Lecture 2 Measurement of Intelligence Flashcards
Wechsler tests (1939)
Wechsler (1939): Wechsler-Bellevue Scale
standardised on 1500 adults
Revised as two tests (1955):
Wechsler Adult Intelligence Scale (WAIS)
Wechsler Intelligence Scale for Children (WISC)
All administered on a one-to-one basis
Widely used today, now the WAIS-IV and WISC-V
Improvement on previous intelligence tests:
Single set of tests: Designed so that people of different ages could take them
The idea of ‘deviation IQ’
Deviation IQ: based on how people score relative to the average scores of other people of the same age
IQ = Test score/Expected score for age X 100
Deviation IQ
Two steps to standardised scores:
Determine the expected score for each age
Transform the scores to a standardised form
Wechsler transformed the data for each age band so that IQ is normally distributed, with a mean of 100 and a standard deviation of 15
Verbal comprehension index
Vocabulary
What do certain words mean?
chair, hesitant, presumptuous
Similarities
What do two things have in common?
In what way are a lion and a tiger alike?
Information
(general knowledge)
Covering people, places and events
Name three oceans
Comprehension (understanding verbal info)
Everyday life problems
Why do people require driving licences?
Working memory index
Arithmetic
(mental arithmetic problems)
If you have 15 apples and give 7 away, how many are left?
Digit span
Repeat digits in exact or reverse order (2 – 9)
3 – 9 – 1 – 7 – 4 – 5
Letter-number sequencing
Repeat numbers then letters, in numeric / alphabetical order
F – 7 – K – 2 – E – 8
Perceptual Reasoning Index
Picture completion, block design, matrix reasoning
Processing Speed Index
Digit-symbol coding (change symbols to numbers)
Symbol search (search for a symbol)
The Wechsler tests for children
Wechsler Intelligence Test for Children (WISC-V)
for children aged 5 - 16 years old
Wechsler Primary and Preschool Scale of Intelligence (WPPI- IV)
for children 2 - 7 years old
Similar tasks to the WAIS. Takes around 1 hour
Full-scale IQ plus 5 index scores:
Verbal Comprehension Index
Visual–Spatial Index
Fluid-Reasoning Index
Working Memory Index
Processing Speed
The Stanford-Binet Intelligence Scale
Last week: original developed in 1916 by Lewis Terman
Recent versions can be used with children (age 2+) and adults (up to 85 years old)
Administered one to one with
a psychologist
SB5 (2003) maps onto
CHC theory
Raven’s Progressive Matrices
Raven’s matrices are based ‘g’
Test abstract ability
free of culture
free of language influences
Non-verbal problems
Abstract reasoning
Require no knowledge of a particular culture
Summary of Raven’s Progressive Matrices
Intelligence assessed by ability to work out the underlying rules
60 items of increasing difficulty (used on all ages)
Overall IQ: deviation from a standardized score
Psychometric vs. cognitive approaches to intelligence
WAIS, Stanford-Binet and Raven’s matrices are all examples of Psychometric approaches to intelligence
-Based on the findings of factor analytic studies
-Look at the psychometric properties of intelligence scales
Cognitive approach to intelligence focusses on biological and physiological processes
Cognitive and biological measures
Simple biological and physiological factors may be good indicators of intelligence
Advantages:
-quick and easy to administer
-Measures abilities that underlie performance on psychometric tests
brain size and elementary cognitive tasks
Brain size
Tiedmann (1836) suggested that there is an “indisputable” link between brain size and mental energy
Magnetic resonance imaging
(MRI) scans have made it possible
to test this hypothesis
McDaniel (2005) Meta analysis of 37 studies: correlation between brain size and IQ to be 0.33
Causation?
Does a large brain cause
high IQ, or does having a
high IQ create a large brain?
Cognitive approach to intelligence
Suggests that ‘g’ is supported by multiple cognitive mechanisms
-Working memory
Maintain, update, and manipulate information
-Processing speed
Speed at which cognitive operations can be performed
-Explicit associative learning
Ability to remember associations between stimuli
Multiple Cognitive mechanisms
Included in factor-analysis theories of intelligence (e.g., CHC has measures of speed, short term memory, etc.)
All are correlated with ‘g’
people with high ‘general intelligence’ tend to have good working memory, fast processing speed and are good at learning
Elementary cognitive tasks (ECTs)
Simple tasks that measure cognitive processes
Understanding and identifying a stimulus
Discrimination tasks
Visual Search tasks
Memory tasks
Assessed using Reaction Times (RT), Standard Deviations in Reaction Times (SDRT) and Inspection Times (IT)
Reaction times on ECTs correlates with IQ test performance (around 0.35)
Jensen (1998) ECTs are more accurate and analytical than IQ tests - not dependent on past learning
What makes a good IQ test?
