How to Determine Sample Size for a Study

Created:February 26, 2026

Last Updated:March 5, 2026

One of the most common questions researchers ask is: "How many participants do I need?" Too few participants and your study won't be able to detect real effects. Too many and you waste time, money, and resources. Getting the sample size right is one of the most important decisions in study design.

This guide walks you through a practical, step-by-step process for determining the right sample size for your study, whether it's a clinical trial, a survey, an A/B test, or an academic experiment.

Ready to calculate? Use our Sample Size & Power Analysis Calculator to compute the exact sample size for your study parameters.

Why Sample Size Matters

Sample size directly impacts the reliability and validity of your study's conclusions. Here's what can go wrong:

Too Small

High risk of missing real effects (Type II error)
Unreliable estimates with wide confidence intervals
Wasted resources on inconclusive results
Ethical concerns about participant burden without benefit

Too Large

Unnecessary expense of time and money
Detects trivially small, meaningless differences
Exposes more participants to potential risks than needed
Delays research completion and publication

The goal is to find the minimum sample size that gives your study enough statistical power to detect a meaningful effect. This is where power analysis comes in.

5 Steps to Determine the Right Sample Size

Define Your Research Question and Hypotheses

Before you can calculate a sample size, you need to clearly define what you're testing. Your research question determines the type of analysis you'll perform, which in turn determines the sample size formula.

Ask yourself:

Are you comparing two groups or more than two groups?
Are you measuring a continuous outcome or a proportion?
Are the groups independent or paired/matched?
What is the primary outcome variable?

Example

"Does a new tutoring program improve math test scores compared to the standard curriculum?" This is a comparison of two independent group means, so you'd use a two-sample t-test for the sample size calculation.

Choose Your Statistical Test

Different statistical tests have different sample size formulas. The table below helps you match your research design to the right test:

Research Design	Outcome Type	Statistical Test
2 independent groups	Continuous (means)	Independent t-test
2 paired/matched groups	Continuous (means)	Paired t-test
3+ independent groups	Continuous (means)	One-way ANOVA
2 independent groups	Proportions	Two-proportion z-test
Categorical association	Frequencies	Chi-square test
Predictive model	Continuous	Multiple regression

Set Your Key Parameters

Every sample size calculation requires three core parameters. Together with the statistical test, these four elements form the "power analysis quartet" — knowing any three lets you solve for the fourth.

Significance Level ( $\alpha$ )

The probability of rejecting the null hypothesis when it is actually true (a Type I error).

Common choice: 0.05 (5%). Use 0.01 for more conservative testing or 0.10 for exploratory research.

Statistical Power ( $1 - \beta$ )

The probability of correctly detecting an effect when one truly exists. Higher power means a lower chance of a Type II error (missing a real effect).

Common choice: 0.80 (80%). Use 0.90 or 0.95 for critical studies like clinical trials.

Effect Size

The magnitude of the difference or relationship you expect to find. This is often the hardest parameter to specify. You can estimate it from:

Prior research: Published studies in your field
Pilot data: A small preliminary study
Practical significance: The smallest difference that would be meaningful
Conventions: Cohen's guidelines (small, medium, large)

Test	Small	Medium	Large
t-test (Cohen's d)	0.2	0.5	0.8
ANOVA (Cohen's f)	0.10	0.25	0.40
Chi-square (Cohen's w)	0.10	0.30	0.50
Regression (Cohen's $f^2$ )	0.02	0.15	0.35

Learn more about effect sizes in our Effect Size tutorial.

Calculate the Sample Size

With your test type, significance level, power, and effect size determined, you can now calculate the required sample size. There are three ways to do this:

Option A: Use an Online Calculator (Recommended)

The easiest approach. Our Sample Size Calculator supports t-tests, proportion tests, ANOVA, chi-square, and regression, and generates power curves automatically.

