Linking outcomes from peabody picture vocabulary test forms using item response models

Area of Science:

• Psychometrics
• Educational Psychology
• Developmental Psychology

Background:

• The Peabody Picture Vocabulary Test (PPVT) is widely used to assess receptive vocabulary.
• Different forms of the PPVT exist, necessitating methods for score comparability.
• Item response theory (IRT) offers a framework for creating common metrics across different assessments.

Purpose of the Study:

• To establish a common latent metric for vocabulary ability using three forms of the PPVT.
• To enable valid comparisons of vocabulary ability across different PPVT forms, samples, and time points.
• To develop a method for equating scores from PPVT-Revised (PPVT-R) Form M, PPVT-3 Form A, and PPVT-3 Form B.

Main Methods:

• Analysis of responses from 2,625 cases (697 individuals) across 459 unique PPVT items.
• Utilized a 2-parameter logistic item response theory (IRT) model for data analysis.
• Included items from PPVT-R Form M, PPVT-3 Form A, and PPVT-3 Form B.

Main Results:

• All three PPVT forms demonstrated high reliability, covering approximately ±3 standard deviations of vocabulary ability.
• Identified differences in item difficulty and discrimination between PPVT-3 Forms A and B.
• Established that IRT modeling can yield comparable vocabulary ability estimates across different test forms.

Conclusions:

• IRT modeling provides a robust method for equating vocabulary ability scores from various PPVT forms.
• A freely available SAS program is provided to facilitate IRT-based ability estimation using the study's item parameters.
• This resource allows researchers to leverage existing PPVT data for standardized ability comparisons.