Factor Analysis

Types of Validity and Validation

Content validation is employed when it seems likely that test users will want to draw references from observed test scores to performances on a larger domain of tasks similar to items on the test. Typically, it involves asking expert judges to examine test items and judge the extent to which these items sample a specified performance domain. There are two types of content validity: face validity and logical validity. A test has face validity if an examination of the items leads to the conclusion that the items are measuring what they are supposed to be measuring. Logical or sampling validity is based on a careful comparison of the items to the definition of the domain being measured.

Criterion-related validation is a study of the relationship between test scores and a practical performance criterion that is measurable. The criterion is the thing of interest or the outcome researchers are concerned about. When a test score, X, can be related to a criterion score, Y, criterion-related validity can be determined. The validity coefficient, ΔXY can be based on a predictive or a concurrent study. A predictive-validity coefficient is obtained by giving the test to all relevant people, waiting a period of time, collecting criterion scores, and calculating the validity coefficient. When a test is used to predict future behavior, predictive validity can be calculated.

Construct validation is appropriate whenever the test user wants to draw inferences from test scores to a behavior domain that cannot be adequately represented by a single criterion. A test's construct validity is the degree to which it measures the behavior domain or traits that it was designed to measure. More specifically, construct validity can be understood as the extent to which the behavior domain or the constructs of theoretical interest have been successfully operationalized. For example, a researcher may be interested in determining clients' satisfaction with healthcare services. Since “satisfaction with healthcare services” is a construct that cannot be adequately represented by a criterion or defined by a universe [Page 382]of content, the researcher chooses to develop a questionnaire of 20 items in order to tap the construct “satisfaction” and proceeds to collect the data. The question is how does the researcher know that what he or she is measuring through the questionnaire is actually and purely clients' satisfaction with healthcare services and not something else or a mixture with other constructs such as clients' degree of confidence in the medical profession? In this case, a construct validation is appropriate.

...