Dominance occurs in a choice task when one concept is as good as or better than every other concept in the task, on every single attribute, so that no respondent should choose any of the other concepts. When clients notice dominated tasks, they might worry that asking questions with obvious answers is a waste of questionnaire real estate. Recently Bliemer, Rose and Chorus (2014) showed that dominance could also cause a serious bias: if dominated tasks lead to respondents always choosing the dominating concept, then utilities in dominated tasks might be inflated relative to those in non-dominated tasks. This would cause utility inflation overall, and more so as the proportion of dominated tasks is larger. They go on to recommend two fixes: (a) making designs with no dominant choice tasks and (b) a fancy statistical model that adjusts the utility scale for just those choice tasks that are dominated. Using two empirical studies I want to find out if dominance really is a problem, and if it is, under what conditions do we need to worry about it and how might we best fix it.
