Data without (operational) variables

Quantitative sociology has grown by borrowing methods from the experimental sciences even though, for the most part, our data are observational. Where the procedures of experimental science can be applied, data analysis can e xploit simplifying assumptions because good experimental design removes cor relations among independent variables that exist nature, outside the experi ment, as well as effects of unmeasured variables. Where the science is, of necessity, observational, these simplifications can not be guaranteed and, as a result, the analyses reached through by use of some of the standard "w orkhorse" techniques of the statistical repertoire may not be valid and con clusions reached by the application of these techniques are in doubt. This paper explores an alternative framework for data analysis in quanitati ve sociology, bypassing the statistics associated with experimental methods . Specifically, it explores generalizations of the method and quantitative theory used by physical surveyors, generalizing them to the needs of observ ational data. Application of this framework to text, including editorials and free answer s to questionnaires as well as application to (social) survey data, support s their its for these purposes and the cartographic methods suggest that mi cro theories embedded in these methods reduce the load of a priori assumpti ons "normally" required for both text analysis and survey analysis. The app lications suggest a research path applicable to "ordinary" sociological var iables, including education, income, occupation, and gender, that shifts th e burden of argument away from variance explained criteria and toward an in tegration of theory and method, guided by principles of parsimony and consi stency.