Triangulation

The Origins of Triangulation

In their classic article on determining the construct validity of various measures of psychological traits, Donald Campbell and Donald Fiske wanted to [Page 752]provide a practical way to assist in determining whether a set of measures were, in fact, measuring actual formulated constructs (psychological traits). To accomplish this, these quantitative researchers conceptualized two new forms of validity—convergent and discriminant—as subcategories of construct validity. By employing different tools to measure different psychological traits and then employing correlational analyses, the independence of the methods and traits could be established. In effect, this enabled the researchers to determine if the variance obtained across the measurements was reflective of the actual trait(s) measured or of the methods employed to measure the trait. Campbell and Fiske suggested the application of a multitrait-multimethod matrix, wherein several different tests or measures of targeted psychological traits would be administered within the same time period. If the tests designed to measure the same construct (trait) correlated more highly among themselves than with those designed to measure a different construct, then convergent validity was established. If the construct being measured by a test did not correlate highly with different constructs—even if the measuring tools across the constructs employed the same methods of measurement (e.g., both employed multiple choice questions)—then discriminant or divergent validity was established. If more than one trait and more than one method was used so that both convergent and discriminant validity could be determined through correlational methods and the relative contributions of the trait or method-specific variance could be determined, then there could be a strong and defensible inference that the construct (trait) under consideration and the tools used to measure it possessed construct validity.

This article had profound implications for methods in the social sciences, and it was the first effective application of employing multiple methods to search for convergence in the same data set. Several years later, Campbell and colleagues coined the term triangulation for this strategy of trying to use multiple methods to look at the same data set so as to create valid results and interpretation of their results. The term triangulation is a metaphorical description for the research strategy taken from navigational or military practices wherein multiple reference points and the principles of geometry were used to locate an object's exact location. That is, as with navigators, researchers could improve the accuracy of their judgments and findings by collecting different kinds of data bearing on the same phenomenon.

...