Option B: Use Statistical Software

Python

from statsmodels.stats.power import TTestIndPower

analysis = TTestIndPower()
sample_size = analysis.solve_power(
    effect_size=0.5,   # Cohen's d
    alpha=0.05,        # Significance level
    power=0.8,         # Desired power
    alternative='two-sided'
)
print(f"Required sample size per group: {int(sample_size) + 1}")
# Output: Required sample size per group: 64

R

library(pwr)

result <- pwr.t.test(
    d = 0.5,              # Cohen's d
    sig.level = 0.05,     # Significance level
    power = 0.8,          # Desired power
    type = "two.sample",
    alternative = "two.sided"
)
cat("Required sample size per group:", ceiling(result$n))
# Output: Required sample size per group: 64

Option C: Use the Formula Directly

For a two-sample t-test with equal groups:

n = \frac{2(z_{\alpha/2} + z_{\beta})^2}{d^2}

See our Sample Size Formula tutorial for detailed formulas for all test types with worked examples.

Adjust for Practical Factors

The calculated sample size is the minimum needed under ideal conditions. In practice, you should inflate it to account for real-world issues:

Dropout / Attrition

Participants may withdraw, miss follow-ups, or provide unusable data. Adjust with:

n_{\text{adjusted}} = \frac{n_{\text{required}}}{1 - \text{dropout rate}}

For example, if you need 100 participants and expect 20% dropout: 100 / (1 - 0.20) = 125 participants to recruit.

Budget and Timeline

Calculate total cost: n × cost per participant
Estimate recruitment timeline: n ÷ recruitment rate per week
If budget is insufficient, consider a smaller (but still meaningful) effect size target, or a more efficient design

Design Modifications

If you need to reduce the sample size, consider:

Paired/repeated measures design: Typically requires fewer participants than independent groups
Covariates (ANCOVA): Reducing variance increases power
One-sided test: Requires fewer participants, but only if you can justify the direction a priori
Stratified sampling: More efficient than simple random sampling

Common Scenarios

Clinical Trial

Scenario: Testing whether a new blood pressure medication reduces systolic BP by at least 5 mmHg more than the standard treatment (SD = 15 mmHg).

Effect size: d = 5/15 = 0.33 (small-to-medium)
Power: 0.90 (higher for clinical studies)
Alpha: 0.05, two-sided
Result: ~192 per group (384 total)
With 15% dropout: 384 / 0.85 = 452 total to recruit

A/B Test (Website Conversion)

Scenario: Testing whether a new checkout flow increases conversion from 3% to 4% (a 1 percentage point lift).

Test type: Two-proportion z-test
Baseline rate: 0.03, expected rate: 0.04
Power: 0.80, Alpha: 0.05, two-sided
Result: ~6,678 per group (13,356 total visitors)

Survey / Questionnaire Study

Scenario: Comparing job satisfaction scores between remote and in-office workers, expecting a medium effect (d = 0.5).

Test type: Independent t-test
Effect size: d = 0.5 (medium)
Power: 0.80, Alpha: 0.05, two-sided
Result: 64 per group (128 total)
With 25% non-response: 128 / 0.75 = 171 surveys to send

Quick Reference: Sample Sizes for Common Designs

The following table shows approximate sample sizes per group for common study designs at 80% power and $\alpha = 0.05$ (two-sided). These are rough estimates; always run a proper power analysis for your specific study.

Test	Small Effect	Medium Effect	Large Effect
Independent t-test	394	64	26
Paired t-test	199	34	15
ANOVA (3 groups)	322	53	22
Chi-square (df=1)	785	88	32

Use our Sample Size & Power Analysis Calculator to get exact numbers for your specific parameters, complete with power curves.

How to Determine Sample Size for a Study

Created:February 26, 2026

Last Updated:March 5, 2026

This guide walks you through a practical, step-by-step process for determining the right sample size for your study, whether it's a clinical trial, a survey, an A/B test, or an academic experiment.

Ready to calculate? Use our Sample Size & Power Analysis Calculator to compute the exact sample size for your study parameters.

Why Sample Size Matters

Sample size directly impacts the reliability and validity of your study's conclusions. Here's what can go wrong:

Too Small

High risk of missing real effects (Type II error)
Unreliable estimates with wide confidence intervals
Wasted resources on inconclusive results
Ethical concerns about participant burden without benefit

Too Large

Unnecessary expense of time and money
Detects trivially small, meaningless differences
Exposes more participants to potential risks than needed
Delays research completion and publication

The goal is to find the minimum sample size that gives your study enough statistical power to detect a meaningful effect. This is where power analysis comes in.

