Thurstone Scaling

Background

A large part of social psychology focuses on the study of attitudes. Attitude research covers a wide variety of topics, such as racism, sexism, attitudes about school, religion, politics, among others. Unlike weight, height, or age, attitudes are not easily quantified, complicating the social research questions at hand. Moreover, there is a great deal of confusion and ambiguity attached to the concept of attitude. In fact, few social psychologists even agree on the definition of attitude. In his early work, in 1928, Louis Thurstone defines attitude as “the sum total of a man’s inclination and feelings, prejudices and bias, preconceived notions, ideas, fears, threats, and convictions about any specified topic” (p. 531).

Attitudes are simply one of many hypothetical constructs used in the social sciences. They are unobservable and therefore cannot be measured [Page 1699]directly. If attitudes could be observed directly, then studying the relationships between attitudes and other variables of interest would be straightforward. Although attitudes cannot be observed directly, hundreds of methods for quantifying attitudes have been suggested in the psychological measurement literature.

One of the earliest methods for the measurement of attitudes was suggested by Thurstone in his 1928 paper, “Attitudes Can Be Measured.” As in Thurstone’s originally proposed methods, most attitude measurement methods assume that the range of attitudes falls on a single bipolar continuum or scale. This continuum can represent attitudes such as the liberal, conservative political scale, levels of prejudice, or how important a student thinks studying is, and countless others.

Assuming the attitude of interest lies on a unidimensional scale, the goal of attitude scaling is to estimate the position of individuals on the continuum. Scaling starts by constructing or selecting a set of items or stimuli that represent varying levels of the attitude. Items can be physical objects, but are usually a set of opinion statements, the verbal expressions of attitude. “All college students should be required to take a statistics course” is an example of an opinion statement that might be used to study attitudes about the importance of studying statistics. Once a set of items that the researcher believes and measures the attitude, or attribute of interest, is constructed, the locations of the stimuli and individuals’ attitudes along the unidimensional continuum are estimated. Thurstone’s LCJ is one of the earliest methods to accomplish this goal.

...