A 1964 review of tests and scales used in family research found serious deficiencies (Straus 1964), and subsequent reviews showed very little improvement (Straus 1992; Straus and Brown 1978). However, changes in the nature of the field have contributed to an increase in the use of standardized tests to measure characteristics of the family. This is an important development because standardized tests are vital tools for both clinical assessment and research. New tests tend to produce a flowering of research focused on the newly measurable concept. Examples of tests that have fostered much research include measures of marital satisfaction (Spanier 1976), adequacy of family functioning (Olson, Russell, and Sprenkle 1989), and family violence (Straus 1990a). Hundreds of family measures are abstracted or reproduced in compendiums such as Family Assessment (Grotevant and Carlson 1989), Handbook of Measurements for Marriage and Family Therapy (Fredman and Sherman 1987), and Handbook of Family Measurement Techniques (Touliatos, Perlmutter, and Straus 2001). There is also a growing methodological literature on techniques for constructing measures of family characteristics, such as those by Karen S. Wampler and Charles F. Halverson, Jr. (1993) and Thomas W. Draper and Anastascios C. Marcos (1990). The state of testing in family research, however, is not as healthy as these publications might suggest. In fact, the data indicate that the validity of tests used in family research is rarely known.
For purposes of this entry, the term measure includes test, scale (such as Likert, Thurstone, Guttman, and Semantic Differential scales), index, factor score, scoring system (when referring to methods of scoring social interaction such as Gottman 1994 or Patterson 1982), and latent variables constructed by use of a structural equation modeling program. The defining feature is that they "combine the values of several items [also called indicators, questions, observations, events] into a composite measure . . . used to predict or gauge some underlying continuum which can only be partially measured by any single item or variable" (Nie et al. 1978, p. 529).