, the first principal
  component will provide a resonable summary.
• Technically, it will account for a substantial
  proportion of the differences between one person
  and another (variance) and more than any other
  possible principal component
• It is Spearmans g factor (general intelligence)
• So, the claim is that differences in g (IQ)
  explain differences in test performance quite
  well.

10
BURT AND VERNON

• Proposed a hierarchical model, with three levels
• g (as in Spearmans model)
• group factors, suggested by the fact that some
  subtests correlate more strongly than others (but
  identified indirectly using the principal
  components method)
• ve, verbal-educational ability
• km,spatial-mechanical ability
• specific factors (as in Spearmans model)

11
PRINCIPAL COMPONENTSANALYSIS (REPEATED)
g
km tests
ve tests
2nd principal component
12
THE SECOND PRINCIPAL COMPONENT

• The second principal component has to be
  orthogonal (at right angles in the diagram) to
  the first).
• If we take the second component to run from left
  (negative) to right (positive), in the example
  given the ve tests load negatively on that
  component and the km tests load positively.
• The second component, therefore, provides an
  indirect representation of group factors.

13
THURSTONE - PRIMARY MENTAL ABILITIES

• Thought g was not real enough
• Disliked the idea of negative loading
• Thought that abilities should be identified
  positively

14
THURSTONE - cont

• Started with 56 different tests and identified(7)
  primary mental abilities, which he original
  wanted to claim were independent
• S - spatial
• P - perceptual speed
• N - numerical reasoning
• V - verbal meaning
• W - word fluency
• M - memory
• I - inductive reasoning

15
THURSTONE - cont

• Argued that g has no real significance, but just
  reflects correlations between test results (which
  may change if different tests are used).
• Spearman and Burt later applied the same argument
  to Thurstones PMAs
• Also argued that the use of a composite score, or
  a system of overall ranking (i.e. an IQ score)
  is inappropriate

16
THURSTONE - cont

• Invented a method of rotating factors so that
  each one fell close to a cluster of strong
  correlations.
• Produces a so-called simple structure (not
  inevitable that a matrix of correlations will
  yield a simple structure).
• Wanted to eliminate negative projections of
  tests onto factors.

17
ROTATED, ORTHOGONAL FACTORS

• Note that factors are at right angles (zero
  correlation) so lie outside clusters of tests if
  all tests positively correlated

Verbal PMA
Maths PMA
18
THURSTONE -cont

• Thurstone later related g to the results of
  second order factor analysis
• He realised that correlated factors would lie
  closer to his clusters than uncorrelated ones.
• i.e. he realised there were correlations between
  his PMAs, even though he originally suggested
  they were independent

19
ROTATED, NONORTHOGONAL FACTORS
Verbal PMA
Maths PMA

• Factors now lie within clusters, but are not
  independent, so harder to interpret

20
WHY SUCH DIFFERENT CONCLUSIONS?

• There is no single correct way to do factor
  analysis
• Spearmans method is bound to find a single
  factor, provided subtests are correlated to some
  degree
• Thurstones will find separate factors if the
  subtests group to some degree

21
WHY SUCH DIFFERENT CONCLUSIONS? - cont

• Exploratory factor analysis (the kind used in
  work on intelligence) is just a mathematical
  technique of data reduction and summarisation
• By itself it cannot show the existence of
  things in the world that correspond to the
  factors it identifies (general intelligence or
  Thurstones PMAs)

22
LATER DEVELOPMENTS

• Within the psychmetric framework
• Guilford - Structure of the Intellect
• Cattell - Crystallised and Fluid Intelligence

23
GUILFORD - STRUCTURE OF THE INTELLECT

• Attempted a more systematic description of the
  Structure of the Intellect
• He classified tasks/tests according to
• content (4 types) e.g. visual figures, verbal
  meanings
• operations (5 types) e.g. evaluation, memory
• products (6 types), e.g. relations, implications

24
GUILFORD cont.

• So, identified 4 x 5 x 6 120 distinct mental
  abilities
• Guilford tried to develop tests for each
• But scores on those tests were often strongly
  correlated

25
CATTELL - FLUID AND CRYSTALLISED INTELLIGENCE

• Cattell 1963 - distinguished between crystallised
  and fluid intelligence, which has subsequently
  been of some importance
• He regarded them as subdivisions of g, and so
  labelled them gf and gc

26
FLUID INTELLIGENCE

• Deals with abstract relations
• Is not taught
• Is culture free
• Declines with age in adulthood
• Represents the ability to deal with new problems

27
CRYSTALISED INTELLIGENCE

• Represents cumulative learning experience
• wisdom?
• expertise?
• Increases with age

28
DOES INTELLIGENCE DECLINE WITH AGE?

• According to Cattell, fluid intelligence
  decreases with age and crystallised intelligence
  increases.
• Previous, cross-sectional work had tended to
  conclude that intelligence in general declines
  with age
• However, this finding was probably an artefact of
  increasing educational opportunities in the 19th
  and 20th centuries

29
DOES INTELLIGENCE DECLINE WITH AGE? - cont

• The older people in the studies had poorer
  education and so less accumulated knowledge, so
  overall their intelligence measured lower than
  expected.
• Younger people had more accumulated knowledge
  that might have been expected because their
  education had been better.