Effect Size Calculator

Introduction

An effect size tells you how big or meaningful the difference is between two groups. It goes beyond just asking "is there a difference?" and answers "how large is that difference?" This is important in statistics because even a tiny difference can seem significant with a large enough sample. The Effect Size Calculator on this page helps you quickly find common effect size measures like Cohen's d, which compares the means of two groups using their standard deviations. Simply enter your data, and the calculator does the math for you. Whether you are working on a school project, a research paper, or just trying to understand your data better, this tool makes it simple to measure and interpret the real-world importance of your results.

How to Use Our Effect Size Calculator

Enter the data from two groups below, and this calculator will compute the effect size so you can see how big the difference between them really is.

Mean of Group 1: Type in the average value for your first group. This is the sum of all values in the group divided by the number of values. If you need help computing this, our Mean Median Mode Calculator can do it for you.

Mean of Group 2: Type in the average value for your second group. This works the same way as Group 1.

Standard Deviation of Group 1: Enter the standard deviation for your first group. This number tells you how spread out the data is from the mean. You can use our Standard Deviation Calculator to find this value from your raw data.

Standard Deviation of Group 2: Enter the standard deviation for your second group. Again, this shows how spread out the data is.

Sample Size of Group 1 (optional): Enter how many items or people are in your first group. This helps give a more accurate result.

Sample Size of Group 2 (optional): Enter how many items or people are in your second group. Like Group 1, this improves accuracy.

Once you hit calculate, the tool will give you the effect size value, commonly known as Cohen's d. A small effect size is around 0.2, a medium effect size is around 0.5, and a large effect size is 0.8 or higher. This helps you understand if the difference between your two groups is small, medium, or large in a meaningful way.

What Is Effect Size?

Effect size is a number that tells you how big the difference or relationship between two groups really is. While a p-value from a statistical test only tells you whether a result is likely real or due to chance, effect size tells you whether that result actually matters. A study can find a statistically significant difference that is so tiny it has no practical importance. Effect size solves this problem by putting a measurable value on the strength of a finding.

Why Effect Size Matters

Imagine two schools try different teaching methods. Both find a "significant" improvement in test scores. But one method raised scores by 2 points and the other by 20 points. The p-value alone cannot tell you which method worked better in a meaningful way. Effect size can. Researchers, doctors, teachers, and scientists all use effect size to decide if a treatment, program, or change is worth using in the real world.

Common Types of Effect Size

Cohen's d is the most widely used effect size measure. It shows the difference between two group means divided by a pooled standard deviation. Jacob Cohen, the statistician who popularized it, created simple guidelines for interpreting the result: a d of 0.2 is considered small, 0.5 is medium, and 0.8 is large. A related measure called Hedges' g applies a small correction that makes the estimate more accurate when sample sizes are small. Glass's Δ uses only the control group's standard deviation, which is useful when the treatment might change variability.

Eta squared (η²) is used with ANOVA tests. It tells you what fraction of the total variation in your data is explained by the group differences. Cohen's benchmarks for η² are 0.01 (small), 0.06 (medium), and 0.14 (large). A less biased version called omega squared (ω²) adjusts for sample size and is often preferred in published research.

The correlation coefficient r is itself an effect size. It measures the strength of a relationship between two variables on a scale from −1 to +1. Values of 0.10, 0.30, and 0.50 correspond to small, medium, and large effects. You can convert between r and Cohen's d using simple formulas, which is helpful when comparing results across different types of studies.

For categorical data in 2×2 tables, the odds ratio (OR) and risk ratio (RR) describe how much more likely an event is in one group compared to another. The phi coefficient (φ) and Cramér's V measure the strength of association in chi-square tests, with Cramér's V being especially useful when tables are larger than 2×2.

Additional Useful Measures

The Common Language Effect Size (CLES) turns Cohen's d into a probability that is easy for anyone to understand. For example, a CLES of 66% means that if you randomly pick one person from each group, there is a 66% chance the person from the treatment group will score higher. The Number Needed to Treat (NNT) is widely used in medicine. It tells you how many patients need to receive a treatment for one additional person to benefit compared to the control group. A lower NNT means the treatment is more effective.

When to Use Each Measure

• Comparing two group means: Use Cohen's d, Hedges' g, or Glass's Δ
• Comparing means from a t-test output: Use the d-from-t conversion
• Before-and-after measurements on the same people: Use paired samples d (d_z)
• ANOVA results: Use eta squared, partial eta squared, or omega squared
• Categorical outcomes (yes/no data): Use odds ratio, risk ratio, or phi
• Chi-square test results: Use phi or Cramér's V
• Clinical decision-making: Use NNT

Important Things to Keep in Mind

Cohen's benchmarks (small, medium, large) are general guidelines, not strict rules. In some fields, a "small" effect size can be very important. For example, in medicine, a small effect on survival rates can save thousands of lives. Always interpret effect size within the context of your specific field and research question. Also, effect size calculations assume your data is roughly normally distributed. You can use our Normal Distribution Calculator to explore how your data fits a bell curve. When data is heavily skewed or contains major outliers, the results may be less reliable. Additionally, when planning a study, knowing your expected effect size is essential for determining the right sample size, and reporting a confidence interval around your effect size estimate provides valuable context about the precision of your results.

