Should parsimony really still be the gold standard in modelling?

stats.stackexchange.com