This paper is concerned with how to indicate Socioeconomic status (SES) among elderly people with census or survey data from less developed societies. It reports findings from a study of independent living among elderly people 65 years and over in Brazil in 1980 and 1995 that uncommonly had comparable data on education, income and housing attributes. It found that the potential indicators of SES are not substitutable. Rather, although they tend to be positively related to each other, each can have a different relationship with independent living, one being positively related even as another has no relation and a third has a negative relation. I recommend that the indicators not be combined into one measure, and that if only one is used, that choice not be called 'Socioeconomic status' but rather precisely what it is (e.g. education). [One}… stance involves a measurement-by-fiat approach through which the investigator selects whatever remotely connected indicators he or she can locate and then merely announces that these will serve as measures [or proxies] of some highly abstract theoretical construct. The presumed rationale is that since the concept is difficult to measure, almost any indicators will do. (Blalock, 1982, p. 19). However, insofar as different perspectives lead to conflicting conclusions concerning the 'same' issue, the sensitivity of findings to (…) measurement (…) is clearly a matter of concern. (World Bank, n.fd.)