Fast and frugal versus regression models of human judgement

Statistical and Psychological Models of Doctors’ Judgments of Heart Patients

“…The other group of studies -the judgment studies -is the mirror image of the ecological studies; instead, the dependent variable is judgments data rather than ecological facts [6,14,16,19]. As far as we know, no cross-validation has been used in this group of studies for comparing the validity of LR and F&F models.…”

“…Empirical results are seemingly contradictory of whether F&F models provide higher predictive accuracy than LR models [6,10,[12][13][14][15][16]. Two groups of studies can be discerned that may be denoted as ecological and judgment studies, respectively.…”

“…as LR 5, p<0.10, denoted as LR 10, and inclusion of all cues, denoted as LR Enter, respectively) and two F&F heuristics, the matching heuristic reported by Dhami and Harries [6], denoted as F&F 1, and extended matching heuristic reported by Backlund et al [12], denoted as F&F 2). Moreover, we used cross-validation for testing the validity of fitted models by first fitting the models to a subset of the patient cases (fitting set) and then testing the predictive accuracy of the fitted models for the remaining patient cases (prediction set).…”

“…One of the most well known F&F models is the take the best heuristic: cues are searched in order of their importance until one discriminates, then search stops and all other cues are ignored [5]. The take the best heuristic has been operationalized in terms of the matching heuristic [6,11] that will be focused in the present study.…”

“…the cues). Instead, they base their decision on one cue, hence the name frugal [6][7][8][9][10]. Marewski and Gigerenzer discuss how F&F models could be used in the medical domain.…”