Validity

• Definition: Judgment about how adequately a test measures what it claims to measure.

• Validation: The process of gathering and evaluating evidence about validity.

Types of Validity

1. Construct Validity

   • Measures how test scores relate to other test scores and how they fit within a theoretical framework.

   • Example Constructs: Intelligence, anxiety, personality, self-esteem, motivation, creativity.

   • Questions to consider:

     • Is the test measuring what it claims?

     • Does the construct change with age as predicted?

   • Evidence of Construct Validity: Age differentiation, convergent evidence, discriminant evidence, evidence from distinct groups, factor analysis, evidence of pretest-posttest changes.

2. Criterion-Related Validity

   • Evaluates the relationship between test scores and other measures (criteria).

   • Types:

     • Concurrent Validity: Correlation with existing criteria evaluated at the same time.

     • Predictive Validity: Ability of a test to predict a future outcome.

   • Example Constructs:

     • Construct: Intelligence / Criterion: GWA, exam scores.

     • Construct: Aggression / Criterion: # of school offenses.

   • Considerations: The phenomenon of criterion contamination and implications on test effectiveness.

3. Content Validity

   • Ensures that a test covers the behavior domain it is intended to measure.

   • Involves expert review of test items against objectives.

   • Test Blueprint: Specifications on content areas, the number of items to be covered, etc.

   • Face Validity: Concerns the relevance and presentation of test items.

   • Content Validity Ratio (CVR): A numeric value indicating the validity of test items based on expert ratings. CVR of at least 0.78 is considered valid.

Challenges & Solutions in Validity Testing

• Age Differentiation Evidence:

  • When measuring constructs expected to change with age, the test should reflect these changes.

  • Solution: Administer tests across different age groups and analyze score variations.

• Convergent Validity Evidence:

  • Test should show correlations with other established measures of similar constructs.

  • Solution: Correlate the new test with established tests to see similarity.

• Discriminant Validity Evidence:

  • Invalidities occur if test scores correlate with unrelated constructs.

  • Solution: Ensure that the test does not show significant correlations with unrelated tests.

• Factor Analysis:

  • Helps determine how many dimensions or factors a test consists of.

  • Types:

    • Exploratory Factor Analysis: Identifies underlying factors and their correlation with test items.

    • Confirmatory Factor Analysis: Tests how well the predicted model fits actual data.

Additional Concepts

• Incremental Validity: The additional explanatory power from incorporating a new predictor compared to existing predictors.

• Test Bias:

  • Systems that prevent impartial measurement, leading to unfair or invalid assessments.

  • Types of rating errors (e.g., leniency error, severity error, central tendency error) can affect judgment.

• Test Fairness: The impartial and equitable use of a test in various contexts.

Conclusion: When to Use Different Types of Validity

• Use Construct Validity to check if a test measures what it claims.

• Use Criterion-Related Validity to predict future behaviors or outcomes.

• Use Content Validity to ensure test participants take the test seriously and it covers the necessary content areas.

Final Thoughts

• Validity studies are crucial for the psychological assessment process, ensuring that tests not only measure what they intend but also predict relevant behaviors and are fair in their administration.