Sattler (2002): three main qualities that good intelligence tests have in common
A variety of tasks
Standardisation of administration
Norm referencing
Also need good reliability (consistent results) and validity (they measure what they say they measure)
Variety of tasks
To assess a full range of abilities: need a number of different tasks e.g., 12 subtests on the WAIS, 10 subtests on SB5
Provides fuller understanding of overall intelligence
Also highlights strengths and weaknesses
Standardisation of administration
Important to have control over the intelligence test environment
Setting: Comfortable setting, no distractions e.g. noise in the testing room, number of people present
Administration: Instructions, Wording of questions, consistency in the time taken to complete the test, supervised by a qualified administrator
Norm referencing
A good intelligence test needs to have ‘norms’
Based on the average score for people in each age group – large numbers needed
It is then possible to compare an individual’s intelligence score with the expected score on the test for a particular age group
Reliability and validity
Reliability: Does the test consistently measure intelligence?
Validity: Does the test measure intelligence?
Reliability
There are two types of reliability we are interested in when considering intelligence tests:
Internal reliability – positive correlation between items on test – assesses whether they are measuring the same construct
Test-retest reliability – a good intelligence test will show good reliability across time
Internal reliability
Developers of standardised intelligence tests focus on ensuring high internal reliability
i.e., they remove items that do not positively correlate with other items
On tests such as the WAIS and Stanford-Binet, internal reliability is likely to be high
Reliability over time
Short-term IQ test fluctuation: up to 15 IQ points (Benson, 2003)
BUT: longitudinal studies suggest that IQ is reasonably stable over long-term
-Berkeley growth study – longitudinal study looking at people in Berkeley born between 1928 and 1929
-Children tested yearly on a number of intelligence tests
-IQ scores at 18 years of age positively correlated with those at 12 years (0.89) and 6 years (0.77)
Reliability over time
Similarly, Deary et al. (2000) found a strong correlation between IQ at 11 and 77 years old (r = 0.73)
Validity of intelligence tests
Validity = does the test measure what it claims to measure?
Do intelligence tests measure intelligence?
Three types of validity:
-face validity
-concurrent validity
-predictive validity
Face validity
‘On the face of it’, does the intelligence test appear to measure what it sets out to measure?
IQ tests undergo substantial development work
- an item that requires mental
arithmetic probably does measure
mental arithmetic ability
Concurrent validity
Refers to the relationship the intelligence test has to other intelligence tests
A good intelligence test should correlate well with other intelligence tests
Predictive validity
Predictive validity refers to whether the test can predict future performance
A good intelligence test has to be able to predict future performance
Do intelligence tests measure intelligence?
Tend to have all three types of validity
But depends on the definition of intelligence used by the test developers
Benson (2003) Many tests based on ‘g’, but still debate over whether ‘g’ exists
Boring (1923)
“Intelligence as a measurable capacity must at the start, be defined as the capacity to do well in an intelligence test. Intelligence is what the tests test”
Gardner’s multiple intelligences evaluation
But… very time consuming to evaluate all of them
Little research on validity and reliability
Salovey & Mayer’s (1990) model
Four-branch model
perceiving, facilitating, understanding, managing
Mayer-Salovey-Caruso Emotional Intelligence Test version 2, (MSCEIT2),
141-item scale, produces an overall score of emotional intelligence plus subscale scores
Perceiving: ability to identify emotion
Facilitation: Ability to use emotion
What mood(s) might be helpful to feel when meeting in-laws for the first time?
Understanding: ability to understand emotion
Managing: ability to manage emotions
Goleman’s theory of emotional intelligence
Measured using the Emotional Competence Inventory (Goleman and Boyatzis, 2005)
designed for use in the workplace
360° instrument – other people evaluate individuals on their emotional intelligence
e.g., the assessor determines whether the person:
respects, treats with courtesy and relates well to people of diverse backgrounds
accurately reads people’s moods, feelings or non-verbal cues
Emotional-social intelligence model
Emotional Quotient Inventory (EQ-i; Bar-On, 1997)
First measure of emotional intelligence
Scores can be converted into standard scores, mean of 100 and s.d. of 15
133 items, e.g., “I’m in touch with my emotions”
Five-point response scale
very seldom or not true of me (1) to
very often true of me or true of me (5)
Tests of emotional intelligence evaluation
No objective measure for validity of emotional intelligence tests
Less research, so less information on reliability and validity
What else is missing?
What else contributes to the ability to perform tasks?
What else affects success and achievement?
Summary
Typically, IQ tests measure a specific type of problem-solving, verbal, mathematical ability
This ignores many other traits that could be considered ‘intelligent’
Practical difficulties in measuring multiple intelligences
Emotional intelligence scales do exist but are difficult to validate
Several other things could be considered (e.g., creativity)
