Abstract
We develop a structural equation model (inspired from an exploratory directed acyclic graph) from a dataset consisting of a cross section of 6,637 UK households in order to uncover the links among expenditures on various forms of gambling, income, and other household expenditures. We find evidence for the existence of latent variables that we refer to as hard gambling, soft gambling, and myopic behavior (myopia). Our central finding is that greater myopia increases hard gambling, but not soft gambling. We also find that a greater income has a strong and negative effect on myopia and hard gambling, but not on soft gambling. Greater spending on necessities, such as food and clothing, is associated with increased myopia, soft gambling, and hard gambling. Finally our results reveal a moderate but significant positive association between soft gambling and hard gambling, providing some support for the "slippery slope" hypothesis. We discuss the results and new questions they suggest.