Effect Size Calculator: Cohen’s d and Pearson’s r
Calculate d and r Using Means and Standard Deviations
The average value for the first group.
The average value for the second group.
The measure of data spread for Group 1.
The measure of data spread for Group 2.
The number of observations in Group 1.
The number of observations in Group 2.
Calculated Results
0.33
0.16
Interpretation of d: Small effect size
Mean Difference (M₁ – M₂): 5.00
Pooled Standard Deviation (SDₚ): 15.00
What are Cohen’s d and Pearson’s r?
When comparing two groups, such as a treatment group and a control group in a study, it’s not enough to know if their means are different. The magnitude of that difference, known as the effect size, is crucial for practical interpretation. This is where you would want to calculate d and r using means and standard deviations. Both Cohen’s d and Pearson’s r are standardized measures that quantify this effect.
Cohen’s d measures the difference between two means in terms of standard deviations. For example, a d value of 0.5 means the difference between the two group averages is half a standard deviation. It’s one of the most common effect size indicators used in research.
Pearson’s r, or the correlation coefficient, measures the strength and direction of a linear relationship between two variables. When used in the context of comparing two groups, it’s referred to as the point-biserial correlation, which quantifies the relationship between group membership (e.g., Group 1 vs. Group 2) and the outcome variable. Values range from -1 (perfect negative correlation) to +1 (perfect positive correlation), with 0 indicating no correlation. Our correlation coefficient calculator provides more detail on this metric.
The Formulas for Cohen’s d and Pearson’s r
To calculate these effect sizes from summary data, you need the mean, standard deviation, and sample size for each of the two groups. The process involves first calculating an “average” standard deviation across both groups.
1. Pooled Standard Deviation (SDₚ)
The first step is to calculate the pooled standard deviation, which is a weighted average of the two groups’ standard deviations. The formula is:
SDₚ = √[((n₁ - 1) * s₁² + (n₂ - 1) * s₂²) / (n₁ + n₂ - 2)]
This pooled standard deviation formula gives more weight to the group with the larger sample size.
2. Cohen’s d Formula
Once you have the pooled standard deviation, calculating Cohen’s d is straightforward:
d = (M₁ - M₂) / SDₚ
3. Converting Cohen’s d to Pearson’s r
You can then convert Cohen’s d to the effect size correlation r using the following formula:
r = d / √(d² + (n₁ + n₂)² / (n₁ * n₂))
| Variable | Meaning | Unit | Typical Range |
|---|---|---|---|
| M₁, M₂ | The mean (average) of Group 1 and Group 2 | Unitless (or depends on data context) | Any real number |
| s₁, s₂ | The standard deviation of Group 1 and Group 2 | Unitless (or same as mean) | Positive real numbers |
| n₁, n₂ | The sample size (number of participants) of Group 1 and Group 2 | Count | Integers > 2 |
| SDₚ | Pooled Standard Deviation | Unitless (or same as mean) | Positive real numbers |
Practical Examples
Example 1: Educational Intervention
A researcher tests a new teaching method. A class of 30 students (Group 1) uses the new method, while a class of 32 students (Group 2) uses the traditional method. At the end of the semester, they take a standardized test.
- Group 1 Inputs: Mean (M₁) = 85, Standard Deviation (s₁) = 8, Sample Size (n₁) = 30
- Group 2 Inputs: Mean (M₂) = 81, Standard Deviation (s₂) = 7.5, Sample Size (n₂) = 32
Using the calculator, we find:
- Cohen’s d ≈ 0.52 (a medium effect size)
- Pearson’s r ≈ 0.25
This indicates the new teaching method had a medium-sized positive effect on test scores.
Example 2: Clinical Drug Trial
A pharmaceutical company is testing a new drug to lower blood pressure. 100 patients receive the drug (Group 1) and 100 receive a placebo (Group 2).
- Group 1 Inputs: Mean Systolic BP (M₁) = 130 mmHg, SD (s₁) = 12 mmHg, Sample Size (n₁) = 100
- Group 2 Inputs: Mean Systolic BP (M₂) = 140 mmHg, SD (s₂) = 13 mmHg, Sample Size (n₂) = 100
The calculator would show:
- Cohen’s d ≈ -0.79 (a medium-to-large effect size)
- Pearson’s r ≈ -0.37
The negative sign indicates the drug group had a lower mean, which in this case is the desired outcome. The magnitude shows the drug had a substantial effect. The statistical significance calculator could then be used to determine the p-value.
