ATS American Thoracic Society
Recent Abstracts

Measurement Properties

Measurement involves using rules to assign numbers to characteristics of objects or entities1. An understanding of basic concepts related to the process of measurement improves the critical reading of clinical and research papers as well as the selection and use of particular measurement instruments and interpretation of scores. In the following discussion of measurement properties, the terms "instruments" and "scores" are used generically. "Instruments" refers to questionnaires, tests, rating scales, observation procedures, and other devices or techniques used to measure or assess specific characteristics or attributes of people, objects, or events. "Scores" refers to the numbers resulting from the use/application of measurement instruments; these scores may represent continuous or categorical attributes.

Measurements may be made directly or indirectly. Physical, biological, and chemical characteristics, in general, have physical properties that can be measured directly. Standards for measurement of physical properties are set by the U.S. National Bureau of Standards, and relevant measurement properties include accuracy, precision, sensitivity, and error of measurement. In contrast, many social and psychologic characteristics, including many human behaviors, are more abstract constructs with no physical referent. These constructs can be measured only indirectly. Primarily because they are more abstract, such characteristics tend to be more difficult to define. As a result, multiple conceptual definitions arise. Instruments purporting to measure the same constructs (i.e., constructs with the same name) may differ as a result of differences in how the construct is conceived or in how conceptually similar constructs are defined, in practice, by virtue of instrument content. For instance, many health-related quality of life (HRQOL) instruments include some measure of symptoms. At a broad level, they share a conception of health-related quality of life as being directly affected by the frequency and severity of symptoms; at a more specific level, their definition of the relevant symptoms may differ (one, for example, may include items assessing fatigue while another does not).

Measurement standards for educational and psychological testing are set forth by the Committee on the Standards of the American Psychological Association, American Educational Research Association, and the National Council on Measurement in Education2. Relevant measurement properties addressed in the Standards include the measurement framework, the reliability of the measurement, and how one establishes the validity of particular interpretations of test scores. Thoughtful consideration of these properties in relation to instrument(s) used in or under consideration for a particular application facilitates better decision-making regarding their use, whether these decisions be in the critical review of research or the selection of instruments for use in clinical or research settings.

Misconceptions often arise in the course of interpreting scores and using measurement instruments. These misconceptions may be associated with the underlying framework or purpose for which the instrument was designed or with measurement properties such as reliability and validity. The first misconception is the assumption that the interpretation of scores on an instrument necessarily remains the same when the instrument is used for purposes other than those for which it was developed. Even though evidence regarding an instrument's score distributions, reliability, and the validity of particular interpretations of its scores may have been reported by its developer, these properties are not fixed and cannot be assumed to be relevant to all subsequent measurement situations, regardless of differences in the purposes, populations, settings for which the instrument is adopted. Another mistaken assumption is that certain methods of assessing measurement properties (e.g. test-retest reliability) are appropriate for all measurement situations.

The following discussion briefly reviews properties relevant to the measurement of health related quality of life. A more complete treatment of measurement reliability, as well as validity and other relevant topics, can be found in the APA-AERA-NCME Standards for Educational and Psychological Testing 2 and in other measurement texts (e.g., Thorndike et al., 1991 1).

Copyright © 2007 American Thoracic Society · Web Site Requirements
Questions or comments? Contact Us.