Abstract
Past literature investigating the determinants of tourist expenditure have made a wide use of econometric models to assess conditional relationships of a set of regressors in predicting individual spending. However such techniques provide little or no information about the relationships within the total set of regressors as determinants of such expenditure. This paper applies graphical models to investigate the links that occur within a set of variables derived from an official survey of Uruguayan statistics of visitor expenditures. Symmetric conditional dependence structures within socio-demographic and trip-related variables are first investigated. Then a chain graph assessment of asymmetric conditional dependencies of these two categories of variables is used to explain individual expenditure items. The output is displayed through graphs that allow for an easy interpretation of the data. The results highlight the marginal role of socio-demographic variables and the direct importance of accommodation type and destination as determinants of tourist expenditure.