cohens-d-effect-size 1.2.0

A Python package for calculating Cohen's d effect size

Cohen's d Effect Size Calculator

A Python package for calculating Cohen's d effect size with comprehensive options for handling missing data, different axes, and pooled vs unpooled standard deviations.

Features

• One-sample and two-sample Cohen's d: Calculate effect sizes for single samples against zero or between two independent groups
• Flexible data handling: Support for 1D and multi-dimensional arrays with axis specification
• Missing data policies: Choose how to handle NaN values (propagate, raise, or omit)
• Pooled and unpooled variance: Option to use pooled standard deviation or first sample's standard deviation
• NumPy compatibility: Full integration with NumPy arrays and broadcasting rules
• Comprehensive testing: Extensive test suite ensuring numerical accuracy and edge case handling

Installation

Install from PyPI:

pip install cohens-d-effect-size

Install from source:

git clone https://github.com/DawitLam/cohens-d-scipy.git
cd cohens-d
pip install -e .

Quick Start

import numpy as np
from cohens_d import cohens_d

# Two-sample Cohen's d
group1 = np.array([1, 2, 3, 4, 5])
group2 = np.array([2, 3, 4, 5, 6])
effect_size = cohens_d(group1, group2)
print(f"Cohen's d: {effect_size:.3f}")

# One-sample Cohen's d (against zero)
sample = np.array([0.5, 1.0, 1.5, 2.0, 2.5])
effect_size = cohens_d(sample)
print(f"One-sample Cohen's d: {effect_size:.3f}")

Parameters

• x: First sample or the sample to compare against zero
• y: Second sample (optional). If not provided, calculates one-sample Cohen's d
• axis: Axis along which to compute the effect size (default: None)
• nan_policy: How to handle NaN values ('propagate', 'raise', 'omit')
• ddof: Delta degrees of freedom for standard deviation calculation (default: 1)
• keepdims: Whether to keep reduced dimensions as size 1 (default: False)
• alternative: Alternative hypothesis ('two-sided', 'less', 'greater')
• pooled: Whether to use pooled standard deviation for two-sample case (default: True)

Examples

Basic Usage

import numpy as np
from cohens_d import cohens_d

# Create sample data
np.random.seed(42)
control = np.random.normal(0, 1, 100)
treatment = np.random.normal(0.5, 1, 100)

# Calculate Cohen's d
d = cohens_d(control, treatment)
print(f"Effect size: {d:.3f}")

Multi-dimensional Arrays

# 2D arrays - calculate along different axes
data_2d = np.random.normal(0, 1, (10, 5))
treatment_2d = np.random.normal(0.3, 1, (10, 5))

# Along rows (axis=0)
d_rows = cohens_d(data_2d, treatment_2d, axis=0)
print(f"Effect sizes along axis 0: {d_rows}")

# Along columns (axis=1)  
d_cols = cohens_d(data_2d, treatment_2d, axis=1)
print(f"Effect sizes along axis 1: {d_cols}")

Handling Missing Data

# Data with missing values
data_with_nan = np.array([1, 2, np.nan, 4, 5])
control_with_nan = np.array([2, 3, 4, np.nan, 6])

# Omit missing values
d_omit = cohens_d(data_with_nan, control_with_nan, nan_policy='omit')
print(f"Effect size (omit NaN): {d_omit:.3f}")

# Raise error on missing values
try:
    d_raise = cohens_d(data_with_nan, control_with_nan, nan_policy='raise')
except ValueError as e:
    print(f"Error: {e}")

Pooled vs Unpooled Standard Deviation

# Using pooled standard deviation (default)
d_pooled = cohens_d(group1, group2, pooled=True)

# Using first sample's standard deviation only
d_unpooled = cohens_d(group1, group2, pooled=False)

print(f"Pooled: {d_pooled:.3f}, Unpooled: {d_unpooled:.3f}")

Interpretation

Cohen's d effect size interpretation:

• Small effect: d ≈ 0.2
• Medium effect: d ≈ 0.5
• Large effect: d ≈ 0.8

def interpret_cohens_d(d):
    \"\"\"Interpret Cohen's d effect size.\"\"\"
    abs_d = abs(d)
    if abs_d < 0.2:
        return "negligible"
    elif abs_d < 0.5:
        return "small"
    elif abs_d < 0.8:
        return "medium"
    else:
        return "large"

d = cohens_d(control, treatment)
interpretation = interpret_cohens_d(d)
print(f"Effect size: {d:.3f} ({interpretation})")

Requirements

• Python ≥ 3.8
• NumPy ≥ 1.19.0

Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

Development Setup

git clone https://github.com/DawitLam/cohens-d.git
cd cohens-d
pip install -e ".[dev]"

Running Tests

pytest tests/

Code Formatting

black cohens_d/ tests/
flake8 cohens_d/ tests/

License

This project is licensed under the MIT License - see the LICENSE file for details.

References

1. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.
2. Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs. Frontiers in Psychology, 4, 863.

Changelog

0.1.0 (2024-01-XX)

• Initial release
• Support for one-sample and two-sample Cohen's d
• Comprehensive parameter options
• Full NumPy compatibility
• Extensive test coverage