Psychometrics: Validity Overview

Study Notes

Validity

Validity measures how well a test measures its intended target in a specific context.
There is no universal validity; tests can be valid only for specific uses and populations.
High validity coefficients generally correlate with acceptable reliability coefficients.

Types of Validity

Face Validity
Content Validity
Criterion Validity
Construct Validity

Face Validity

Evaluates whether a measure superficially appears to reflect the intended construct.
It is not based on established theories; relies on a cursory review.

Content Validity

Assesses the alignment of a measure with the theoretical domain of a construct.
Important for specific abilities; less useful for broad psychological constructs.
Example: A math test lacking addition questions would not exhibit high content validity.

Quantifying Content Validity

Lawshe's method involves raters assessing the relevance of each item as essential, useful, or unnecessary.
The Content Validity Ratio (CVR) quantifies the agreement among raters; positive CVR indicates majority agreement.

Criterion Validity

Measures the correlation of an instrument with specific outcome criteria.
Includes two forms:
- Concurrent Validity: Correlation with another measure at the same time (e.g., essay vs. multiple-choice tests).
- Predictive Validity: Correlation with future performance (e.g., SAT predicting GPA).

Validity Coefficient

Indicates the correlation between a test score and the criterion score.
Methods include simple regression and multiple regression for assessing incremental validity.

Incremental Validity

Refers to the additional explanation provided by new predictors beyond existing ones.
Example: Assessing if adding study hours improves the predictive power of SAT scores on GPA.

Construct

Constructs are unobservable concepts used to explain behaviors (e.g., self-esteem, depression).
Construct Validity measures how effectively a test assesses a psychological construct, distinct from established abilities.

Evidence of Construct Validity

Tests demonstrate homogeneity and consistent measurement of a single construct.
Scores should vary predictably with age, time, or experimental changes (e.g., effects of antidepressants).
Top-down theoretical predictions can be validated through score variation among distinct groups.

Convergent and Divergent Validity

Convergent Validity assesses how well the test relates to other related measures (e.g., IQ and achievement).
Discriminant Validity examines the lack of correlation with unrelated constructs, ensuring distinct measure characteristics.

Description

This quiz covers the concept of validity in psychometrics, explaining how well tests measure their intended targets in specific contexts. You will explore types of validity, including face, content, criterion, and construct validity, as well as methods for quantifying content validity. Gain a deeper understanding of these essential measurement principles in psychology.