simply'' assigning numbers to things.'' As I noted at the outset, Dawson et al. similarly advocated the value of adhering to the classical definition of measurement, although they expressed much more optimism than Stam about the possibility of developing such'' strong'' measures.
I believe that it is not possible to develop measures that meet the criteria for'' real'' measures and that we should not aim to develop such measures. These claims follow directly from the first point above about interpretation. All research is interpretive, and this certainly includes the key research process of measurement. While it may or may not be the case that the classical notion of measurement can and should apply in some natural sciences, it does not apply in research in the human sciences. But, we can employ measurement procedures and, therefore, make use of quantification in our investigations-so long as we understand what we are doing in a novel way, which could be called a different'' theory of measurement.''
Here, the second point about interpretation comes into play. We can make use of measurement so long as we recognize that our measures are indexical, that is, interpretive in the sense that they always refer beyond themselves to our prior familiarity with practices. Such measures can be of very different kinds, which, very roughly speaking, mark out a continuum ranging from the very concrete to the obviously meaning-laden. Measures of decibel levels lie quite far to the'' concrete'' end of this continuum, the coding category'' yells'' moves away from that end, and global ratings of'' behaves in a hostile manner'' lie well to the '' Meaning-laden'' end. Note, however, that because all measures are indexical, all points along this continuum are ultimately both concrete and meaningful. They are all examples of phenomena of interest that, in varying ways, concretely specify the phenomena but at the same time reflect the fact that the concrete specifications are never exhaustive. A measure on the concrete end of this continuum based on decibel levels might be the dependent variable in an experimental paradigm within which high-decibel verbalizations are examples of angry behavior. At the other end of the continuum, global ratings will be based on a manual that uses concrete examples to define the phenomenon of interest.
Given this'' theory of measurement,'' I think that it is misleading to say-as Stam and Dawson et al. claimed-that I call for'' weak'' measures rather than'' strong'' ones. As I see it, I am offering a different framework that incorporates many measures that might well be called'' strong'' measures (those near the'' concrete'' end of the continuum), although they do not conform to the classical defin...