Unit 2 Review

1
Unit 2 Review

• Chapters 7-9, 11

2
Theories of Intelligence

• Two classical theories of intelligence
• Spearmans g or two-factor theory
• Thurstones 9 Primary Mental Abilities
• Guilfords Structure of Intellect is another
  multi-factor theory of intelligence
• Also known as the psychometric theories of
  intelligence because of reliance on data
  relationships

3
Chapter 7

• Theories of Intelligence

4
Spearman

• Developed first formal theory about human mental
  ability
• One general, g, factor accounted for correlations
  among tests of simple sensory functions
• Each test also had a specific component, s,
  unique to that test plus error

5
Thurstones Primary Mental Abilities

• S spatial
• P perceptual (esp. speed of visual perception)
• N Numerical (speed accuracy of computation)
• V Verbal M - Memory

• W Words (word fluency or disarranged words)
• I Induction (finding a rule)
• R Reasoning (arithmetic)
• D Deduction (application of rule)

6
Hierarchical Models

• Are compromise between and one vs. many argument
• Acknowledge there are many separate abilities but
  can be arranged so only a few dominant factors
  are at the top of the hierarchy.
• Includes Cattell (crystallized and fluid), Vernon
  (verbal educational and spatialmechanical), and
  Carroll (three stratum theory with g as the top
  stratum)

7
Other Theories

• Developmental (e.g., Piaget)
• Information Processing--based on elementary
  cognitive tasks, ECT, such as reaction time
  (e.g., Jensen, Sternberg)
• Biological Theories (e.g., Gardners theory of
  multiple intelligences)

8
Differences by Sex

• Differences minimal on total scores
• Males outperform females on tests of spatial
  ability (effect size of 5-7)
• Females outperform males on verbal tests during
  childhood and much of adolescence
• Greater variability in intelligence for males

9
Differences by Racial/Ethnic Group

• Compared to whites
• Blacks are about 1SD below
• Hispanics and native Americans are .5-1 SD below
  on verbal and at the mean on performance tests
• Asians have a similar verbal mean and are about 1
  SD above on non-verbal tests

10
Heredity and Environment

• Intelligence results from an interaction of
  heredity and environment
• Estimates of heritability range from .4 to .8
  (median of about .5 or .6)
• Heritability increases with age
• g has a higher heritability index than specific
  abilities

11
Correlation of IQ scores From Bouchard McGue
(1981)

• Identical twins reared together .86
• Identical twins reared apart .72
• Same sex fraternal twins reared together .62
• Opposite sex fraternal twins reared together
  .57
• Non-twin siblings reared together .47
• Unrelated (adopted) siblings reared together .30

12
Individual Tests of Intelligence

• Chapter 8

13
Common Characteristics of Individual Intelligence
Tests

• individually administered
• administration requires advanced training
• tests cover wide range of age and ability
• examiner must establish rapport
• immediate scoring of items
• usually requires about one hour
• allows opportunity for observation

14
Two Main Individually Administered Intelligence
Tests

• Stanford-Binet
• He wanted to create a process for identifying
  intellectually limited children so they could be
  removed from the regular classroom and put in
  special education.
• Wechsler scales
• Developed in response to the perceived
  shortcomings of the Stanford-Binet

15
Early Binet Scales

• 1905 30 items ordered by difficulty. Test
  lacked
• adequate measuring units to express results
  (only used idiot, imbecile, and moron)
• adequate normative data (only used 50 subjects)
• evidence of validity
• 1908 Grouped items according to age level
  rather than simply increasing difficulty.
  Introduced concept of mental age.
• Increased norm group to 203.
• Criticized because it produced only one score
  almost exclusively related to verbal, language,
  and reading ability

16
Modern Binet scale

• Totally revised in 1986 by Thorndike et al.
• Used Thurstones multidimensional model (1938)
• G made up of crystallized ability (verbal
  quantitative reasoning), fluid-analytic abilities
  (abstract-visual reasoning) and short term memory.

17
Structure of the SB-IV

• Verbal reasoning included vocabulary test,
  comprehension test, absurdities test, and verbal
  relations test.
• Abstract-visual reasoning included pattern
  analysis test, copying test, matrices test,
  paper-folding and cutting test.
• Quantitative reasoning included quantitative
  test, number series test, equation-building test.
• Short-term memory included bead memory, memory
  for sentences, memory for digits, and memory for
  objects
• Composite included all areas combined

18
Psychometric properties of SB-IV

• Standardization sample stratified based on 1980
  census geographic region, community size,
  ethnic group, age, and gender.
• Internal consistency reliability is .98 for
  composite and .93-.97 for area scores. Some
  individual test scores are lower .73 for memory
  for objects is the lowest.
• Test-retest reliabilities for composite score
  were .91 and .90 for 5 and 8-year-olds.
• Factor analysis supports the structure of the
  test.
• Correlations with other IQ tests are generally in
  the 70s and 80s

19
Wechsler Scales

• David Wechsler worked at NYs Bellevue Hospital.
  He wasnt happy with the Stanford Binet with its
  focus on children or on the production of a
  single score.
• In 1939, he created the Wechsler-Bellevue, later
  called the WAIS.
• In 1949, he created the childrens version, the
  WISC.
• In 1967, he added the WPPSI for children ages
  2.5-7.

