Analyzing differences between parent- and self-report measures with a latent space approach.

Area of Science:

• Psychometrics
• Developmental Psychology
• Child Psychiatry

Background:

• Cross-informant discrepancies between child and parent reports are common when assessing behavioral and emotional problems.
• Traditional methods struggle to capture the complex dependencies between child and parent perspectives.
• The Child Behavior Checklist (CBCL) and Youth Self Report (YSR) are widely used but show informant variations.

Purpose of the Study:

• To propose and demonstrate a new method, the Latent Space Item Response Model (LSIRM), for analyzing cross-informant discrepancies.
• To investigate the dependency structures between items and respondents from both child and parent perspectives.
• To illustrate the utility of LSIRM using the CBCL and YSR datasets.

Main Methods:

• Development and application of a Latent Space Item Response Model (LSIRM).
• Estimation of an interaction map to visualize respondent and item positions in latent spaces.
• Analysis of differential item and respondent functioning across child (YSR) and parent (CBCL) reports.

Main Results:

• The LSIRM successfully captures differential positions of items and respondents in latent spaces.
• The interaction map reveals complex dependencies between respondents and items, offering insights beyond conventional analyses.
• LSIRM analysis highlights the benefit of uncovering distinct response structures from child and parent viewpoints.

Conclusions:

• The LSIRM provides a powerful framework for understanding cross-informant discrepancies in child behavioral assessments.
• Integrating both child and parent perspectives through LSIRM leads to a more comprehensive evaluation of behavioral problems.
• This approach enhances the nuanced understanding of children's emotional and behavioral challenges.