More Power for Less Money: Statistical, Power, and Cost Analyses That Account for Intracluster Correlation in...

Area of Science:

• * Biostatistics
• * Laboratory Animal Science
• * Experimental Design

Background:

• * Cohoused animals in research can exhibit correlated outcomes due to intracluster correlation.
• * Ignoring intracluster correlation in same-group cage mate experiments leads to pseudoreplication and increases false-positive results.
• * This issue mirrors challenges found in cluster randomized trials in human studies.

Purpose of the Study:

• * To provide a tutorial on statistically accounting for intracluster correlation in same-group cage mate experiments.
• * To demonstrate effect size and sample size calculations for proper experimental power.
• * To highlight the efficiency and cost-effectiveness of experimental designs that mitigate intracluster correlation.

Main Methods:

• * Utilizing linear mixed models by including cage identifiers as independent variables.
• * Performing power analyses, including effect size and sample size calculations.
• * Conducting cost analyses comparing different experimental designs.
• * Demonstrating statistical analyses using JASP, R, and SAS software.

Main Results:

• * Accounting for intracluster correlation reduces false-positive statistical errors.
• * Experimental designs with more cages and fewer animals per cage are statistically more efficient (higher power).
• * Optimized designs can lead to reduced animal numbers, aligning with the 3Rs of animal research.
• * Cost analyses indicate that more cages with fewer animals per cage are often less expensive overall.

Conclusions:

• * Properly accounting for intracluster correlation is crucial for accurate statistical analysis in same-group cage mate experiments.
• * Efficient experimental design improves statistical power, reduces costs, and minimizes animal usage.
• * The methods presented offer a practical approach to enhance the rigor and ethics of animal research.