86
permitted psychological scientists to avoid addressing the criticisms of the committee, and gave them license to continue making claims of measurement—along with
at least implicit claims to the epistemic authority and social status that measurement
had earned among the general public—and had ripple effects that effectively
unmoored the development of the theory and practice of measurement in the psychological sciences from the rest of science and philosophy (McGrane, 2015).
But Stevens’ definition of measurement inherited the limitations of both operationalism and representationalism, and in effect trivialized the requirements for
claiming that one has successfully achieved measurement; to abbreviate an example
from Borsboom (2009), one could divide individuals’ shoe sizes by their postal
codes and assign to them the resulting numerals; according to Stevens’ definition,
this would be an instance of measurement, though it would be difficult to state what,
if anything, is being measured. Viewed this way, the identification of a given process
as an instance of measurement is simply unrelated to the issue of whether its results
are authoritative or worthy of trust. Thus representationalism, like operationalism,
is incapable of explaining the societal role played by measurement, or of providing
a justification for its acknowledged epistemic authority.
4.3 The concept of validity in psychosocial measurement
As has been described in the previous sections, the way in which measurement is
commonly defined in the human sciences—i.e., as rule-based numerical assignment—is, in and of itself, not usable as a basis for justifying the epistemic authority
of measurement processes and their results. Additionally, as was discussed in Chap.
1, the human sciences generally lack anything like the systems of lawful relationships (sometimes referred to as “nomic nets”, or by the VIM as “systems of quantities”; JCGM, 2012: 1.3) that serve as the basis for identifying transduction effects
that themselves serve as the basis for physical measurement processes (see also
Finkelstein, 2003, 2005), and thus also serve as justifications for claims regarding
the existence of measured properties (see also Sect. 6.6) and the validity of measurement processes involving these properties and the relevant laws. However, this does
not mean that human scientists have been unconcerned with such justifications;
rather, particularly in the contexts of educational testing and psychological research,
a separate concept (and literature) has arisen that addresses the issues of when and
why some measurement results are (or are thought to be) epistemically authoritative
in their role of providing information about a property intended to be measured: that
is, the concept of validity, which is often described as “the most fundamental consideration in developing and evaluating tests” in the human sciences (quote from the
The representationalist stereotype: Measurement is any process that maps
properties to informational entities in a consistent way, so as to represent the
former by means of the latter.
4 Philosophical perspectives on measurement
permitted psychological scientists to avoid addressing the criticisms of the committee, and gave them license to continue making claims of measurement—along with
at least implicit claims to the epistemic authority and social status that measurement
had earned among the general public—and had ripple effects that effectively
unmoored the development of the theory and practice of measurement in the psychological sciences from the rest of science and philosophy (McGrane, 2015).
But Stevens’ definition of measurement inherited the limitations of both operationalism and representationalism, and in effect trivialized the requirements for
claiming that one has successfully achieved measurement; to abbreviate an example
from Borsboom (2009), one could divide individuals’ shoe sizes by their postal
codes and assign to them the resulting numerals; according to Stevens’ definition,
this would be an instance of measurement, though it would be difficult to state what,
if anything, is being measured. Viewed this way, the identification of a given process
as an instance of measurement is simply unrelated to the issue of whether its results
are authoritative or worthy of trust. Thus representationalism, like operationalism,
is incapable of explaining the societal role played by measurement, or of providing
a justification for its acknowledged epistemic authority.
4.3 The concept of validity in psychosocial measurement
As has been described in the previous sections, the way in which measurement is
commonly defined in the human sciences—i.e., as rule-based numerical assignment—is, in and of itself, not usable as a basis for justifying the epistemic authority
of measurement processes and their results. Additionally, as was discussed in Chap.
1, the human sciences generally lack anything like the systems of lawful relationships (sometimes referred to as “nomic nets”, or by the VIM as “systems of quantities”; JCGM, 2012: 1.3) that serve as the basis for identifying transduction effects
that themselves serve as the basis for physical measurement processes (see also
Finkelstein, 2003, 2005), and thus also serve as justifications for claims regarding
the existence of measured properties (see also Sect. 6.6) and the validity of measurement processes involving these properties and the relevant laws. However, this does
not mean that human scientists have been unconcerned with such justifications;
rather, particularly in the contexts of educational testing and psychological research,
a separate concept (and literature) has arisen that addresses the issues of when and
why some measurement results are (or are thought to be) epistemically authoritative
in their role of providing information about a property intended to be measured: that
is, the concept of validity, which is often described as “the most fundamental consideration in developing and evaluating tests” in the human sciences (quote from the
The representationalist stereotype: Measurement is any process that maps
properties to informational entities in a consistent way, so as to represent the
former by means of the latter.
4 Philosophical perspectives on measurement