Weighting method · Diakoulaki et al., 1995

CRITIC

CRiteria Importance Through Intercriteria Correlation

Objective weights straight from the data: criteria that vary more and correlate less carry more weight.

How it works

1. Normalize the matrix direction-aware to [0, 1].
2. Measure each criterion's contrast (standard deviation) and conflict (Σ of 1 − correlation with the other criteria).
3. Weight_j ∝ contrast × conflict, normalized to sum to 1.

Use it when

• No decision-maker preference is available, or you want a judgment-free baseline to compare against.
• You suspect some criteria are redundant (highly correlated) and should count less.

Watch out for

• Statistical importance is not the same as decision importance a criterion can vary a lot and still not matter to you.

Parameters

None fully data-driven.

Cite the method

Diakoulaki, D., Mavrotas, G., & Papayannakis, L. (1995). Determining objective weights in multiple criteria problems: The CRITIC method. Computers & Operations Research, 22(7), 763-770.

@article{diakoulaki1995critic,
  author  = {Diakoulaki, Danae and Mavrotas, George and Papayannakis, Lefteris},
  title   = {Determining objective weights in multiple criteria problems: The {CRITIC} method},
  journal = {Computers \& Operations Research},
  volume  = {22},
  number  = {7},
  pages   = {763770},
  year    = {1995},
  doi     = {10.1016/0305-0548(94)00059-H}
}