Triangulation

Telling Case 1: Multimethod and Multimeasure Research

[Page 1718]In 1959, Donald T. Campbell and Donald W. Fiske introduced the concept of triangulation as critical for validating variables defined as constitutive of psychological traits of individuals through methods including pencil-and-paper tests, observations, and/or performance measures. They argued that researchers could confirm and/or disconfirm assumptions about the reality or validity of the phenomena being assessed by using a multitrait or multimethod approach that they called triangulation. Since that time, this argument has been expanded to include multimethod and multimeasure approaches to assessing or measuring educational traits or phenomena. Triangulation is undertaken to ensure that the result of the study is not dependent on characteristics of a single measure or of a measurement method.

Triangulation of constructs and/or traits is undertaken by constructing a statistical matrix consisting of a table of correlations in which the relationship within and across variables or constructs by methods is examined. This table provides a basis for facilitating and/or assessing the interpretation of convergent and discriminant validity of actions, which are assumed to reflect the traits or phenomena being assessed or measured. This process focuses on construct validity, by confirming the degree to which two measures of constructs that theoretically should be related are in fact related (convergent validity). Discriminant validity provides a basis for confirming that a particular test of a concept is not highly correlated with other tests designed to measure theoretically different concepts; that is, the two measures are unrelated.

In 2012, Robert Coe provided a summary of multiple forms of triangulation that are used to assess a broad range of quantitative forms of validity, including internal validity (causal relationship definitions), external validity (population and ecological), construct validity (causal), measurement forms of validity (e.g., face, content, criterion related, predictive, concurrent, and systemic), and construct validity (measurement—convergent, divergent, and factorial). In this program of research, triangulation is undertaken to validate constructs assessed by measurement instruments as well as the reliability of particular measurements and to construct warrants to confirm or disconfirm the validity of particular measures, instruments, evaluation processes, or relationships among variables. The logic in use used by the researcher to construct the claims based on this triangulation process provides a level of transparency for assessing the warrants of the interpretations and the conclusions drawn.

...