Planned missingness in intensive longitudinal studies: Extensions and comparisons of multiform designs

Area of Science:

• Psychometrics
• Longitudinal Data Analysis
• Statistical Modeling

Background:

• Intensive longitudinal studies (ILSs) are increasingly feasible due to technological advances in data collection.
• High measurement frequency in ILSs can increase participant burden, leading to nonresponse and reduced data quality.
• Planned missingness designs offer a strategy to optimize data collection in ILSs.

Purpose of the Study:

• To extend existing multiform designs and propose a new random sampling design for ILSs.
• To evaluate the performance of these multiform designs within the dynamic structural equation modeling (DSEM) framework.
• To demonstrate the utility of multiform designs in reducing participant burden and costs in ILSs.

Main Methods:

• Examined three multiform designs: anchor test, matrix sampling, and random sampling.
• Varied item subset combinations across measurement occasions for participants.
• Conducted simulation studies within the DSEM framework, varying key parameters like sample size and missingness proportion.

Main Results:

• Multiform designs yielded unbiased parameter point estimates.
• Acceptable credible interval coverage rates were maintained.
• Sufficient statistical power was preserved.

Conclusions:

• Extended and proposed multiform designs effectively reduce costs and participant burden in ILSs.
• These designs provide adequate data for capturing dynamic processes.
• Multiform designs offer a flexible and efficient approach to planned missingness in longitudinal research.