Using Ant Colony Optimization for the Optimal Design of Experimental Studies With Complex Power Formulas

• In Event: Estimation Methods for Multilevel Models

Abstract

Optimal design frameworks seek to identify optimal sample allocations that can produce the maximum statistical power under a fixed budget. Conventionally, the optimization is done through the first-order derivative approach by setting the first-order derivatives equal to zero and solving for the optimal design parameters. However, the derivative method is not capable of handling the optimization task when statistical power formulas are complex, such as those for designs detecting mediation effects under the joint significance test. This study proposes using the ant colony optimization algorithm to optimize the design of experiments with complex power formulas. Preliminary results show that the ant colony optimization algorithm can efficiently identify optimal sample allocations. The method is implemented in the R package XXX.

Authors

• Zuchao Shen, University of Georgia

• Walter L. Leite, University of Florida

• Huibin Zhang, University of Tennessee

• Jia Quan, University of Florida

• Huan Kuang, University of Florida