Minimum Detectable Effect Calculator
Know what your study can actually detect before you launch it — or find out exactly how many respondents you need to measure the effect that matters to you.
Learn
What is Minimum Detectable Effect?
The Minimum Detectable Effect (MDE) is the smallest true difference between two groups that your study is statistically powered to detect — given your sample size, confidence level, and power setting. Think of it as the resolution of your experiment. Just as a camera with low resolution can't capture fine detail, an underpowered study can't detect small effects. The MDE tells you exactly how fine that resolution is before you launch. If your MDE is 5 percentage points, and the true effect of your campaign is 3 percentage points, your study will likely miss it — even if the effect is real. Understanding your MDE upfront prevents you from running tests that can't answer the questions you're asking.
The four factors that drive MDE
MDE is determined by four inputs that are always in tension with one another: 1. Sample size — the more respondents you have, the smaller the effects you can detect. Doubling your sample reduces your MDE by roughly 30%. 2. Significance level (α) — how much false-positive risk you'll accept. A 95% confidence threshold (the standard) means a 5% chance of detecting an effect that isn't real. Tightening this to 99% requires a larger sample. 3. Statistical power (1-β) — the probability of detecting an effect that is real. 80% power (the standard) means a 20% chance of missing a real effect. Increasing power to 90% requires a larger sample. 4. Baseline metric — for binary metrics (rates), MDE is largest when the baseline is near 50%. For continuous metrics (scores), higher variability (standard deviation) means a larger MDE. These four factors form a system: fix three, and the fourth is determined. This calculator lets you explore both directions — given your sample, what can you detect? Or, given your target effect, how many respondents do you need?
How to interpret and use your MDE
Before launching a study, always ask: is my MDE smaller than the effect I expect to see? If your MDE is 8% but you only expect your campaign to shift intent by 2–3%, your study won't have the resolution to pick that up — even if the effect is real. You have two options: increase your sample size, or adjust your expectations. After a study, MDE helps you interpret null results. A "no significant effect" finding means something different depending on your MDE: • MDE of 2%: strong evidence that the effect (if any) is very small • MDE of 15%: inconclusive — a meaningful effect may exist but your study wasn't powered to see it In attitudinal and survey research, a useful rule of thumb: aim for an MDE of no more than half the effect size you consider meaningful. If a 5-point shift in brand sentiment matters to you, design for an MDE of 2–3 points.
Calculator
Select your mode, fill in your inputs, and get your result instantly — no data upload needed.
Designed for survey & attitudinal research. This calculator is optimised for studies with sample sizes in the low thousands to tens of thousands — typical of consumer surveys, brand trackers, and attitudinal panels. It is not intended for large-scale behavioural or digital datasets (millions of events), where different tooling applies.
What do you want to calculate?
Metric type
Use for metrics like agreement rate, purchase intent, or any yes/no response.
Inputs
Enter as a decimal — e.g. 0.35 for 35%