COMPARATIVE-HISTORICAL SOCIOLOGYothers haunt with more force interpretive com-
parisons (e.g., nonrepresentative cases; analytical
looseness and non replicability; and excessive par-
ticularity at the expense of generalization and
theoretical development).
Granting their very real import, these difficul-
ties are not normally intractable. As we have shown,
methodologically self-conscious comparative-his-
torical scholars, both qualitative and quantitative,
have devised strategies to: construct and select the
most relevant cases for analysis; exploit cross-level
data (e.g., within and between nation-states) and
make causal inferences across levels of analysis;
guard against logically false or self-validating infer-
ences; infuse the analysis of historical narrative
with greater theoretical and analytical rigor; and
increase the number and representativeness of
their cases. They have also devised or imported
useful ways to ensure maximum measurement
equivalence across cultures; sample historical rec-
ords; estimate data missing from statistical series
and to correct for truncated data; capitalize on
case interdependence with new statistical meth-
ods; historically ground time-series analysis; etc.
(see Bloch [1928] 1969; Carr 1961; Przeworski and
Teune 1970; Zelditch 1971; Tuma and Hannan
1984; Skocpol 1984b; Ragin 1987; Kohn 1989;
Isaac and Griffin 1989; McMichael 1990; Kiser and
Hechter 1991; Griffin 1993; Western 1994; Deane,
Beck, and Tolnay 1998).
Some problems, admittedly, have no satisfac-
tory solution, and nothing can compensate for the
utter dearth of valid information on particular
happenings or variables. This, though, is hardly
different from any other form of social research.
The litany of methodological difficulties discussed
above, in fact, afflict all social research. Survey
analysts, for example, deal with questionnaire re-
sponses of dubious accuracy and unknown mean-
ing taken from geographically and temporally lim-
ited samples, and scholars working with institutional
records of any sort must deal with information
collected by institutionally self-interested actors.
It is important to recognize that although
comparative-historical inquiry is no more ‘‘error-
prone’’ or unreliable than are other forms of social
research, it often adopts a very different underly-
ing ‘‘working philosophy’’ from most theory-driv-
en quantitative analysis. Usually in highly deduc-
tive and statistical analysis (including comparative
strategies), theory construction and theory testing
are two clearly defined and quite separate phases
of a research program (Kiser and Hechter 1991;
Stryker 1996). Segmenting research practice in
this way is thought necessary to ensure that hy-
pothesis testing produces a fair empirical test,
unbiased by prior analysis, or even intimate fore-
knowledge of the data. Given these rules and
objectives, then, concepts, indicators, hypotheses,
and theory are not usually changed in the middle
of analysis; that would invalidate the statistical or
logical test.In contrast, comparative-historical inquiry that
is qualitative and inductive, though often also
analytically formal, establishes a continual dia-
logue between social theory and empirical evi-
dence, so that conceptual abstraction and com-
parative history are mutually defined and constructed
in reference to each other. In The Protestant Ethic
and the Spirit of Capitalism, for example, Weber
([1904] 1958) used historical evidence on econom-
ic and religious practices and beliefs to refine his
ideal types and his hypotheses and, simultaneous-
ly, conceptually frame that empirical data to ren-
der it intelligible and explicable. Many compara-
tive-historical sociologists follow Weber’s general
strategy, that is, they start their inquiry with one set
of concepts, indicators, or hypotheses, or all three,
and end with another, modified set because they
mutually adjust data and theory, ensuring a good
‘‘fit’’ between the two as analysis progresses. In
their study of capitalist economic development
and democratic political practice in many nations,
for example, Rueschemeyer, John Stephens, and
Evelyne Huber Stephens (1992) amended their
theoretical definition of democracy, both broad-
ening its historical scope and specifying more
precisely its meaning, as they moved in their analy-
sis from Europe to Latin America. Thus what may
look like scientific ‘‘cheating’’ to a deductive or
quantitative scholar is an appropriate, fruitful strate-
gy for comparative-historical sociologists whose
goals are concept formulation and reformulation,
theory construction, and empirical understand-
ing, rather than strict hypothesis testing.
Analytic rigor does not rest with the mechanis-
tic application of technique per se, but, instead,
with a systematic and methodologically self-disci-
plined stance toward both data collection and
analysis and the use of evidence to advance infer-
ences. Most comparative-historical scholars adhere