Abstract

Outliers in datasets severely distort the detection of multicollinearity-an issue arising from strong correlations between explanatory variables in linear regression analysis. Conventional diagnostic tools, including correlation matrices, classical variance inflation factors (CVIF), and robust variance inflation factors (RVIF), fail to maintain accuracy and reliability in the presence of outliers. This study presents the Modified Fuzzy Robust Variance Inflation Factors (MFRVIF) method, a powerful and adaptive approach that combines fuzzy with robust statistical techniques to detect multicollinearity while neutralizing the disruptive effects of outliers. Empirical evidence demonstrates that MFRVIF decisively outperforms CVIF and RVIF, delivering superior accuracy and consistency across contaminated datasets. This groundbreaking method redefines multicollinearity diagnostics, providing an essential and robust solution for precise regression analysis in complex, real-world data environments.