Skip to main content icon/video/no-internet

Construct-Related Validity Evidence

Construct-related validity evidence demonstrates whether a test measures its intended construct, where a construct can be defined as a conceptual abstraction used to understand the unobservable latent variable that is responsible for scores on a given measure. Constructs are said to be situated within the nomological network, which was originally proposed by Lee Cronbach and Paul Meehl in 1955. Nomologic refers to rules of nature, and the nomological network situates a construct in terms of its relationship to other, known constructs and behaviors in order to provide a theoretical context for the construct. This theoretical context in turn suggests avenues through which construct-related validity evidence can be provided, for example, in terms of its relationship to other constructs or traits.

According to our current understanding of validity, construct validity is the only type of validity, and thus, construct-related validity evidence encompasses all possible types of validity evidence. Samuel Messick’s 1989 framework redefined validity as a unified concept by defining all validity as construct validity; this definition effectively subsumes all possible types of validity evidence into the larger category of construct validity evidence. However, the term construct validity evidence is also sometimes used to refer to specific sources of validity evidence; this sense of the term recalls the historical definition of construct validity as a specific type of validity. In earlier decades when validity was conceptualized as having multiple types, construct validity frequently appeared alongside content validity and criterion-related validity as one of the main types of validity, and construct validity had its own sources of validity evidence. Although this conceptualization of types of validity is not the modern view, construct-related validity evidence is still discussed in the literature. The following sections outline the sources of construct-related validity evidence, provide an overview of the historical definition of construct validity as a type of validity, and provide an overview of our current understanding of construct validity as the sole type of validity in the unified theory.

Sources of Construct-Related Validity Evidence

Construct-related validity evidence supporting an item’s nomological validity attempts to provide quantitative evidence to position the construct within the nomological network. In order to assemble nomological validity evidence, it is useful to consider both convergent and discriminant validity evidence. Convergent validity evidence rests on the assumption that constructs that are closely related in the theoretical framework of the nomological network should also be correlated when measured in reality. Convergent validity evidence can be provided in terms of a measure’s correlation with other measures with strong validity arguments that assess theoretically related constructs. For example, if the construct of intelligence is thought to be closely related to working memory, then examinees’ results on a test thought to measure intelligence should be highly correlated with their results on a measure of working memory.

Discriminant validity evidence is the counterpart of convergent validity evidence. This type of validity evidence is used to demonstrate that constructs that have no relationship or an inverse relationship in the nomological net are also not correlated in reality. Discriminant validity evidence can consist of a measure’s low or negative correlation with other measures assessing theoretically opposed concepts. For example, if extraversion and introversion are assumed to lie at opposite ends of a spectrum, then examinees’ results on a measure of extraversion should negatively correlate with their results on a measure of introversion.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading