molecalculator

Moles & Molar Mass

Sensitivity Power Analysis Calculator: How It Works

Calculator No. 01-10047 · Updated · Formula and assumptions shown below

The calculator on this page is switched off while it is checked. The formula is below, and the worked example shows how to do the sum by hand.

For study and lab planning. The result follows from the numbers you enter and the formula shown. Check it against your method before you weigh, mix or dilute anything, follow the safety data sheet and your supervisor, and do not use it for clinical, dosing or medication work.

A sensitivity power analysis calculator determines the minimum detectable effect size (or required sample size) given your chosen statistical power, significance level, and sample size. You enter your sample size, desired power (typically 0.80), and alpha (typically 0.05), and the calculator returns the smallest effect size your study can reliably detect. This is essential for planning studies where the sample size is fixed (e.g., due to budget or ethics) and you need to know what magnitude of effect is worth looking for.

Formula

For a two-sample t-test (equal group sizes), the minimum detectable standardized effect size (Cohen’s d) is calculated as:

d = (z1-β + z1-α/2) / √(n/2)

Where:

  • d = Cohen’s d (effect size = difference in means divided by pooled standard deviation)
  • z1-β = z-score corresponding to the desired power (e.g., for power = 0.80, z = 0.8416)
  • z1-α/2 = z-score for the significance level (e.g., for α = 0.05 two-tailed, z = 1.96)
  • n = total sample size (both groups combined)

This formula assumes equal group sizes, a two-sample t-test, and normally distributed data. The calculator above uses the same logic but also allows one-tailed tests. For one-tailed tests, the z for alpha is z1-α (e.g., 1.645 for α = 0.05).

Worked Example

Suppose you have a fixed sample size of 100 (50 per group), want 80% power, and use a two-tailed test with α = 0.05.

  1. Find the z-scores: z0.80 = 0.8416, z0.975 = 1.9600
  2. Sum them: 0.8416 + 1.9600 = 2.8016
  3. Divide by √(n/2) = √(50) = 7.0711
  4. d = 2.8016 / 7.0711 = 0.396

With 100 total participants, your study can detect an effect size of d ≈ 0.40 (a moderate effect) with 80% power. If the true effect is smaller, your study is underpowered to detect it.

Assumptions

  • Two-sample t-test (independent groups) with equal sample sizes per group.
  • Normally distributed data and equal variances between groups.
  • The effect size is defined as Cohen’s d (mean difference / pooled standard deviation).
  • No covariates or repeated measures; for more complex designs, use specialized software.

Why Use a Sensitivity Power Analysis?

Most power analyses ask: “Given an expected effect size, what sample size do I need?” But sometimes sample size is fixed—because of budget, participant availability, or data already collected. In that case, the question reverses: “Given my sample size, what effect size can I reliably detect?” This is a sensitivity analysis because it tells you how sensitive your study is to detecting small effects. It’s particularly useful when planning a pilot study, evaluating a completed null result, or reviewing a grant where you need to justify feasibility. A sensitivity analysis can also help you decide whether a study is worth doing at all: if the minimum detectable effect is larger than what is theoretically plausible, the study may be a waste of resources.

How to Choose Inputs

  • Sample size (n): Total number of participants across both groups. Use the actual number you have or plan to collect.
  • Power (1-β): Conventionally 0.80, meaning 80% chance of detecting a true effect. Higher power reduces false negatives but may require a larger sample.
  • Alpha (α): Usually 0.05, the probability of a false positive. Lower alpha (e.g., 0.01) makes the test more conservative and increases the detectable effect.
  • Tails: Two-tailed tests are standard because they detect effects in either direction. One-tailed tests are only appropriate when you have a strong directional hypothesis and never for post-hoc checks.

If you have unequal group sizes, the formula changes slightly. For groups of size n1 and n2, the denominator becomes √(n1*n2/(n1+n2)) instead of √(n/2). The calculator above assumes equal groups, which is the most common design.

