João Pedro Azevedo

João Pedro Azevedo

Chief Statistician @ UNICEF | Development Economist

FACTORTEST

Test whether a correlation matrix is suitable for factor analysis using Bartlett's sphericity test (is the matrix the identity?) and the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy.

SSC Stata SSC GitHub Issues

Installation

Stata ssc install factortest

Description

factortest evaluates whether factor analysis is suitable for a given dataset by performing two key tests:

  1. Bartlett’s Test for Sphericity: Examines the correlation matrix’s determinant by converting it to a chi-square statistic. The null hypothesis is that variables are non-collinear.

  2. Kaiser-Meyer-Olkin (KMO) Measure: Compares observed correlation coefficients against partial correlation coefficients with the following interpretation scale:

    • 0.90+: Excellent
    • 0.80–0.89: Meritorious
    • 0.70–0.79: Middling
    • 0.60–0.69: Mediocre
    • 0.50–0.59: Miserable
    • Below 0.50: Unacceptable

Example

* Bartlett's sphericity and KMO tests
factortest price mpg rep78 rep77 headroom rseat trunk

Citation

Azevedo, J.P. (2003). “FACTORTEST: Stata module to perform tests for factor analysis suitability.” Statistical Software Components S436001, Boston College Department of Economics.