Scaling Problems in the Cognition of Urban Distances

Abstract

Sources of measurement error in ratio estimates of physical driving distances are identified. In the ratio estimation procedure, the subject estimates each distance in relation to a standard line distance. When the positions of the standard line distances are in the middle range of the variable driving-distance stimuli, the functional relationship between cognitive and physical distance is biased by measurement error. Also, the estimates of distance are not independent of the physical magnitudes of the standard line distances. In the light of the results, distance stimuli of approximately equal physical magnitude and in the lower range of the set of distance stimuli should act as standard line distances in future cognitive distance experiments using ratio estimation. COMMON to most models of spatial behaviour is the view that the spatial separation of places has an influence upon behaviour and such models include a distance variable measuring this separation. But spatial behaviour results from the perceived and cognitive values of stimuli which are produced by the interaction of the cognitive processes of an individual with the environment (Briggs, 1973). Thus, movements of individuals should be understood in terms of a cognitive measure of distance rather than a physical distance measure, since movement depends on cognition. If researchers attempt to predict behaviour directly from a stimulus such as physical distance, then they may be invoking a false assumption about cognition of distance. To look for a function which transforms objective reality to the subjective space, we must devise methods for eliciting estimates of distances from the individual's cognitive map. The aim of the present study is to evaluate a method of measurement, known as ratio estimation, which has been used frequently to elicit measures of cognitive space (e.g. Briggs, 1973; Howard, Chase and Rothman, 1973; Lowrey, 1970). Until recently, ratio estimation has been accepted without question as providing accurate measures of cognitive distances, even though, as with all measurement procedures, it is possible that measurement error exists in the data collected in this way (Nunnally, 1967). A more serious possibility is that substantive conclusions are reflections of the measurement error in the data rather than of the mental constructs which are evoked to explain them. Cognitive distance research has been conducted at several different spatial levels and the function relating cognitive to physical distance may be different for each level (Cadwallader, 1973; Canter and Tagg, 1975). But the apparent difference between functions calibrated at different spatial levels may be caused by varying amounts of measurement error in the distance data. Therefore, possible sources of measurement error need to be identified in cognitive distance data elicited by ratio estimation. This measurement error problem will be addressed with the data of one experiment (Phipps, 1973), but the results will have general application to other distance experiments employing the ratio estimation procedure. Experimental design Ratio estimation is one of a number of scaling procedures developed by Stevens (1956) to elicit