This paper reports three experiments. Our first experiment compares several versions of an item modeled on the ACS ethnicity item to explore whether examples may help clarify whether specific borderline instances are acceptable answers. Our next two experiments focus on whether examples may suggest features of the desired answers. One experiment compares three versions of an item on ethnicity; the other is a conceptual replication that focuses on reports about wine consumption. Wine consumption was chosen because it was consistent with the overall topic of the questionnaire, which was health and diet. In all three studies, we also examine what might be called the “focusing hypothesis,” the idea that examples bias respondents to report the examples they were presented with.