Robust Factor Analysis
Also known as: robust factor analysis, outlier-resistant factor analysis, MCD-based factor analysis, Robust Faktör Analizi
Robust Factor Analysis recovers the latent factor structure of multivariate continuous data while resisting the distorting pull of outliers. Introduced by Pison, Rousseeuw, Filzmoser and Croux (2003), it replaces the classical sample covariance with a robust estimator such as the Minimum Covariance Determinant (MCD) or an S-estimator before extracting factors.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Robust Factor Analysis with multivariate continuous data when you suspect outliers or non-normality and still want to uncover or describe a latent factor structure. It assumes continuous variables and is designed precisely for the situation where contamination or heavy tails would distort a classical solution. A reasonably large sample is needed — at least about 100 observations — because the robust covariance estimate must be reliable.
Strengths & limitations
- Extracts a factor structure that is not pulled toward outliers, unlike classical factor analysis.
- Does not require multivariate normality, making it suitable for contaminated or heavy-tailed data.
- Built on principled robust covariance estimators (MCD, S-estimators) with established breakdown properties.
- Requires a fairly large sample (about 100+) for the robust covariance to be trustworthy.
- With fewer than 30 observations the robust covariance estimate becomes unreliable and classical PCA is preferable.
- When the proportion of outliers is very high (above roughly 25%), even an MCD-based covariance can break down.
Frequently asked
How is this different from ordinary factor analysis?
The model is the same, but the covariance it factors is estimated robustly. Classical factor analysis uses the ordinary sample covariance, which a few outliers can distort; Robust Factor Analysis uses an MCD or S-estimator covariance that is computed from the bulk of the data, so the extracted factors are not driven by extreme cases.
What is the MCD estimator?
The Minimum Covariance Determinant fits the location and covariance to the subset of observations whose covariance has the smallest determinant — effectively the densest, most typical core of the data. Because outliers are excluded from that core, the resulting covariance, and the factors derived from it, resist contamination.
How large a sample do I need?
Plan for at least about 100 observations. Below roughly 30 the robust covariance estimate is not trustworthy, and classical PCA is the recommended fallback in that small-sample situation.
What if a large share of my data are outliers?
When the outlier ratio climbs above about 25%, even an MCD-based covariance can break down. In that regime robust PCA is the preferred alternative for recovering structure.
Sources
- Pison, G., Rousseeuw, P. J., Filzmoser, P., & Croux, C. (2003). Robust factor analysis. Journal of Multivariate Analysis, 84(1), 145-172. DOI: 10.1016/S0047-259X(02)00007-6 ↗
- Hubert, M., Rousseeuw, P. J., & Vanden Branden, K. (2005). ROBPCA: A new approach to robust principal component analysis. Technometrics, 47(1), 64-79. DOI: 10.1198/004017004000000563 ↗
How to cite this page
ScholarGate. (2026, June 1). Robust Factor Analysis. ScholarGate. https://scholargate.app/en/statistics/robust-factor-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Factor AnalysisResearch Statistics↔ compare
- Influence DiagnosticsStatistics↔ compare
- Principal Component AnalysisMachine learning↔ compare
- Robust Covariance (MCD)Statistics↔ compare
- Robust PCAStatistics↔ compare