Frequently Asked Questions

What is Cohen's d?

Cohen's d is a number that shows how big the difference is between two group averages. It is calculated by dividing the difference between the two means by the pooled standard deviation. A Cohen's d of 0.2 is small, 0.5 is medium, and 0.8 or higher is large.

What is the difference between Cohen's d and Hedges' g?

Cohen's d and Hedges' g use the same basic formula, but Hedges' g adds a small correction factor (called J) that makes it more accurate when your sample sizes are small. For large samples, the two values are nearly the same. Hedges' g is often preferred in meta-analyses.

What is Glass's Delta and when should I use it?

Glass's Δ divides the mean difference by only the control group's standard deviation instead of a pooled SD. Use it when you think the treatment may have changed how spread out the data is. Enter your control group as Group 2 in the calculator.

What does CLES mean in the results?

CLES stands for Common Language Effect Size. It tells you the probability that a randomly picked person from Group 1 will score higher than a randomly picked person from Group 2. For example, a CLES of 64% means there is a 64% chance of this happening.

Can I calculate effect size if I only have a t-test statistic?

Yes. Use the "Cohen's d from t-test Statistic" section. Enter your t value and the sample sizes for both groups. The calculator uses the formula d = t × √(1/n₁ + 1/n₂) to convert it into Cohen's d.

What is the difference between d_z and d_av for paired samples?

d_z divides the mean difference by the standard deviation of the differences. d_av corrects for the correlation between the two measurements, giving an effect size closer to an independent-groups Cohen's d. Use d_z when reporting paired t-test results, and d_av when you want to compare your result with independent-groups studies.

What is eta squared and how is it different from partial eta squared?

Eta squared (η²) is the ratio of the between-group variability to the total variability in your data. Partial η² divides the between-group variability by only the between-group plus the error variability, ignoring other factors. In a one-way ANOVA they are the same, but in more complex designs partial η² is usually larger.

What is omega squared and why is it used?

Omega squared (ω²) is a less biased version of eta squared. Eta squared tends to overestimate the true effect size, especially with small samples. Omega squared corrects for this, so many researchers prefer to report it.

How do I read the odds ratio result?

An odds ratio (OR) of 1 means no difference between groups. An OR below 1 means the event is less likely in the treatment group. An OR above 1 means the event is more likely in the treatment group. For example, an OR of 0.43 means the treatment group has roughly 57% lower odds of the event.

What is the Number Needed to Treat (NNT)?

NNT tells you how many people need to receive a treatment for one extra person to benefit compared to the control group. A lower NNT means the treatment works better. An NNT of 5 means you need to treat 5 people for 1 additional person to have a good outcome.

How do I convert between correlation r and Cohen's d?

Use the r ↔ d Converter section. To go from r to d, the formula is d = 2r / √(1 - r²). To go from d to r, the formula is r = d / √(d² + 4). Just type in your value and click the convert button.

What is Cramér's V and how is it different from phi?

Both measure the strength of association in a chi-square test. Phi (φ) works best for 2×2 tables. Cramér's V adjusts for larger tables by dividing by the minimum number of rows or columns minus one. For a 2×2 table, phi and Cramér's V are the same value.

What does the confidence interval for Cohen's d tell me?

The confidence interval gives a range where the true effect size likely falls. For example, a 95% CI of [0.05, 1.04] means you can be 95% confident the real effect size is somewhere between 0.05 and 1.04. If the interval includes 0, the effect may not be meaningful.

What does the overlap visualization show?

It shows two bell curves representing the two groups. The more they overlap, the smaller the effect size. The chart also displays the Overlap coefficient (OVL), CLES, and U₃, which is the percentage of the treatment group that scores above the control group's average.

Can a negative Cohen's d be valid?

Yes. A negative Cohen's d simply means Group 2 has a higher mean than Group 1. The sign shows direction. The absolute value tells you the size of the effect. For interpretation, a d of -0.5 is the same magnitude as +0.5.

What does Cohen's f represent?

Cohen's f is an effect size measure used with ANOVA. It is calculated as f = √(η² / (1 - η²)). Cohen's guidelines say 0.10 is small, 0.25 is medium, and 0.40 is large. It is commonly used in power analysis for ANOVA designs.

Do I need sample sizes to calculate effect size?

For basic Cohen's d with equal groups, you only need the means and standard deviations. However, sample sizes are needed for Hedges' g correction, confidence intervals, and the t-test conversion. Adding sample sizes gives you more accurate and complete results.

What is the weighted pooled standard deviation?

When group sizes are unequal, the pooled SD gives more weight to the larger group. The formula is Sp = √[((n₁-1)SD₁² + (n₂-1)SD₂²) / (n₁+n₂-2)]. This produces a more accurate estimate than simply averaging the two SDs.