# A power analysis framework to aid the design of robust semi-field vector control experiments

## TL;DR

This paper introduces a power analysis framework to help design effective semi-field experiments for testing vector control tools.

## Contribution

A novel power analysis methodology and tutorial for optimizing semi-field vector control experiments.

## Key findings

- The number of chambers and variance between chambers most strongly influence statistical power.
- Short-term and long-term experiments require similar numbers of chambers per treatment.
- Adding an additional intervention increases the required number of chambers.

## Abstract

Semi-field experiments are an efficient way of assessing the impacts of potential new vector control tools (VCTs) before field trials. However, their design is critically important to ensure their results are unbiased and informative. An essential element of the design of semi-field experiments is power analysis, which empowers researchers to ensure that only designs with adequate statistical power are adopted. In this study, a methodology was developed, and its use was demonstrated in a tutorial, to determine the required number of semi-field chambers, sampling frequency and the number of mosquitoes required to achieve sufficient power for evaluating the impact of a single VCT or two in combination.

By analysing data simulated from a generalized linear mixed-effects model, power was estimated for various experimental designs, including short- (24 h) vs. long-term (3 months) experiments and single vs. combined application of interventions (e.g., insecticide-treated nets combined with pyriproxyfen autodissemination).

Although power increased with increasing number of chambers, sampling frequency and the number of mosquitoes, the number of chambers and variance between chambers were the dominant factors determining power relative to all other design choices. High variance between chambers decreased power, highlighting the importance of making conditions similar among chambers, by reducing variation if possible and by rotating variables if not. As compared to a single intervention, an additional intervention required an increase in the number of chambers, while short and long experiments were similar in terms of key aspects such as the number of chambers per treatment.

Determining the most efficient experimental design for a semi-field experiment will depend on a balance of design choices and resource constraints. The power analysis framework and tutorial provided here can aid in the robust design of these widely used experiments and ultimately facilitate the development of new vector control tools (VCTs).

The online version contains supplementary material available at 10.1186/s12936-025-05454-y.