Psychometric Properties

Appraisal of human characteristics—such as achievement, ability, proficiency, attitude, belief, or another construct—is routinely accomplished through administration of a test, which is itself often carefully developed and administered by standardized protocols. Examinees and other test users are usually interested only in the results yielded by a test administration; generally they are not attuned to characteristics or technical features of the instrument itself. Still, many persons who use test results realize that the usefulness and appropriateness of test-score interpretation is a direct result of the test's internal characteristics. The internal attributes of a test are technically termed its psychometric properties. Psychometric properties are characteristics of tests and other measures of human characteristics that identify and describe attributes of an instrument, such as its reliability or appropriateness for use in a particular circumstance.

Most commonly, psychometric properties provide information about a test's appropriateness, meaning-fulness, and usefulness—in other words, its validity. As illustration, suppose a test is advertised as a measure useful for diagnosing a personality disorder such as schizophrenia. The test's psychometric properties provide test makers and users with evidence of whether the instrument performs as portrayed.

Although numerous psychometric properties exist for describing the technical qualities of tests, they are not catalogued into a convenient, definitive list. Instead, whenever a particular feature of a test is described in terms of scientific standards (i.e., not just mentioned in casual conversation), it may be considered as a psychometric property. In its most general usage, therefore, psychometric property as a scientific term refers to some essential attribute of the test.

Psychometric properties focus on particular features of a test. Some properties provide evidence as to the quality of the whole instrument, while others provide evidence about its constituent parts, its sections, or even its individual items. For example, when looking at a whole test, a psychometric property could indicate whether the instrument measures a single construct or many. The attribute of a test measuring only a single dimension—or more than one dimension if that is the case—is a psychometric property of the complete instrument. Another psychometric property of the whole test could indicate whether the instrument appraises the target construct equally well for women and men. This is a psychometric property of sex (or gender) equality. Still other psychometric properties provide evidence of whether a test measures a construct consistently (i.e., reliability). These examples illustrate some common psychometric properties of an entire test: its dimensionality, its equality, and its reliability.

Psychometric properties also exist for individual test questions (i.e., items). Indicators of a particular test item's difficulty, its ability to discriminate among people having differing amounts of the construct being measured (i.e., discrimination), and the likelihood that the correct answer could be chosen by guessing are each a psychometric property of an item.

Another feature for psychometric properties is that they are generally—but not exclusively—expressed quantitatively. Often an index, a coefficient, or some other numerical quantity is given to convey the property. Many students and professionals are familiar with a reliability coefficient, for example. The reliability coefficient is a numerical value. “Reliability” is the psychometric feature for the test, but it is expressed as a quantitative value. So, too, most other psychometric properties are indicated numerically. However, a quantitative value is not always the best means of conveying a particular psychometric property. Validity, for instance, is a complex phenomenon that cannot be meaningfully reduced to a single index or value. Validity is the overarching psychometric property, but a thorough discussion that summarizes an extensive body of evidence is necessary to describe test validity.

...