20
Structure of the WAIS

• The WAIS yields separate verbal and performance
  IQs
• The WAIS-III has four index scores Verbal
  comprehension, working memory, perceptual
  organization, and processing speed.

21
Scales and Norms for the WAIS

• Determine raw score for each subtest.
• Convert raw scores to standard scores, called
  scaled scores (M10, SD3)
• There are conversions for 13 age groups. This
  method of conversion obscures any differences in
  performance by age.
• Subtest scaled scores are added, then converted
  to WAIS-III composite scores.
• Three composite scores verbal, performance,
  full scale, each with M100, SD15
• Four index scores verbal comprehension,
  perceptual organization, working memory,
  processing speed

22
Reliability of the WAIS

• Internal consistency and test-retest
  reliabilities are about .95 or higher for full
  scale and verbal scores.
• Theyre about .90 for performance and three other
  index scores perceptual organization, working
  memory, and processing speed.
• Internal consistency reliability for the
  subtests range from upper .70s to low .90s.
  Test-retest is about .83.
• Generally, performance reliabilities are lower
  than verbal reliabilities on the subtests.

23
Validity of the WAIS

• Great deal of information on criterion-related
  and construct validity.
• Factors analyses support use of 4 index scores.
• Comparison studies show the pattern of WAIS-III
  scores for many special groups, e.g., Alzheimers
  Disease, Parkinsons, learning disabled, brain
  injury.
• Is the top test used today

24
Group Differences in IQ

• Test scores that demonstrate differences among
  people may suggest that people are not created
  with the same basic abilities.
• Biggest problem Some ethnic groups obtain lower
  average scores on some psychological tests. On
  average African Americans score 15 points lower
  than whites on IQ tests.
• Dispute is not whether differences occur but why
  they occur.environment vs. biology

25
Problems with Biology Argument

• IQ scores are improving (called the Flynn
  effect), more so for African Americans than
  whites.
• Victimization by stereotyping could affect test
  performance and grades.
• Construct of race has no biological meaning based
  on evidence from studies in population genetics,
  the human genome and physical anthropology.

26
Criticisms related to Content Validity

• Looking at specific items, it was thought that
  they might be biased because some children
  wouldnt have the opportunity to learn the
  material
• Members of ethnic groups might answer some items
  differently but still correctly
• Scores affected by language skills inculcated as
  part of a white, middle-class upbringing foreign
  to inner city children

27
Responses to Content Validity Criticisms

• Some evidence suggests that the linguistic bias
  in standardized tests does not cause the observed
  differences (Scheuneman, 1987).
• Elimination of biased items from a test didnt
  change the test scores (Bianchini, 1976).
• Cant find classes of items most likely to be
  missed by minority group members (Wild, et al.,
  1989)

28
Group Tests of Mental Ability

• Chapter 9

29
Characteristics of Group Mental Ability Tests

• Administered to a large group
• Composed of multiple choice items that can be
  machine-scored
• Content similar to individual tests
• Fixed time limit and number of items
• Usually yield a total score and some subscores
• Principal purpose is prediction

30
Advantages of individual tests

• Provide information beyond the test score
• Allow the examiner to observe behavior in a
  standard setting
• Allow individualized interpretation of test scores

31
Advantages of group tests

• Are cost-efficient
• Minimize professional time for administration and
  scoring
• Require less examiner skill training
• Have more objective and more reliable scoring
  procedures
• Have a very broad application
• Group tests far outnumber individual tests and
  group tests vary widely among themselves

32
Scoring Information for the OLSAT7

• Yields verbal, Nonverbal, and Total scores
• Converted to School Ability Index (SAI) with
  M100 and SD16
• SAIs determined separately for age groups at 3
  month interval from ages 5-19
• Score reports also include anticipated
  achievement comparisons (AAC) to predict
  performance on the Stanford tests

33
Psychometric Properties

• About half a million cases are part of the
  research base for the OLSAT7
• High internal consistency with nothing lower than
  .87 for total score (higher at higher grades)
• KR-20 for Verbal and Nonverbal in the high .80s
  for upper grades and low .80s for lower grades
• No test-retest reliability data
• High correlations between the OLSAT7 and the
  Stanford tests, but other validity evidence is
  weak

34
College Admissions The SAT

• The College Board oversees the development of the
  test called the SAT
• ETS (Educational Testing Service) actually
  develops the SAT
• The SAT is a cluster of tests
• SAT I includes the well-known Verbal and Math
  tests
• SAT II has tests in 12 subject fields

35
SAT I Structure

• Includes verbal (SAT-V) and math (SAT-M) summed
  to get total score
• Uses correction for guessing
• Range for each subtest is 200-800 with M500 and
  SD100. Total M1000 and SD200
• Norms based on test users, not any well-defined,
  predetermined population
• Scaled score norms last determined in 1994.
  Percentile norms adjusted on an annual basis.

36
Reliability and validity of the SAT

• Internal consistencies of .91-.93
• SEMs of about 30 points
• Poor predictive power regarding grades of
  students scoring in middle ranges
• Number of English or math units doesnt correlate
  significantly (maybe due to coaching)
• Validity coefficients are about .40 with 1st year
  grades
• On old SAT, African-American and Latino students
  scored lower, sometimes by as much as 80 points.
  New test MAY have reduced that.

37
The ACT

• ACT provides more emphasis on school-based skills
• Have scores for English, Math, Reading, Science
  Reasoning, and Composite which is an average of
  the 4 tests
• Does NOT use a correction for guessing
• Score range is 1-36 with M20 and SD5

38
Psychometric properties of the ACT

• Norms based on users, usually about a million
  annually
• Reliabilities ranges from .84-.91 with the
  Composite score reliability being .96
• SEMs are 1.5-2 points for each subtest and 1
  point for the Composite score
• Correlates about .80 with SAT
• Like SAT, high school GPA is generally as good a
  predictor as test scores

39
Graduate and Professional School Selection The
GRE

• Includes General Test, Subject Test, and Writing
  Assessment
• General tests includes Verbal, Quantitative, and
  Analytical
• General tests intended to measure developed
  abilities that have been acquired over a long
  period of time
• Gradually moving to computer adaptive testing

40
GRE Scores and Norms

• Scaled score set at M500 and SD100, with a
  range of 200-800
• Like SAT, average scores have gradually drifted
  downward
• Norms are user norms
• Internal consistency is in the low 90s
• SEMs are about 30-40 points

41
Validity of the GRE

• GRE tests correlate with first-year GPA in the
  range of mid-20s to low 30s.
• Tests correlate with each other. Lowest
  correlation is V-Q at .45, higher is Q-A at .66
• Undergraduate GPA is a better prediction than any
  of the tests and about equal to total test score
  in predictive validity.

42
Ravens Progressive Matrices

• Designed to measure the fluid dimension of
  intelligence may be best measure of g
• Not used more widely because
• too many manuals and norm groups
• conflicting evidence about what the test is
  measuring
• Hasnt really eliminated differences between
  majority and minority group examinees

43
Achievement Tests

• Chapter 11

44
Achievement vs. Aptitude tests

• Achievement tests
• Evaluate the effects of a known or controlled set
  of experiences
• Evaluate the product of a course of training
• Rely heavily on content validation procedures
• Aptitude tests
• Evaluate the effect of an unknown, uncontrolled
  set of experiences
• Evaluate the potential to profit from a course of
  training
• Rely heavily on predictive criterion validation
  procedures

45
Classification of Achievement Tests

• Achievement battery
• Single area achievement tests
• Certification, licensing exams
• State, national, international tests
• Psycho-educational batteries
• Teacher-made tests

46
Establishing cutscores

• Norm-referenced approach
• select a percentile and everyone above that point
  is in
• Criterion-referenced approach
• many approaches
• Most popular approach is Angoffs where judge
  looks at each item and assesses the probability
  of minimally competent person getting it right.
  Probabilities summed to get total score
• Original judgments often changed after results
  are seen

47
TIMSS design

• TIMSS 1999 used a matrix sampling technique to
  achieve broad coverage the total of 308 items
  were systematically distributed across 8 test
  booklets and the booklets were distributed
  randomly to students
• Each student completed one 90-minute test
  booklet.
• Approximately one-third of the items were
  constructed-response format, and the remaining
  items were multiple-choice

48
Nagging questions about achievement tests

• Is there some other way to measure content
  validity?
• Is there really a difference between achievement
  and ability?
• How motivated were examinees to perform well?
• How can we get diagnostic information from a
  short test?
• What is the difference between constructed
  response and selected response? Is it important?