Measurement

Measurement may be defined as the process that connects theoretical concepts with empirical indicator(s) designed to represent those concepts. As such, it is vitally important to social science research. This entry presents the most important properties of measurement, validity, and reliability, in their various forms. In addition, measurement levels, scales, indices, and related statistical techniques are briefly discussed.

It has been argued that inadequate measurement, more than mistaken concepts or faulty hypotheses, has hindered progress in the social sciences. The sources of inadequate measurement are complex but may be rooted in an oft-cited definition of measurement. In 1951, Stanley Smith Stevens characterized measurement as assigning numbers to objects or events according to particular rules. This is in itself an inadequate conceptualization of the process. It presents measurement as an empirical, almost mechanistic process that overlooks the important role of theory in social science research. Until researchers have worked through their measurement problems, they may not really have a theory capable of generating testable hypotheses. The difficult process of measuring theoretical concepts can make theories clearer, richer, and more subject to empirical investigation. A fuller and more appropriate definition of measurement, then, is the representation of abstract, theoretical concepts with concrete, empirical indicators. This is accomplished through the process of construct building by linking abstract, theoretical, and unobservable concepts with empirical indicators for which researchers have direct observations. This definition involves both theoretical and empirical considerations. Empirically, the focus is on the observable response—answers on a questionnaire, observed behavior in an experiment, material from an archive. Theoretically, the interest is in the underlying unobservable (and not directly measurable) concept that is used in the explanation of some social phenomenon and is represented by the response.

When the relationship between the theoretical concept and the observable response is strong, analysis of empirical indicators can lead to useful inferences about the relationships among the underlying concepts and a greater understanding of the phenomenon under investigation. When the relationship between concept and indicator is weak or faulty, analysis of the indicators leads to incorrect inferences and misleading conclusions concerning the underlying concepts. From this perspective, the auxiliary theory specifying the relationship between concepts and indicators is as important to social research as the substantive theory linking concepts to one another.

Measurement issues arise in many contexts within political science. For example, when respondents are asked in public opinion surveys which party or candidate they voted for in the previous election, questions of measurement are immediately apparent. Do some of the respondents who voted for the losing party or candidate systematically mis-remember and claim that they voted for the winner? If so, then, as we will see below, this evidence [Page 1513]would raise serious questions about the validity of recall questions that purport to measure previous voting behavior. Other measurement issues arise with instruments such as that used in the Polity Project, which codes the authority characteristics of states in the world system for purposes of comparative, quantitative analysis. Coders are instructed to assign yearly scores to all major, independent states in the global system, measuring features of these regimes such as constraints on executive authority, openness of political competition, and peaceful changes in government. Obviously, a crucial concern in this project is the degree to which different coders assign identical or widely different scores when evaluating these characteristics of these countries, an issue of reliability as we will also see below. These examples illustrate the ubiquitous nature of measurement in the social sciences generally and political science in particular.

...