OBJECTIVE: A challenge in DCEs is to capture how individuals process choices and attributes. Chorus, Rose & Hensher (2013) put forth the framework and empirical proof that some attributes may be processed using one type of decision rule (e.g., random utility maximization-RUM), while others using another rule (e.g., random regret minimization-RRM) using the hybrid RUM-RRM modeling approach. This study explores if heterogeneity exists in the way community representatives process choice attributes while assessing preferences for hemophilia therapies. METHODS: Representative members of the US general population from the RAND American Life Survey panel completed a discrete-choice survey in two waves (N = 227; N= 344). The survey presented a series of 5 trade-off questions, each including a pair of hypothetical treatment profiles and an opt out option. The treatment profiles were described using five attributes: 1. costs, 2. dose adjustment, 3. side-effects, 4. efficacy and dosing frequency  and 5. type of dosage. Preferences were analyzed using RUM, RRM and hybrid RUM-RRM modeling strategies. Models were compared using the Ben-Akiva and Swait test for comparison of non-nested models, out of sample predictive ability and willingness-to-pay (WTP) estimates.  RESULTS: The hybrid RRM-RUM models, containing both regret-based and utility-based attribute decision rules, outperform choice models where all attributes are assumed to be processed by means of one and the same decision rule i.e. either utility maximization or regret minimization rule; in terms of model fit (p <0.01 Ben-Akiva and Swait test) and out-of-sample predictive ability. The hybrid WTP measures also differ substantially from conventional utility-based WTP measures (p<0.05). CONCLUSION: The community sample processes the ‘cost’ of treatment attribute in order to ‘minimize regret’ while other treatment attributes are processed differently in order to ‘maximize utility’.