Fuzzy DEMATEL

Description

Experts rate how strongly each factor influences every other; fuzzy matrix algebra then ranks factors by importance and separates causes from effects.

DEMATEL uses graph theory and matrix algebra to screen influencing factors in a complex system. Experts rate the direct influence of each factor on each other on a 0 to 4 scale. Because such judgements are vague, the fuzzy version converts ratings to triangular fuzzy numbers, defuzzifies each expert's matrix with the CFCS method and averages them into a direct relation matrix. This is normalised and the total relation matrix computed as T = D(I - D) inverse, capturing indirect influence. Row sums R and column sums C give centrality R + C (importance) and cause degree R - C (positive means cause, negative means effect). A causal diagram plots both. In UX it turns a long list of experience factors into a small set of causal levers.

Attributes

Loss type
Siloed
Effort
Weeks

Steps

  1. Build the factor list from literature and prior research, then trim it with a short questionnaire.
  2. Have each expert rate the influence of every factor on every other: 0 none, 1 weak, 2 moderate, 3 strong, 4 very strong.
  3. Convert ratings to triangular fuzzy numbers (for example none = 0, 0, 0.25; weak = 0, 0.25, 0.5).
  4. Defuzzify each expert's matrix with CFCS and average across experts to obtain the direct relation matrix Z.
  5. Normalise Z by the largest row or column sum to obtain D, then compute the total relation matrix T = D(I - D)^-1.
  6. Compute row sums R and column sums C, centrality R + C and cause degree R - C.
  7. Plot the cause-effect diagram and select key factors by centrality, prioritising high-centrality cause factors as design levers.

Sources

  • DUXU 2022: UX Research, Design, and Assessment (Soares et al., 2022)

Also known as

  • DEMATEL
  • Decision-Making Trial and Evaluation Laboratory
  • Fuzzy-DEMATEL model

Categories

  • Strategy & Collaboration
  • Game Theory & Decision Science