Validity of Measurement

The Strength of Validity

Perhaps one of the reasons it is difficult to understand measurement validity is the belief that science involves the discovery of truth. Truthful discoveries might be the ultimate aim of research; yet investigators examine a variety of truths and realities. Research in this information age is a process of fighting for language that will help individuals better understand one another and the universe in which they live. At best, new research can offer improvements in the accuracy with which phenomena are explained, and measurement is the step in the research process whereby investigators describe what they understand. Valid measurement arguments, in other words, can offer strong or weak descriptions of the concepts used to generate and defend new ideas.

With this in mind, it is helpful to remember that the relative strength of a validity argument is likely to differ depending on how well something is understood and on the qualities of the entities [Page 1592]being measured. Imagine two strong validity arguments that can be made for a well-understood construct. In the simplest case, just as a light switch can turn a bulb on and off, constructs might reflect a single dimension that can be assessed as present or absent, right or wrong. As occurs when electric frequencies vary, the underlying cause of such events might be complex, but indicators of these simple constructs are readily apparent and can be recorded accurately. More commonly in the social sciences, a construct might be well known but has complex properties. Strong validity arguments about such multidimensional constructs can be made accurately only if indicators of each dimension are included in the measurement instrument. Validity arguments about complex constructs include comparisons across dimensions as well as across the indicants of each dimension. Both simple and complex arguments are supported by evidence that all dimensions and indicants are recorded accurately and consistently across repeated uses of a measurement tool, and that the construct is captured fully.

Regardless of whether a construct has one or several dimensions, well-understood constructs are the easiest to measure because there are fewer opportunities for multiple interpretations of their existence. In mathematics, 2 + 2 will equal 4 in a base 10 system, and the accuracy with which individuals can answer such problems can be evaluated across repeated use of an addition test with relative ease. If the construct of addition were to be broadened to include other basic mathematical operations, more dimensions would be needed to represent fully the construct. Along with the inclusion of addition problems, for example, a measurement argument might include subtraction, multiplication, and division problems. A seemingly simple change in the definition of a construct invariably results in a dramatic revision of how validity is established.

...