Objective Measurement of Subjective Phenomena
5. Items, Levels of Measurement, and Methods of Scale Construction
Comparisons of methods of scale construction
Sometimes one is attempting to assess a complex, multifaceted conception, as occurs in diagnostic settings that requires documentation of presence of several symptoms. This research approach would favor use of Method 1 (Empirical keying).
If one has a commitment to empirical validation and for the inductive derivation of theory from data, then Method 2 (Factor analytic) might be optimal.
If the research aims concentration on development of scales to test a priori theories, then Method 3 (Rational) is probably the method to choose.
Importantly, in empirical comparisons, all three methods arrive at scales that have comparable levels of reliability and validity (topics that will be discussed soon). Thus, if strong psychometric properties for a scale are the goal of a project, all three methods of scale construction can be recommended.
See Hase and Goldberg (1967) for an early extensive test of alternative methods of scale construction, and Burisch (1984) for a review and synthesis of research on this topic.