Validity/Validation

The degree to which one measures what one sets out to measure in a given sample or, alternatively, the accuracy of the interpretations and decisions that are made based on test or measurement instrument scores. Several kinds of validity-related evidence can be used to support validity claims, including criterion-, construct-, and content-related evidence.

Criterion-related validity examines the degree to which a measure predicts a known construct or criterion of interest. For example, a standardized achievement test given to schoolchildren would have criterion-related validity if it adequately predicted academic performance, such as grades in school. Criterion validity includes two subtypes, predictive and concurrent validity. Predictive validity refers to the degree to which scores on one measure predict scores on a future criterion measure. When examining predictive validity, time must elapse between administration of the predictive and criterion measures. Concurrent validity refers to the degree to which a measure correlates with a measure of a related construct administered at the same point in time. For example, concurrent validity could be examined by looking at the correlation between standardized achievement test scores and current school GPA (grade point average).

Construct validity is the overarching type of measurement validity; it is the extent to which a higher-order construct is represented by a particular measure. Evidence of construct validity is gained by examining the relation ships between a measure of the construct of interest and other measures designed to measure similar and different constructs. Specifically, convergent evidence and discriminant evidence are used to establish construct validity. Convergent evidence is obtained by investigating the relationship between the measure of interest and separate measures of the same construct. Discriminant evidence is obtained by investigating the relationship between the measure of interest and measures of theoretically different constructs. Construct validity exists when both convergent evidence and divergent evidence have been established.

Content-related validity evaluates the degree to which a measure examines a construct in its entirety. Content validity looks at all aspects of the research idea being studied. For example, a measure of child aggression would have content validity if it measured all compo nents of aggression, such as both physical and verbal aggression.