Glossary · Design

Design

Plan tests that can answer a useful question without creating avoidable validity risks.

Glossary terms

Browse this topic

Adaptive Experimentation

Changing allocation or design as evidence arrives while preserving valid inference.

Adaptive Sequential Design

Planned interim analyses and adaptations that maintain statistical guarantees.

Always-Valid Inference

Methods designed for continuous monitoring without invalid optional stopping.

Before-and-After Analysis

What change-over-time comparisons can show and why they rarely establish causality alone.

Blocking

Grouping similar units before randomization to improve precision and balance.

Cluster Randomized Experiment

Design choices, assumptions, and practical guidance for experiments.

Cluster-Level Randomization

Design choices, assumptions, and practical guidance for experiments.

Conversion Lift

Design choices, assumptions, and practical guidance for experiments.

CUPED

Design choices, assumptions, and practical guidance for experiments.

Difference-in-Differences

Design choices, assumptions, and practical guidance for experiments.

Experiment Duration

Business cycles, seasonality, novelty, and stopping rules.

Factorial Design

Design choices, assumptions, and practical guidance for experiments.

Fixed-Horizon Test

Design choices, assumptions, and practical guidance for experiments.

Fractional Factorial Design

Design choices, assumptions, and practical guidance for experiments.

Full Factorial Design

Design choices, assumptions, and practical guidance for experiments.

Geo Experiment

Design choices, assumptions, and practical guidance for experiments.

Geo Lift Test

Design choices, assumptions, and practical guidance for experiments.

Group Sequential Design

Design choices, assumptions, and practical guidance for experiments.

Heterogeneous Treatment Effect

Design choices, assumptions, and practical guidance for experiments.

Hypothesis

How to write falsifiable, mechanism-based product hypotheses.

Multiple Testing

Bonferroni, Holm, FDR, variants, metrics, and segments.

Sample Size

Baseline, MDE, power, alpha, allocation, and duration.