Multivariate analysis of user perceptions about the serviceability of urban roads: case of Barranquilla

Abstract

ABSTRACT A user's perception of road surface conditions is usually measured by a subjective 1-to-5 scale defined as the Present Serviceability Rating (PSR). The subjectivity associated with the scale can result in different PSR values for the same pavement section. In this paper, we use multivariate data analysis to identify potential groups of users sharing similar perceptions on the condition of urban roads. We conducted a survey of 137 drivers of different modalities (bus, car, taxi, truck) after they had driven over some selected road sections in Barranquilla, Colombia. The survey included socioeconomic questions and a rating questionnaire containing a list of statements related to the ride quality and the acceptance of possible pavement defects. The study included 115 urban road sections with different pavement surfaces and geometric characteristics (e.g. slopes, lane width). We used factor analysis to identify two distinct user latent preferences while driving over urban roads. Then, we applied cluster analysis on the latent preferences. Results suggest a possible classification for surveyed participants into (a) those wanting more infrastructure investment to guarantee an excellent ride quality, and (b) those who are more tolerant toward pavement surfaces in poor condition. We found significant differences in the mean rate of acceptance of urban roads within groups, suggesting that more experienced drivers tend to be more sensitive when riding over pavements in poor condition. Ignoring these differences in the preference of raters could lead to biased results when evaluating the level of pavement serviceability in urban contexts.