researchpy

ResearchPy is a Python package that is designed to be easy to use and deliver univariate, bivariate, and multivariate models and statistical tests with clear and informative outputs ready for interpretation, further analysis, and downstream export without additional formatting. Open-source, clear function names, intuitive parameters, and robust documentation.

Built for researchers, analysts transitioning from SPSS/Stata/R, and students learning statistics.

Features

• Descriptive statistics — summarize(), summary_cont(), summary_cat(), codebook()
• Difference tests — Independent t-test, paired t-test, Welch's t-test, Wilcoxon signed-rank via difference_test()
• ANOVA — Type I, II, and III sum of squares with anova()
• Correlation — Correlation matrices and testing
• Crosstabs — Cross-tabulation with chi-square testing
• OLS regression — Ordinary least squares modeling
• Effect sizes — Cohen's d, Hedge's g, Glass's delta, eta/epsilon/omega squared, and more
• Confidence intervals — Included by default in descriptive and inferential output
• Structured output — Results returned as pandas DataFrames, ready for reporting

Installation

pip install researchpy

Quick Start

import pandas as pd
import researchpy as rp

# Load example data
df = pd.DataFrame({
    "score": [88, 92, 75, 85, 90, 78, 95, 70, 82, 87,
              72, 68, 80, 76, 74, 69, 71, 77, 73, 79],
    "group": ["treatment"]*10 + ["control"]*10
})

ANOVA

model = rp.anova("score ~ C(group)", data=df) # Type III sum of squares by default
descriptives, results = model.results()
print(descriptives, results, sep = "\n"*2)

Codebook

rp.codebook(df)

Difference Test

desc, results = rp.difference_test("score ~ C(group)", data=df).conduct(effect_size="all")
print(desc, results, sep = "\n"*2)

Summarize

# Continuous summary statistics
rp.summarize(df["score"], stats=["N", "Mean", "SD", "SE", "CI"])

# Categorical frequency table
rp.summary_cat(df["group"])

Requirements

• Python ≥ 3.12
• pandas ≥ 3.0.1
• numpy ≥ 2.5.0
• scipy ≥ 1.17.1
• statsmodels ≥ 0.14.0
• patsy ≥ 1.0.2 (for legacy formula parsing, will be removed in future versions)
• requests ≥ 2.34.2
• formulaic ≥ 1.2.2 (for new formula parsing)

Documentation

Full documentation is available at researchpy.readthedocs.io.

License

ResearchPy is released under the MIT License.

Citation

If you use ResearchPy in your research, please cite:

Bryant, C. (2018–2026). researchpy (Version X.Y.Z) [Python package]. https://github.com/Corey-Bryant/researchpy

To find your installed version for citation:

import researchpy
print(researchpy.__version__)

Current citation with version number:

Bryant, C. (2018–2026). researchpy (Version 0.3.7.2) [Python package]. https://github.com/Corey-Bryant/researchpy