psymetrics 0.1.0

Reliability, factor analysis and SEM for psychologists with an R-like API

psymetrics

Reliability, factor analysis and SEM for psychologists, with an R-like functional API. Part of the ANOVA Methods family (psystats, psymetrics, psyreport).

Free and open (MIT). Results print R-style summaries and render to APA-7 through psyreport.report().

Install

pip install psymetrics

What's inside

Reliability — alpha (Cronbach's alpha with the Feldt confidence interval, corrected item-total correlations, alpha-if-dropped, reverse keying), like psych::alpha.

Factorability — kmo (Kaiser-Meyer-Olkin, overall and per item) and bartlett (test of sphericity).

Exploratory factor analysis — efa with selectable estimator (minres, ml, principal) and rotation (varimax, promax, oblimin, quartimax, ...); returns loadings, communalities, eigenvalues, variance explained, and the factor correlation matrix for oblique rotations.

Confirmatory factor analysis and SEM — cfa using lavaan-style syntax; returns standardized loadings and the usual fit indices (χ², df, p, CFI, TLI, RMSEA).

Example

import pandas as pd
from psymetrics import alpha, kmo, efa, cfa

df = pd.read_csv("scale.csv")
items = df[[f"q{i}" for i in range(1, 11)]]

print(alpha(items))                 # like psych::alpha
print(kmo(items))                   # sampling adequacy
print(efa(items, 3, method="minres", rotation="promax"))

model = """
visual  =~ x1 + x2 + x3
textual =~ x4 + x5 + x6
speed   =~ x7 + x8 + x9
"""
print(cfa(model, df))               # lavaan-style CFA

EFA is validated against factor_analyzer and CFA fit indices against semopy.

Note: factor_analyzer (0.5.x) requires scikit-learn < 1.6; this pin is declared as a dependency so efa works out of the box.