ABSTRACT
This chapter discusses how to choose empirical variables. When the theoretical structure is well-defined, the theoretical variable may be confined to a single dimension. But in fields in which the researcher works less with a well-structured theory and more with free-ranging imagination, the theoretical variable is likely to be multidimensional. There are two ways to justify the measurement of a given variable as a proxy for another variable that is clearly defined but cannot itself be measured. The first is to demonstrate an empirical association between the proxy and the hard-to-measure variable. The second is to demonstrate a logical link with a chain of reasoning between the proxy and the conceptual variable. The greater the similarity between the proxy and the actual variable, the better the proxy. The branch of economics called “welfare economics” offers a particularly interesting and thorny problem of relating the proxy to the theoretical variable.
