85
an epistemic foundation, rather than an ontic foundation such as was proposed by
the realists discussed previously.
Finally, the representationalist emphasis on rules also had significant consequences for the concept of a true value. While a rule (e.g., for numerical assignment) could be found to be (for example) adequate, effective, or efficient, its
application is in and of itself unrelated to the possible truth of its outcomes. But by
the time of the introduction of RTMs, the idea that measurement is a quest for true
values had become so entrenched in measurement science that renouncing it
appeared to be unacceptable, even in a context in which the search for an ontic
grounding for the concept of a true value had been replaced by epistemic (or even
formal) conditions. The definition of given in the VIM exemplifies
this. According to the first edition (ISO, 1984: 1.18) a true value is “the value which
characterizes a quantity perfectly defined, in the conditions which exist when that
quantity is considered” (though what would count as a perfect definition of quantity
is not explicated). Almost 30 years later, the third edition of the VIM changed the
definition: a true value is a “quantity value consistent with the definition of a quantity” (JCGM, 2012: 2.11). The consistency of something with something else is a
condition that can be obtained by the appropriate application of a rule, but consistency and truth are distinct concepts, and this seems to be a definition of. Thus truth has been maintained in the lexicon
of the VIM but seems to have disappeared in the substance.
Representationalism has had relatively little direct impact on the practices of
either the physical or the human sciences. This fact can be interpreted as a sign of
the practical uselessness of such theories in situations—like most cases of physical
measurement, and many cases of nonphysical measurement as well (see, e.g., Cliff,
1992)—in which the source of complexity, and hence of interest, is actually (also)
the execution of measurement, not (only) the characterization of its preconditions.
However, as already noted, representationalism has had at least an indirect impact
on thinking about measurement through the work of Stevens, who, informed by
both operationalism and early versions of representationalism, defined measurement as “the assignment of numerals to objects according to a rule” (Stevens, 1946:
p. 667). This definition and close variants thereof are ubiquitous in introductory
textbooks on psychology and psychological statistics (for a review, see, e.g.,
Michell, 1997), when indeed a definition of measurement can be found at all (see
also Borsboom, 2009).
Stevens proposed this definition after the Ferguson Committee, which was initially convened in 1930 by the British Association for the Advancement of Science
(Ferguson et al., 1940) with the charge of studying the possibility of providing
“quantitative estimates of sensory events”, ultimately concluded that claims of measurement being made by human scientists of the day—including by Stevens himself, in the context of his work on the measurement of sensations—were at best
premature, and at worst “not merely false, but misleading” (p. 345) given the way in
which measurement was understood by the wider scientific community (for longer
histories than is possible here, see, e.g., McGrane, 2015; Michell, 1999; Rossi,
2007; see also Sect. 6.5). Stevens’ move of redefining measurement arguably
4.2 Characterizing measurement
an epistemic foundation, rather than an ontic foundation such as was proposed by
the realists discussed previously.
Finally, the representationalist emphasis on rules also had significant consequences for the concept of a true value. While a rule (e.g., for numerical assignment) could be found to be (for example) adequate, effective, or efficient, its
application is in and of itself unrelated to the possible truth of its outcomes. But by
the time of the introduction of RTMs, the idea that measurement is a quest for true
values had become so entrenched in measurement science that renouncing it
appeared to be unacceptable, even in a context in which the search for an ontic
grounding for the concept of a true value had been replaced by epistemic (or even
formal) conditions. The definition of
this. According to the first edition (ISO, 1984: 1.18) a true value is “the value which
characterizes a quantity perfectly defined, in the conditions which exist when that
quantity is considered” (though what would count as a perfect definition of quantity
is not explicated). Almost 30 years later, the third edition of the VIM changed the
definition: a true value is a “quantity value consistent with the definition of a quantity” (JCGM, 2012: 2.11). The consistency of something with something else is a
condition that can be obtained by the appropriate application of a rule, but consistency and truth are distinct concepts, and this seems to be a definition of
of the VIM but seems to have disappeared in the substance.
Representationalism has had relatively little direct impact on the practices of
either the physical or the human sciences. This fact can be interpreted as a sign of
the practical uselessness of such theories in situations—like most cases of physical
measurement, and many cases of nonphysical measurement as well (see, e.g., Cliff,
1992)—in which the source of complexity, and hence of interest, is actually (also)
the execution of measurement, not (only) the characterization of its preconditions.
However, as already noted, representationalism has had at least an indirect impact
on thinking about measurement through the work of Stevens, who, informed by
both operationalism and early versions of representationalism, defined measurement as “the assignment of numerals to objects according to a rule” (Stevens, 1946:
p. 667). This definition and close variants thereof are ubiquitous in introductory
textbooks on psychology and psychological statistics (for a review, see, e.g.,
Michell, 1997), when indeed a definition of measurement can be found at all (see
also Borsboom, 2009).
Stevens proposed this definition after the Ferguson Committee, which was initially convened in 1930 by the British Association for the Advancement of Science
(Ferguson et al., 1940) with the charge of studying the possibility of providing
“quantitative estimates of sensory events”, ultimately concluded that claims of measurement being made by human scientists of the day—including by Stevens himself, in the context of his work on the measurement of sensations—were at best
premature, and at worst “not merely false, but misleading” (p. 345) given the way in
which measurement was understood by the wider scientific community (for longer
histories than is possible here, see, e.g., McGrane, 2015; Michell, 1999; Rossi,
2007; see also Sect. 6.5). Stevens’ move of redefining measurement arguably
4.2 Characterizing measurement