hypothesize: Now on CRAN

hypothesize is now on CRAN.

What It Does

hypothesize provides a consistent API for hypothesis testing in R. It defines generic methods that any hypothesis test can implement:

• pval() - Extract the p-value
• test_stat() - Get the test statistic
• dof() - Retrieve degrees of freedom
• is_significant_at() - Check significance at a given level

The package ships with two implementations:

• Likelihood Ratio Test (LRT) for comparing nested models
• Wald Test for testing parameter estimates

Why

When building statistical libraries, I kept implementing ad-hoc hypothesis test structures. Different packages, different interfaces, no composability. hypothesize standardizes this: any package can wrap its tests in a consistent interface, and statistical workflows can treat all tests uniformly.

It’s a small package. That’s the point.

Installation

install.packages("hypothesize")
Copy

Quick Example

library(hypothesize)

# Likelihood Ratio Test
result <- lrt(null_loglik = -100, alt_loglik = -96, dof = 3)
print(result)

# Check significance
is_significant_at(result, 0.05)

# Extract components
pval(result)
test_stat(result)
Copy

Documentation

Full documentation is at queelius.github.io/hypothesize.

What’s Next

I have several other R packages in the pipeline for CRAN submission, including packages for likelihood-based inference and reliability analysis.

Links

• CRAN: CRAN.R-project.org/package=hypothesize
• GitHub: github.com/queelius/hypothesize
• Documentation: queelius.github.io/hypothesize