Preface In order to assess the reliability and especially the validity of social measures, we formulate
a strategy based on the integration of evidence from factor analysis and construct validation... Our ultimate purpose is to
convey the logic underlying this measurement strategy and, more generally, to sensitize researchers to the multiple
sources of nonrandom measurement error that probably influence data in the social sciences.
p.1 The link between observation and formulation is one of the most difficult and crucial of the scientific
enterprises. It is the process of interpreting our theory or, as some say, of "operationalizing our concepts." Our creations
in the world of possibility must be fitted in the world of probability; in Kant's epigram, "Concepts without percepts are
empty." It is also the process of relating our observations to theory; to finish the epigram, "Percepts without concepts are
blind." Scott Greer (1969: 160)"
p.1 The importance of measurement to social research is well stated in an observation by Hauser (1969:127-9):
"I should like to venture the judgment that it is inadequate measurement, more than inadequate concept or hypothesis,
that has plagued social researchers and prevented fuller explanations of the variances with which they are confounded."
p.2 measurement serves a vital theoretical purpose, as aptly described by Blalock (1970:88-9):
"Measurement considerations often enable us to clarify our theoretical thinking and to suggest new variables that should be
considered. It is often thought, prior to actual attempts at measurement, that we really understand the nature of
a phenomenon because we have experienced it directly... Careful attention to measurement may force a clarification
of one's basic concepts and theories.
p.2 The purpose of this book is to explicate an approach to measurement that focuses on the theoretical
as well as the empirical components of the measurement process. Stated more fully, our purpose is to introduce, justify, and
illustrate a systematic and integrated approach to measurement in the social sciences.
p.2 Following Blalock, we define measurement as the process of linking abstract concepts
to empirical indicants.
p.2-3 Stated somewhat differently, measurement has been defined as "the process that permits the social
scientist to move from the realm of abstract, indirectly observable concepts and theories into the world of sense experience."
(McGaw and Watson, 1976:205).
p.3 Abstract concepts do not have a one-to-one correspondence with empirical indicants; they can be operationalized
and measured in an almost infinite variety of ways... The point is that abstract concepts can only be approximated
by empirical indicants. Indeed, it is the very vagueness, complexity, and suggestiveness of concepts that allow them
to be empirically referenced with varying degrees of success at different times and places... However, because concepts can
be neither directly observed nor measured, the systematic exploration, testing and evaluation of social theory requires social
scientists to use empirical indicants, designed to represent given abstract concepts. Thus, in contrast to concepts, empirical
indicants are designed to be as specific, as exact, and as bounded as theoretical formulations and research settings
will allow. Indicants, in other words, are intended to approximate and locate concepts empirically... Any particular
set of empirical indicants that one chooses, therefore, is only a small subset of an almost infinite number of possible indicants
that could be selected to represent a particular concept.
p.4 From this discussion, it is not difficult to see that both abstract concepts and empirical
indicants are necessary if a worthwhile social science is to thrive and develop... By providing an accurate representation
of concepts, empirical indicants contribute to the theoretical development of the social sciences by highlighting those gaps
that exist between theoretical formulations and observed reality.
p.6 we will argue that the ultimate success of social science research depends strategically on
how accurately theoretical concepts are measured. Viewed from this perspective, measurement is not
an esoteric concern unrelated to the central issues of the social sciences but instead is woven into the very fabric
of social research.
p.6-7 Two key terms, reliability and validity, provide the essential language of measurement... "Reliability
concerns the extent to which measurements are repeatable" ... "In a very general sense, a measuring instrument is
valid if it does what it is intended to do." Consequently, if a set of indicants were perfectly valid, it would
represent the intended - and only the intended - concept. A less than perfectly valid measure, conversely, would
imply that it does not fully represent the concept or that it represents something other than the concept.
From this discussion, it is easy to see that validity is more important than reliability...
to have a valid measure, one must have a reliable one
p.7 Classical test theory begins with the basic formulation that an observed score, X, is equal to the true
score, T, plus a measurement error, e. To state this idea as a formula:
X = T + e
p.8 From these assumptions, it follows that the expected (mean) value of the observed score is equal to
the expected (mean) value of the true score. In formula form,
E(X) = E(T) [JLJ - this would be a good way to calibrate any evaluation method
for a machine playing a game.]
p.12 In our reformulation of classical test theory, an observed score, X, is equal to the true score, T,
plus systematic measurement error, S, plus random measurement error, R. In formula form,
X = T + S + R
p.13 Heise (1974:9) notes:
There is no purely mechanical procedure for identifying latent variables with guaranteed theoretical validity
p.13 although one cannot observe a concept, systematic error, or their covariance directly (because they
are, in principle, unobservable), one can estimate these variances and covariances if one is willing to make informed, reasonable
assumptions about the nature of the phenomena underlying a set of indicants.
p.14 validity is the degree to which a set of indicants measures the concept it is intended to measure.
p.80 Predictive validity, on the other hand, concerns a future criterion that is correlated
with the relevant measure. Tests used for selection purposes in different occupations are, by nature, concerned with
predictive validity. Thus, a test used to screen applicants for police work could be validated by correlating their test scores
with future performance in fulfilling the duties and responsibilities associated with police work.