Common Mistakes and Misconceptions

  • Confusing sensitivity analysis with post-hoc power: Post-hoc power uses the observed effect size to compute power after the study, which is often misleading. Sensitivity analysis uses a specified power to find the minimum detectable effect, which is more informative for interpreting null results.
  • Using one-tailed tests without justification: One-tailed tests halve the alpha z-score, making the detectable effect smaller, but they are only valid if you can confidently rule out effects in the opposite direction. Most journals expect two-tailed tests.
  • Ignoring effect size conventions: Cohen suggested d = 0.2 small, 0.5 medium, 0.8 large. But these are arbitrary; the meaningful effect depends on your field. A sensitivity analysis tells you what you can detect, not what is practically important.
  • Rounding too early: Use at least 4 decimal places in intermediate calculations. Rounding z-scores prematurely can change the result by 0.01 or more.

Limitations of This Calculator

This calculator is designed for a simple two-sample t-test. Real studies often involve more complex designs: ANOVA, regression, repeated measures, or clustered data. For those, the minimum detectable effect depends on additional parameters (e.g., number of groups, covariance structure, intraclass correlation). Use dedicated software like G*Power or R for such cases. Also, the formula assumes equal variances and normality. If your data are heavily skewed or variances differ, the actual power may be lower. Finally, this calculator gives the effect size in Cohen’s d; if you need the minimum detectable difference in raw units (e.g., mmHg), multiply d by the expected standard deviation of your outcome.

Frequently Asked Questions

What is the difference between sensitivity power analysis and a priori power analysis?
A priori analysis determines the required sample size for a given effect size, power, and alpha. Sensitivity analysis determines the minimum detectable effect size for a given sample size, power, and alpha. They answer opposite questions.

Can I use this calculator for a one-sample t-test?
No, this calculator is for two independent groups. For a one-sample t-test, the formula is d = (z1-β + z1-α/2) / √n, where n is the total sample size. You can adapt the logic, but the denominator changes.

What is a reasonable power level?
0.80 is the conventional minimum, meaning an 80% chance of detecting a true effect. Some fields require 0.90 or higher for confirmatory studies. Higher power is always better, but it requires a larger sample or a larger detectable effect.

If my study finds no effect, should I do a sensitivity analysis?
Yes, it is highly recommended. A sensitivity analysis tells you the smallest effect your study could have detected. If the minimum detectable effect is large (e.g., d = 0.8), then your null result is not very informative—you only ruled out large effects. If the minimum detectable effect is small (e.g., d = 0.2), then your null result is more convincing.

Does the calculator work for unequal group sizes?
No, it assumes equal group sizes. For unequal groups, use the formula d = (z1-β + z1-α/2) / √(n1*n2/(n1+n2)), where n1 and n2 are the group sizes. You can compute that manually or use a more flexible tool.

What if my outcome is not continuous?
For binary outcomes (e.g., success/failure), the effect size is often expressed as odds ratio or risk difference, and the formula is different. This calculator is only for continuous outcomes analyzed with a t-test. For binary outcomes, use a chi-square or logistic regression power calculator.

Related

Formula used
in n | Total sample size | - | 100 | 2
in power | Desired power | - | 0.80 | 0.01 | 1
in alpha | Significance level (alpha) | - | 0.05 | 0.001 | 1
in tails | Test direction | 2=Two-tailed, 1=One-tailed | 2
out d | Minimum detectable Cohen's d | - | ( (sqrt(n) * ( (power_to_z(power) + power_to_z(1 - alpha/tails)) )) ) / sqrt(n/2) | 3
out n_per_group | Sample per group | - | n / 2 | 0

Each line is one input (in, sel, list) or one result (out: label, unit, formula, decimals). How the calculators work

Method: the standard formula for this calculation with the constants and conversion factors written on the page, rounded as each result states. See method and constants, the disclaimer or report an error.

More in this section

The whole Moles & Molar Mass section →