edstan for Python

A python module that simplifies the fitting of common Bayesian item response theory models using Stan. It is compatible with and extends the functionality of pystan.

Features

• Streamlined interface to common Bayesian item response models using Stan
• Models include: Rasch, two-parameter logistic, (generalized) rating scale, and (generalized) partial credit
• Posterior summaries tailored to item response models

Installation

edstan depends on the successful installation of pystan, so please see the pystan installation guide. Note that compatibility with Windows OS may be limited.

edstan may subsequently be installed with pip:

pip install edstan

API Reference

The API reference may be found at Read the Docs.

Quickstart

Here is an example of running a model using data from a response matrix:

from edstan import EdStanModel
import numpy as np
import pandas as pd

# Simulate a "wide format" data frame of item responses for
# 5 items and 100 persons. Responses are scored 0 or 1.
rng = np.random.default_rng(seed=42)
data = pd.DataFrame(rng.binomial(1, p=.5, size=(100, 5)))
data.columns = [f"Question {i}" for i in range(5)]
data.index = [f"Respondent {i}" for i in range(100)]

# Instantiate the model, selecting the Rasch model
model = EdStanModel("rasch")

# Sample from the model by MCMC
fit = model.sample_from_wide(data)

# View a posterior summary of the item (and person distribution)
# parameters
print(fit.item_summary())

# View a posterior summary of the person parameters
print(fit.person_summary())

Alternatively, this is an example of using long format data:

from edstan import EdStanModel
import numpy as np
import pandas as pd

# Simulate a "long format" data frame of item responses for
# 5 items and 100 persons. Responses are scored 0, 1, or 2.
rng = np.random.default_rng(seed=42)
data = pd.DataFrame(
  {
     "person": [f"Person {j}" for j in range(100) for i in range(5)],
     "item": [f"Item {i}" for j in range(100) for i in range(5)],
     "response": rng.binomial(2, p=.5, size=5*100),
  }
)

# Instantiate the model, choosing the generalized partial
# credit model
model = EdStanModel("gpcm")

# Sample from the model by MCMC
fit = model.sample_from_long(
  ii=data['item'],
  jj=data['person'],
  y=data['response'],
)

# View a posterior summary of the item (and person distribution)
# parameters
print(fit.item_summary())

# View a posterior summary of the person parameters
print(fit.person_summary())