A new arXiv preprint by Fabrizio Maturo and Annamaria Porreca extends distance-based regression to triangular fuzzy responses, where each observation has a center and spreads. The authors define direct, indirect, and total effects separately for the center and the spreads, and they show that the standard product-of-coefficients rule for indirect effects no longer holds when mediator spreads carry part of the association.

In simulations calibrated on real data, percentile bootstrap intervals for indirect effects had coverage between 0.905 and 0.975, and the empirical size of indirect-effect tests did not exceed 0.085. Ignoring the paths through the mediator spreads biased the indirect effect on the center by 16–20% at every sample size. The fuzzy model predicted better than a centers-only model only from n=250.

The method was applied to data from 91 university students collected in 2014. The indirect contrast through a course indicator accounted for about half of the model-based association between prior knowledge and the center of academic performance. Interactions of prior knowledge with sex and with motivation remained after a Holm adjustment, while the spreads of the covariates did not improve leave-one-out prediction.