5 Steps to Determine the Right Sample Size

Define Your Research Question and Hypotheses

Ask yourself:

Are you comparing two groups or more than two groups?
Are you measuring a continuous outcome or a proportion?
Are the groups independent or paired/matched?
What is the primary outcome variable?

Example

Choose Your Statistical Test

Different statistical tests have different sample size formulas. The table below helps you match your research design to the right test:

Research Design	Outcome Type	Statistical Test
2 independent groups	Continuous (means)	Independent t-test
2 paired/matched groups	Continuous (means)	Paired t-test
3+ independent groups	Continuous (means)	One-way ANOVA
2 independent groups	Proportions	Two-proportion z-test
Categorical association	Frequencies	Chi-square test
Predictive model	Continuous	Multiple regression

Set Your Key Parameters

Significance Level ( $\alpha$ )

The probability of rejecting the null hypothesis when it is actually true (a Type I error).

Common choice: 0.05 (5%). Use 0.01 for more conservative testing or 0.10 for exploratory research.

Statistical Power ( $1 - \beta$ )

The probability of correctly detecting an effect when one truly exists. Higher power means a lower chance of a Type II error (missing a real effect).

Common choice: 0.80 (80%). Use 0.90 or 0.95 for critical studies like clinical trials.

Effect Size

The magnitude of the difference or relationship you expect to find. This is often the hardest parameter to specify. You can estimate it from:

Prior research: Published studies in your field
Pilot data: A small preliminary study
Practical significance: The smallest difference that would be meaningful
Conventions: Cohen's guidelines (small, medium, large)

Test	Small	Medium	Large
t-test (Cohen's d)	0.2	0.5	0.8
ANOVA (Cohen's f)	0.10	0.25	0.40
Chi-square (Cohen's w)	0.10	0.30	0.50
Regression (Cohen's $f^2$ )	0.02	0.15	0.35

Learn more about effect sizes in our Effect Size tutorial.

Calculate the Sample Size

With your test type, significance level, power, and effect size determined, you can now calculate the required sample size. There are three ways to do this:

Option A: Use an Online Calculator (Recommended)

The easiest approach. Our Sample Size Calculator supports t-tests, proportion tests, ANOVA, chi-square, and regression, and generates power curves automatically.

Option B: Use Statistical Software

Python

from statsmodels.stats.power import TTestIndPower

analysis = TTestIndPower()
sample_size = analysis.solve_power(
    effect_size=0.5,   # Cohen's d
    alpha=0.05,        # Significance level
    power=0.8,         # Desired power
    alternative='two-sided'
)
print(f"Required sample size per group: {int(sample_size) + 1}")
# Output: Required sample size per group: 64

R

library(pwr)

result <- pwr.t.test(
    d = 0.5,              # Cohen's d
    sig.level = 0.05,     # Significance level
    power = 0.8,          # Desired power
    type = "two.sample",
    alternative = "two.sided"
)
cat("Required sample size per group:", ceiling(result$n))
# Output: Required sample size per group: 64

Option C: Use the Formula Directly

For a two-sample t-test with equal groups:

n = \frac{2(z_{\alpha/2} + z_{\beta})^2}{d^2}

See our Sample Size Formula tutorial for detailed formulas for all test types with worked examples.

Adjust for Practical Factors

The calculated sample size is the minimum needed under ideal conditions. In practice, you should inflate it to account for real-world issues:

Dropout / Attrition

Participants may withdraw, miss follow-ups, or provide unusable data. Adjust with:

n_{\text{adjusted}} = \frac{n_{\text{required}}}{1 - \text{dropout rate}}

For example, if you need 100 participants and expect 20% dropout: 100 / (1 - 0.20) = 125 participants to recruit.

Budget and Timeline

Calculate total cost: n × cost per participant
Estimate recruitment timeline: n ÷ recruitment rate per week
If budget is insufficient, consider a smaller (but still meaningful) effect size target, or a more efficient design

Design Modifications

If you need to reduce the sample size, consider:

Paired/repeated measures design: Typically requires fewer participants than independent groups
Covariates (ANCOVA): Reducing variance increases power
One-sided test: Requires fewer participants, but only if you can justify the direction a priori
Stratified sampling: More efficient than simple random sampling

Common Scenarios

Clinical Trial

Scenario: Testing whether a new blood pressure medication reduces systolic BP by at least 5 mmHg more than the standard treatment (SD = 15 mmHg).

