Visual Preference Survey

Images are curated to represent the specific design choices at stake — street cross-sections, building types, densities, public-space treatments — and to span a realistic range from poor to exemplary. They should isolate the variable of interest while controlling confounds such as weather, time of day, and photographic quality, so respondents react to the design rather than the picture. Because the choice of images strongly shapes the result, the selection should be transparent and balanced rather than steered toward a preferred answer.

The mean rating ranks how much people like an image, but the dispersion around that mean measures consensus. A low variance means the community broadly agrees about an image; a high variance means opinion is split, and the average is hiding a real conflict. Reading mean and spread together — and breaking ratings down by subgroup — lets planners distinguish a genuinely shared vision from an image that merely averages to neutral because the community is polarized.

Not necessarily. A VPS captures stated visual preference under decontextualized conditions, stripped of cost, traffic, climate, maintenance, and how the space is actually used. People may rate an image highly yet behave differently in the built reality. That is why visual preference surveys are best paired with performance-based methods such as behavioral mapping or placemaking evaluation, which test whether the preferred forms actually work in use.