It is well known that the presence of response styles can affect estimates in item response models. Various approaches to account for response styles have been suggested, in particular the tendency to extreme or middle categories has been included in the modelling of item responses. A response style that has been rarely considered is the noncontingent response style, which occurs if persons have a tendency to respond randomly and nonpurposefully, which might also be a consequence of indecision. A model is proposed that extends the Rasch model and the Partial Credit Model to account for a response style that accounts for subject-specific uncertainty when responding to items. It is demonstrated that ignoring the subject-specific uncertainty may yield biased estimates of model parameters. Uncertainty as well as the underlying trait are linked to explanatory variables. The parameterization allows to identify subgroups that differ in response style and underlying trait. The modeling approach is illustrated by using data on the confidence of citizens in public institutions.