In most visual preference surveys, citizens are shown a sample of scenes and asked to rate them on a preference scale. Scenes are then classified by type, and for each scene type, statistics are computed. In the end, results may suggest that one scene type is preferred to another, but that is about all that can be said. In this article, we offer an alternative: a visual assessment study. In our example, we find what qualities distinguish main streets from other highways. Main street stakeholders were shown photos and video clips of state highways and asked to score them on a “main street” scale. We then estimated a cross-classified random effects modelusing main street scores as the dependent variable, and characteristics of scenes and viewers as independent variables. This class of models is new to the planning field and is preferred when random effects are present and an outcome varies systematically in two dimensions, as do ratings of different scenes by different viewers. The model we estimated can now be used to qualify certain highways for special treatment as main streets or to redesign certain highways to be more main street-like.