|Abstract:||Decision making research often heavily relies on deterministic modeling approaches. However, choice data are stochastic and therefore need to be modeled probabilistically. According to one probabilistic modeling approach, a decision maker has a fixed preference, but makes errors when selecting the utility-maximizing option. In another approach, a decision maker makes no errors, but his preference itself is probabilistic. Bhatia and Loomes (2017) refer to the first approach as "response noise" and the second approach as "preference noise." To avoid incorrect conclusions of a decision maker's underlying preferences, Bhatia and Loomes (2017) strongly advocate for modeling both types of noise simultaneously. In this commentary, we discuss the methods of Bhatia and Loomes (2017) and revisit a hybrid model, which models response and preference noise simultaneously, to address some limitations of these methods. Furthermore, we illustrate the hybrid model, discuss further refinements to the model, and illustrate model fit using data from hypothetical decision makers.