How to Use This Effect Size Calculator
This tool makes it easy to calculate d and r using means and standard deviations. Follow these simple steps:
- Enter Group 1 Data: Input the mean (M₁), standard deviation (SD₁), and sample size (n₁) for your first group.
- Enter Group 2 Data: Input the mean (M₂), standard deviation (SD₂), and sample size (n₂) for your second group.
- View Real-Time Results: The calculator automatically updates the results for Cohen’s d, Pearson’s r, and intermediate values as you type. No need to press a calculate button.
- Interpret the Output: The primary results show the final d and r values. A text interpretation of Cohen’s d (e.g., “small”, “medium”, “large”) is provided for quick assessment. You can also see the mean difference and pooled standard deviation.
- Visualize the Data: The bar chart provides a simple visual representation of the difference between the two means.
- Reset or Copy: Use the “Reset” button to clear all fields to their default values. Use the “Copy Results” button to copy a summary to your clipboard.
Key Factors That Affect Effect Size
- Magnitude of Mean Difference: The larger the difference between the two means (M₁ – M₂), the larger Cohen’s d will be. This is the “signal” in the signal-to-noise analogy.
- Data Variability (Standard Deviation): The smaller the standard deviations of the groups, the larger Cohen’s d will be. High variability (or “noise”) makes a given mean difference less impressive. A smaller pooled standard deviation leads to a larger d.
- Sample Size (n): While Cohen’s d itself is not directly influenced by sample size in the same way p-values are, the stability of the estimate is. Larger samples provide more accurate estimates of the true population means and standard deviations. Sample sizes are also required to calculate the pooled SD correctly when n1 and n2 are different.
- Measurement Error: Imprecise measurement tools can increase the observed standard deviation, which artificially lowers the calculated effect size.
- Homogeneity of Variances: The calculation for Cohen’s d assumes that the standard deviations of the two groups are reasonably similar. If they are wildly different, the pooled standard deviation may not be a good representation of the overall variance, and alternative effect size measures might be more appropriate.
- Outliers: Extreme scores can heavily influence both the mean and the standard deviation, potentially distorting the resulting effect size. It’s important to screen data for outliers before analysis.
Frequently Asked Questions (FAQ)
1. What is a “good” Cohen’s d value?
General guidelines interpret d as: 0.2 = small effect, 0.5 = medium effect, and 0.8 = large effect. However, context is crucial; a “small” effect in one field (e.g., medicine) might be considered large in another (e.g., social sciences).
2. Can Cohen’s d be negative?
Yes. A negative d value simply means the mean of the second group (M₂) is larger than the mean of the first group (M₁). The magnitude (the absolute value) of d is what indicates the size of the effect.
3. Why do I need sample size to calculate Cohen’s d?
When the sample sizes of the two groups are different, the pooled standard deviation must be weighted. The group with the larger sample size provides a more reliable estimate of variance, so it contributes more to the pooled value. If your sample sizes are equal, the formula simplifies, but it’s still best practice to use the full formula.
4. What’s the difference between Cohen’s d and a p-value?
A p-value tells you about statistical significance (i.e., whether the observed difference is likely due to chance). An effect size like Cohen’s d tells you about practical significance (i.e., the magnitude or importance of the difference). A tiny, practically meaningless effect can be statistically significant with a large enough sample size.
5. Are the units of my data important?
No, because Cohen’s d is a standardized measure. By dividing the mean difference by the standard deviation, you are expressing the difference in standard deviation units. This makes the result “unitless” and allows for comparison of effect sizes across different studies that may have used different measurement scales.
6. When should I use Pearson’s r instead of Cohen’s d?
Both are valid effect sizes. Pearson’s r is often preferred when you want to express the effect as a correlation. It is bounded between -1 and 1, which can sometimes be easier to interpret. Cohen’s d is not bounded. The choice often comes down to convention within a specific field of study.
7. What if my standard deviations are very different?
If the standard deviation of one group is much larger than the other (e.g., more than double), the assumption of homogeneity of variances is violated. In this case, you might consider using Glass’s Δ (Delta), which uses only the standard deviation of the control group as the denominator.
8. Can I calculate d and r from raw data?
Yes. If you have raw data, you would first calculate the mean, standard deviation, and count (n) for each group, and then use those values in this calculator. You would not need to calculate d and r using means and standard deviations manually.