Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A coefficient alpha for test-retest data.

Samuel B Green1

  • 1Division of Psychology in Education, Arizona State University, Tempe 85287-0611, USA. samgreen@asu.edu

Psychological Methods
|May 14, 2003
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Incorporating Uncertainty Into Parallel Analysis for Choosing the Number of Factors via Bayesian Methods.

Educational and psychological measurement·2021
Same author

Proportion of Indicator Common Variance Due to a Factor as an Effect Size Statistic in Revised Parallel Analysis.

Educational and psychological measurement·2019
Same author

Type I and Type II Error Rates and Overall Accuracy of the Revised Parallel Analysis Method for Determining the Number of Factors.

Educational and psychological measurement·2018
Same author

Accuracy of Revised and Traditional Parallel Analyses for Assessing Dimensionality with Binary Data.

Educational and psychological measurement·2018
Same author

An SEM Assessment of the Internal Structure and Predictive Validity of the Abbreviated Early Adolescent HOME Inventory.

Assessment·2018
Same author

The Problem with Having Two Watches: Assessment of Fit When RMSEA and CFI Disagree.

Multivariate behavioral research·2016
Same journal

Bayesian evaluation for latent variable models: A tutorial on computing information criteria and bayes factors with the r package bleval.

Psychological methods·2026
Same journal

A stochastic block prior for clustering in graphical models.

Psychological methods·2026
Same journal

Three-level vector autoregressive models.

Psychological methods·2026
Same journal

Scaling cognitive modeling to big data: A deep learning approach to studying individual differences in evidence accumulation model parameters.

Psychological methods·2026
Same journal

Best practices in multilevel modeling for within-cluster group comparisons: An evaluation of coding strategies reflecting group composition and heterogeneity.

Psychological methods·2026
Same journal

A unified framework for psychometrics in experimental psychology: The standardized generalized hierarchical factor model.

Psychological methods·2026
See all related articles

Transient errors can inflate reliability estimates. A new test-retest alpha method accounts for these errors, providing a more accurate measure of reliability than traditional coefficient alpha or test-retest correlations.

Area of Science:

  • Psychometrics
  • Psychological Measurement
  • Reliability Theory

Background:

  • Transient errors, stemming from fluctuating feelings and moods, can distort reliability estimates.
  • Coefficient alpha, a common reliability measure, often inflates true reliability when transient errors are present.
  • Existing methods may not adequately address the impact of transient errors on test-retest data.

Purpose of the Study:

  • To develop a new reliability estimation method that accounts for transient errors in test-retest data.
  • To introduce the concept of 'test-retest alpha' as a more accurate reliability measure.
  • To provide an alternative method, the 'test-retest split-half coefficient,' when essential tau equivalency is violated.

Main Methods:

  • A true-score model was developed to incorporate transient errors into test-retest reliability analysis.

Related Experiment Videos

  • The 'test-retest alpha' was derived from this model.
  • A 'test-retest split-half coefficient' was proposed as an alternative under specific assumptions.

    • Main Results:

    • The 'test-retest alpha' provides a reliability estimate lower than coefficient alpha when transient errors are present.
    • 'Test-retest alpha' is less affected by item recall compared to traditional test-retest correlations.
    • The 'test-retest split-half coefficient' serves as a viable alternative when essential tau equivalency is not met.

    Conclusions:

    • The proposed 'test-retest alpha' offers a more precise estimation of reliability by addressing transient error effects.
    • This new method enhances the accuracy of reliability assessment in psychological and educational testing.
    • The 'test-retest split-half coefficient' provides flexibility for reliability estimation when assumptions of tau equivalency are not fully met.