In the present study, we aimed at investigating what factors affect the judgment of a typical reader when he or she deals with numerical information in an ecological context. Participants read a story about a man who was not treated with heparin after hernia surgery and then died. Their task was to assess the liability of the medical staff after receiving ambiguous numerical data based on percentages, and again after receiving unambiguous data based on frequencies. Participants also assessed the likelihood of survival/death for heparin-treated vs. not-treated patients. The unambiguous numerical information they were given was different in terms of numerousness of the reference class and framing. Results show that even when unambiguous frequency-based information is available, the participants' judgments were strongly affected by both frame and reference class. Findings also indicate that likelihood and liability judgments are strongly related, and that liability is accounted for by likelihood, but not vice versa.
