Search for efficient complete and planned missing data designs for analysis of change

Area of Science:

• Biostatistics
• Longitudinal Data Analysis
• Research Design

Background:

• Longitudinal data collection is crucial for studying change over time.
• Efficient study designs maximize statistical power and minimize resource expenditure.
• Complete data (CD) and planned missing (PM) designs are common approaches for longitudinal data.

Purpose of the Study:

• To propose SEEDMC (Search for Efficient Designs using Monte Carlo Simulation), a systematic procedure for identifying efficient longitudinal data collection designs.
• To enable the search for efficient complete data (CD) and planned missing (PM) designs tailored for growth-curve modeling within budget limitations.
• To develop designs robust to missing data due to attrition, specifically under the Missing Completely At Random (MCAR) assumption.

Main Methods:

• SEEDMC utilizes Monte Carlo simulation to systematically explore various longitudinal designs.
• The procedure allows for the identification of efficient designs for single or multiple effects simultaneously.
• Application focuses on linear and quadratic growth models to find efficient designs for key change parameters.

Main Results:

• SEEDMC successfully identified efficient complete data (CD) and planned missing (PM) designs for growth-curve modeling.
• The procedure demonstrated flexibility in optimizing designs for specific effects and robustness against MCAR attrition.
• Efficient designs were determined for key parameters in both linear and quadratic growth models.

Conclusions:

• SEEDMC provides a flexible and systematic approach for optimizing longitudinal study designs under budget constraints.
• The method aids researchers in selecting efficient designs that balance statistical power, cost, and missing data robustness.
• Future extensions of SEEDMC could incorporate more complex growth models and different missing data mechanisms.