experimental-design

Experimental design

Write the primary hypothesis and estimand before choosing measurements. Define the experimental unit, biological and technical replicates, primary outcome, decision threshold, and what result would falsify or materially weaken the hypothesis.

Specify positive, negative, vehicle, batch, and process controls as applicable. Identify confounders and decide how randomization, blocking, matching, blinding, or covariate adjustment will address them.

Base sample-size reasoning on the smallest effect worth detecting, expected variance, desired power, allocation, and planned analysis. Label heuristic sample sizes as heuristics.

Predefine exclusions, stopping rules, missing-data handling, multiplicity, and the analysis plan. Include a pilot when feasibility or variance is too uncertain for a defensible full design.

End with a reproducibility checklist: materials and identifiers, protocol version, instrument settings, software/environment, raw-data location, metadata, and the artifacts that constitute a successful run.