Skip to main content icon/video/no-internet

Construct validity is a logical inference from which some measure or test generates results that sufficiently corroborate theoretical conceptualization or rationale. Often, the process of construct validity centers on the development of a new measure or test. To validate the new measure, researchers generate empirical evidence to support argumentation that the measure is indeed accurately observing what is claimed to be measured. Often in communication studies, constructs typically define various motivation- or attitudinal-based communication such as emotions, cultural values, and perceptions of identity, or performance-based communication such as skills, knowledge, and communicative outcomes such as the persuasive skills that speakers use to influence an audience to act. Consequently, to establish construct validity, researchers assess the accuracy of new measurement through determination of the similarities and/or differences shared among a variety of associated events.

The aim of this entry is to provide an understanding of construct validity by describing the evidence generated from observing the relationship between the new concept and independent, but related communication activities. The aim also includes an explanation of the levels of construct validity in respect to convergent and discriminant validity. Finally, practical examples are included to provide clarity on appropriate application of tests, followed by a variety of common errors in establishing construct validity.

Assessment

As a branch of the social sciences, communication research findings almost always involve social interaction and therefore are dependent on the context within which communication takes place. As new contexts are defined, new measures are in constant demand to advance theoretical assumptions or conclusions about communication activities. The new measures must be tested, or validated, for accuracy. A common practice of construct validation aims to determine similarities and differences between the new construct and various related measures. At times, researchers are concerned with the similarities a new measure shares with various other measures. Tests for similarity commonly include factor analysis, correlation analysis, analysis of variance (ANOVA), and regression analysis. At other times, researchers focus on differences more commonly determined by various t-tests and discriminant analyses. The question becomes whether or not the new measure relates as theoretically expected to the other associated measures within some context.

As an example, consider a communication scholar who wishes to establish a new theory to explain why some individuals express dishonest reasons about why they are emotionally upset in conflict situations. The scholar develops a measure to generate evidence for a construct that involves two levels. The primary level includes attributions associated with dishonest reasons for why an individual is upset, while the secondary level includes attributions associated with more honest reasons. To validate the construct, the scholar chooses a measure of conflict competitiveness. The scholar hypothesizes that the more competitive an individual becomes in conflict situations, the more that same person expresses dishonest reasons to explain his or her emotional state.

The scholar collects data using both the conflict competitiveness measure and the new measure for primary and secondary emotional expressions. Data are collected at the same time, using both measures to generate evidence for why individuals become more or less honest about why they may be upset in conflict situations. The scholar is now ready to begin determining construct validity.

...

  • 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