Construct Validity in Psychological Tests

1
Construct Validity in Psychological Tests

• Lee J. Cronbach and Paul E. Meehl (1955)
• ??? 9737505

2
Four Types of Validity

• 1. Predictive, Concurrent, Content, and Construct
  Validity.
• Predictive the operationalizations ability to
  predict something it theoretically should.
• Concurrent ability to distinguish between groups
  that it should theoretically distinguish.
• (ex farm workers/dictation)

3

• Content we check the operationalization against
  the relevant content domain. Have we included all
  the domains essential for this construct?
• (ex intelligence)
• Construct an assessment of how well you
  translated your ideas/theories into actual
  measures.
• (??)
• Criterion Oriented Predictive and Concurrent
  Validity
• Concern of behavior Content and Construct
  Validity

4
(No Transcript)
5

• 2. Example of construct validation procedure
• palmar sweatings correlation with anxiety
• (0.5) academic aspiration
• (0.45) frat brothers
• (0.55) intellectual inefficiency

6
The Relation of Constructs to Criteria

• All operations have an unquestionable criterion
• Asymmetry between tests and criterion
• (ex Compulsive Rigidity and dot counting)

7
Inadequacy of Validation in terms of Special
Criteria

• If 2 tests correlate highly, then an evaluation
  of the predictive power of the older test may be
  used for the new one
• Clinical field finds it hard to find a criterion

8
Special Criteria Bootstraps Effect

• Vague concept (observations) ? Empirical
  discovery ? another observation with better
  reliability correlation
• Notion of temperature rise from mercury
  expansion
• test sense of touch? taking temperature
• criterion feels hotter? mercury expansion
• 1. statistical relation
• 2. agreement of interobserver
• 3. regularity
• 4. explainable

9

• Binet Scale vs IQ test
• If no error in test procedure can be argued, we
  treat the test score as validour tasks (are)
  finding other variables.

10
Experimentation to Investigate Construct Validity

• Group differences (ex Churchgoing)
• Correlation matrices and factor analysis
• Internal structure
• Change over occasions (test-retest)
• Process of performance
• (ex test of math competence)

11
The Nomological Network

• nomological? lawful
• Definition must be provided as evidence to prove
  your construct has construct validity
• Contents

Theoretical Framework
Empirical Framework
12

• Fundamental Principles
• 1. Make clear what something is so that laws can
• be set
• 2. The laws may relate 3 different attributes
  within the net
• 3. Learn more about a construct to elaborate the
  network
• 4. Different operations overlap if they are
  from the same construct variable

13
Recapitulation

• A construct is defined by a network
• Some statements in the network must lead to
  predicted relations among observables
• Many types of evidence are relevant
• When proposition fails to occur, fault might lie
  in test or network

14

• Cannot be expressed with one coefficient (upper
  lower bounds)
• Constructs may vary or make identifications with
  others
• Investigation of construct validity is similar to
  developing theories