Strategies to manipulate reliability: impact on statistical associations

Area of Science:

• Child and Adolescent Psychiatry
• Psychometric Research
• Developmental Psychology

Background:

• Childhood psychiatric disorders are influenced by various risk factors.
• Accurate measurement is crucial for understanding these associations.
• Previous research may be limited by measurement reliability issues.

Purpose of the Study:

• To investigate how enhancing measurement reliability impacts the observed relationships between risk factors and childhood psychiatric disorders.
• To determine if improved reliability strengthens or clarifies associations with specific disorders and risk factors.

Main Methods:

• Utilized data from a general population sample of 211 parents and their children (aged 6-16).
• Improved measurement reliability through increased scale items, averaging assessments over time, and latent variable modeling.
• Regressed selected risk factors onto parental assessments of oppositional defiant disorder (ODD) and overanxious disorder (OAD).

Main Results:

• Systematic increases in the magnitude of associations between family dysfunction and ODD/OAD were observed with improved reliability.
• Enhanced reliability strengthened the specificity of links between ODD, OAD, family dysfunction, and maternal depressed mood.
• Latent variable methods yielded the largest associations but also larger standard errors.

Conclusions:

• Improving measurement reliability leads to stronger associations between risk factors and childhood psychiatric disorders.
• Enhanced reliability can increase the potential for identifying differential associations between variables.
• These findings underscore the importance of robust psychometric properties in psychiatric research.