Harmonization of cognitive screening tools for dementia across diverse samples: A simulation study

Area of Science:

• Gerontology and Cognitive Science
• Psychometrics and Statistical Modeling

Background:

• Cognitive aging research is global, but sociocultural differences hinder direct comparisons of test scores across nations.
• Harmonization of cognitive data using item response theory (IRT) can facilitate cross-national comparisons.

Purpose of the Study:

• To explore necessary conditions for accurate harmonization of cognitive data using simulation.
• To evaluate how the quality and quantity of linking items affect cross-sample harmonization accuracy.

Main Methods:

• Item response theory (IRT) analysis was applied to US Health and Retirement Study (HRS) and Mexican Health and Aging Study (MHAS) data.
• Simulations were conducted under 10 scenarios varying linking item quality and quantity to assess harmonization accuracy.

Main Results:

• The current linking items in HRS and MHAS data were unsuitable for harmonization, causing significant bias.
• Increased numbers and improved quality of linking items led to less biased and more accurate harmonization outcomes.

Conclusions:

• Linking items must have low measurement error across the ability range for successful co-calibration.
• Optimal harmonization accuracy is achieved with linking items that vary in difficulty and strongly correlate with the measured ability.