One of the major problems of regional science is the lack of availability of reliable quantitative data. Modern spatial data analysis is often hampered by this. Many variables and attributes are only measured or measurable on an ordinal or qualitative scale. This gives rise to a new line of thinking, in which the existence of this data problem is accounted for, such as emphasizing and developing methods and techniques that are able to use qualitative information in a theoretically consistent way. The purpose of this paper is to contribute to this new research paradigm by elaborating a new way of interaction modelling, in which fuzzy spatial characteristics are explicitly taken into account. The number of interactions, for instance, will vary from day to day and from hour to hour. It appears, therefore, much better and even easier to measure this variable on a less demanding - and for this reason more precise - ordinal scale. It can also be argued that the operational ability of many spatial interaction models is handicapped by the qualitative nature of many spatial characteristics (e.g. the "attractiveness" of zones). In this paper a calibration method is outlined, in which both the interactions and the attraction variables are interpreted as ordinal data. Some interesting consequences will be discussed.
Abstract