CRITIC-M
Criteria Importance Through Intercriteria Correlation - Modified (CRITIC-M) · Also known as: CRITIC-M, Modified CRITIC
CRITIC-M (Criteria Importance Through Intercriteria Correlation - Modified) is an objective weight derivation method that extends the classical CRITIC approach. It assigns weights to criteria based on two intrinsic properties of the decision matrix: variance (how much a criterion differentiates alternatives) and correlation (how much a criterion conflicts with or supplements others). Modified variants adjust the formulation to improve robustness or interpretability.
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 CRITIC-M when you want objective, data-driven criterion weights without expert elicitation. It is ideal for large decision matrices where patterns are evident in data, and when criteria independence or redundancy is important. Avoid it when data is sparse, when criteria are naturally correlated by design, or when expert judgment is critical to decision acceptance.
Strengths & limitations
- Fully objective; no subjective expert input required, reducing bias
- Accounts for criterion diversity and discrimination power simultaneously
- Computationally simple and efficient; applicable to large matrices
- Data-driven approach provides empirical justification for weights
- Identifies redundant criteria naturally; highly correlated criteria receive lower weights
- Assumes that variance reflects importance; in some contexts, stable criteria may be more important than variable ones
- Correlation structure may be distorted by outliers; robust variants are needed for contaminated data
- Does not account for preference direction; a criterion that varies but always performs poorly may still receive high weight
- Weights can be unstable if data is sparse or if a few outliers dominate variance calculations
Frequently asked
Should I use CRITIC or CRITIC-M?
Original CRITIC uses max normalization; CRITIC-M variants may use vector normalization or other schemes. Vector normalization is more robust to scale differences. Run both and compare results; if they differ significantly, investigate the normalization scheme.
How do I handle missing values?
Impute missing values using domain knowledge or statistical methods (mean imputation, k-NN imputation). Alternatively, calculate variance and correlation only on complete cases. Perform sensitivity analysis to verify results are robust to imputation method.
Can CRITIC-M weights be negative?
No, conflict scores are always non-negative, so normalized weights are always non-negative. If a criterion has very low variance and very high correlation to others, it receives near-zero weight—often appropriately.
Sources
- Diakoulaki, D., Mavrotas, G., & Papayannakis, L. (1995). Determining objective weights in multiple criteria problems: The CRITIC method. Computers & Operations Research, 22(7), 763-770. DOI: 10.1016/0305-0548(94)00059-H ↗
- Jahan, A., Mustapha, F., Sapuan, S. M., Ismail, M. Y., & Badruddin, I. A. (2012). A comprehensive VIKOR method for material selection. Materials & Design, 32(3), 1215-1221. DOI: 10.1016/j.matdes.2010.10.015 ↗
How to cite this page
ScholarGate. (2026, June 3). Criteria Importance Through Intercriteria Correlation - Modified (CRITIC-M). ScholarGate. https://scholargate.app/en/decision-making/critic-m
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.
Compare side by side →