Multiple-Indicator Measures

MEASUREMENT of theoretical CONSTRUCTS is one of the most important steps in social research. Relatingthe abstract concepts described in a THEORY to empirical INDICATORS of those CONCEPTS is crucial to an unambiguous understanding of the phenomena under study. Indeed, the linkage of theoretical constructs to their empirical indicators is as important in social inquiry as the positing of the relationships between the theoretical constructs themselves.

Multiple-indicator measures refer to situations in which more than one indicator or item is used to represent a theoretical construct in contrast to a single indicator. There are several reasons why a multipleitem measure is preferable to a single item. First, many theoretical constructs are so broad and complex that they cannot be adequately represented in a single item. As a simple example, no one would argue that an individual true-false question on an American government examination is an adequate measure of the degree of knowledge of American government possessed by a student. However, if several questions concerning the subject are asked, we would get a more comprehensive assessment of the student's knowledge of the subject.

A second reason to use multiple-item measures is accuracy. Single items lack precision because they may not distinguish subtle distinctions of an attribute. In fact, if the item is dichotomous, it will only recognize two levels of the attribute.

A third reason for preferring multiple-item measures is their greater RELIABILITY. Reliability focuses on the consistency of a measure. Do repeated tests with the same instrument yield the same results? Do comparable but different tests yield the same results? Generally, multiple-indicator measures contain less RANDOM ERROR and are thus more reliable than single item measures because random error cancels itself out across multiple measurements. For all of these reasons, multiple indicator measures usually represent theoretical constructs better than single indicators.