Effect size: d = 5/15 = 0.33 (small-to-medium)
Power: 0.90 (higher for clinical studies)
Alpha: 0.05, two-sided
Result: ~192 per group (384 total)
With 15% dropout: 384 / 0.85 = 452 total to recruit

A/B Test (Website Conversion)

Scenario: Testing whether a new checkout flow increases conversion from 3% to 4% (a 1 percentage point lift).

Test type: Two-proportion z-test
Baseline rate: 0.03, expected rate: 0.04
Power: 0.80, Alpha: 0.05, two-sided
Result: ~6,678 per group (13,356 total visitors)

Survey / Questionnaire Study

Scenario: Comparing job satisfaction scores between remote and in-office workers, expecting a medium effect (d = 0.5).

Test type: Independent t-test
Effect size: d = 0.5 (medium)
Power: 0.80, Alpha: 0.05, two-sided
Result: 64 per group (128 total)
With 25% non-response: 128 / 0.75 = 171 surveys to send

Quick Reference: Sample Sizes for Common Designs

Test	Small Effect	Medium Effect	Large Effect
Independent t-test	394	64	26
Paired t-test	199	34	15
ANOVA (3 groups)	322	53	22
Chi-square (df=1)	785	88	32

Use our Sample Size & Power Analysis Calculator to get exact numbers for your specific parameters, complete with power curves.

How to Determine Sample Size for a Study

Why Sample Size Matters

Too Small

Too Large

5 Steps to Determine the Right Sample Size

Define Your Research Question and Hypotheses

Ask yourself:

Example

Choose Your Statistical Test

Set Your Key Parameters

Significance Level (α\alphaα)

Statistical Power (1−β1 - \beta1−β)

Effect Size

Calculate the Sample Size

Option A: Use an Online Calculator (Recommended)

Option B: Use Statistical Software

Python

R

Option C: Use the Formula Directly

Adjust for Practical Factors

Dropout / Attrition

Budget and Timeline

Design Modifications

Common Scenarios

Clinical Trial

A/B Test (Website Conversion)

Survey / Questionnaire Study

Quick Reference: Sample Sizes for Common Designs

Test Your Understanding

View Step-by-Step Solution

Think About It

I've thought about it - Show me the solution

How to Determine Sample Size for a Study

Why Sample Size Matters

Too Small

Too Large

5 Steps to Determine the Right Sample Size

Define Your Research Question and Hypotheses

Ask yourself:

Example

Choose Your Statistical Test

Set Your Key Parameters

Significance Level (α\alphaα)

Statistical Power (1−β1 - \beta1−β)

Effect Size

Calculate the Sample Size

Option A: Use an Online Calculator (Recommended)

Option B: Use Statistical Software

Python

R

Option C: Use the Formula Directly

Adjust for Practical Factors

Dropout / Attrition

Budget and Timeline

Design Modifications

Common Scenarios

Clinical Trial

A/B Test (Website Conversion)

Survey / Questionnaire Study

Quick Reference: Sample Sizes for Common Designs

Test Your Understanding

View Step-by-Step Solution

Think About It

I've thought about it - Show me the solution

How to Determine Sample Size for a Study

Why Sample Size Matters

Too Small

Too Large

5 Steps to Determine the Right Sample Size

Define Your Research Question and Hypotheses

Ask yourself:

Example

Choose Your Statistical Test

Set Your Key Parameters

Significance Level (α\alphaα)

Statistical Power (1−β1 - \beta1−β)

Effect Size

Calculate the Sample Size

Option A: Use an Online Calculator (Recommended)

Option B: Use Statistical Software

Significance Level ( $\alpha$ )

Statistical Power ( $1 - \beta$ )

Significance Level ( $\alpha$ )

Statistical Power ( $1 - \beta$ )

Significance Level ( $\alpha$ )

Statistical Power ( $1 - \beta$ )

Significance Level ( $\alpha$ )

Statistical Power ( $1 - \beta$ )