Perception score dataset of 7'002 places in Utrecht, Netherlands

This perception dataset consists of responses from 10,042 individuals evaluating 7,002 places in the urban-rural region of Utrecht, based on street-level photographs. It includes eight perception indicators: urban, rural, safe, natural, beautiful, residential, recreational, and social. Perceptions were measured through pairwise comparisons, where respondents selected which of two randomly shown photos was more urban, beautiful, safe, etc. Individual scores for each photograph were derived using maximum likelihood estimation of the Bradley-Terry model and linearly rescaled between 0 and 10 for comparability. The dataset was collected via an online survey conducted in the Netherlands through a panel service company, following ICC/ESOMAR guidelines and implementing quality checks to mitigate non-response and bot participation. Quota sampling along gender, age, education, and region ensured representativeness of the Dutch population. While the final sample slightly overrepresents female respondents, older individuals, two-person households, and those with continuing education, regional distribution aligns well with national statistics. The study aimed to enhance the spatial extent, resolution, and integration of perception data with existing subjective attributes such as accessibility, building stock, and social indicators. For more in-depth information on the dataset and its analysis consult "Measuring place rather than space: integrating landscape qualities for the urban-rural gradient using people’s perceptions and machine learning" (linked in the "Related Publications" section).