COMPARATIVE-HISTORICAL SOCIOLOGYStates, when contrasted to the European experi-
ence and stacked up against available explanations
for the development of class-conscious labor poli-
tics, appeared both historically and theoretically
anomalous. To explain such puzzles, the original
theory is modified and new concepts and theories
with greater explanatory power are formulated.
The improved theory then serves as the new start-
ing point for subsequent inquiry, so that cumula-
tion of knowledge is facilitated (Stryker 1996).
Thus historical patterns are comparatively situat-
ed, the ‘‘inexplicable’’ residue of time and place
therefore is potentially ‘‘explicable’’ (Sewell 1967),
and sociological theory is further developed.
ANALYTIC TYPES OF HISTORICAL
COMPARISONSContemporary comparative-historical research
practice rests upon diverse, even contradictory,
epistemologies and research strategies (Bonnell
1980, Skocpol 1984b, Tilly 1984; Ragin 1987;
McMichael 1990; Kiser and Hechter 1991; Griffin
1992; Sewell 1996; Stryker 1996; Mahoney 1999).
These can be initially grouped into two basic ap-
proaches, labeled here ‘‘analytical formalism’’ and
‘‘interpretivism.’’ Each displays considerable in-
ternal diversity, and the two can overlap and be
combined in practice. Moreover, they can be syn-
thesized in creative ways, thereby capitalizing on
the strengths of each approach and reducing their
respective weaknesses. The synthesis, ‘‘causal
interpretivism,’’ is sufficiently distinct as to consti-
tute a third approach to the analysis of compara-
tive history.
Analytically formal comparison. Formal com-
parison conforms generally to conventional scien-
tific practice in that causal explanation is the goal.
It is therefore generally characterized by the devel-
opment and empirical examination or testing of
falsifiable theory of wide historical scope, by the
assumption of the preexistence of discrete and
identifiable cases, and by the use of formal logical
or statistical tools and replicable analytic proce-
dures. Historical narration and the ‘‘unities of
time and place’’ (Skocpol 1984b, p. 383) are delib-
erately replaced by the language of causal analysis.
Analytic formal comparison can be used to gener-
alize across time and space, to uncover or produce
limited causal regularities among a set of carefully
chosen cases, and to establish a theory’s scope
conditions. There are two major procedural sub-
types in this genre, statistical comparisons and
formal qualitative comparisons.Statistical analyses of comparative-historical phe-
nomena are logically and inferentially identical to
statistical analyses of any other social phenomena.
Thus, they rely on numerical counts, theoretical
models, and techniques of statistical inference to
assess the effects of theoretically salient variables,
to test the validity of causal arguments, and to
develop parsimonious, mathematically precise gen-
eralizations and explanations. One important way
in which the statistical method appears in com-
parative-historical inquiry is in the form of ‘‘com-
parative time-series’’ analysis, in which statistical
series charting historical change are systematically
compared across countries. Charles, Louise, and
Richard Tilly (1978) and Bruce Western (1994),
for example, first use statistical time-series proce-
dures to map and explain historical variation in
working-class activity within European nations.
They then compare, either qualitatively (the Tillys)
or quantitatively (Western), these national statisti-
cal patterns to detect and explain historical differ-
ences and similarities across these countries. The
second major use of statistical procedures relies
on quantitative data from many social units for
only one or a few time points. Emphasis in this
‘‘cross-national’’ tradition (e.g., Chase-Dunn 1979;
Jackman 1984; Korpi 1989) is on detecting causal
generalizations that are valid for a large sample or
even an entire population of countries. Though
the results of such studies typically resemble a
static, cross sectional snapshot of historical proc-
ess, this analytic strategy sometimes is necessitated
because complete time-series data do not exist for
very many nations, especially those now undergo-
ing economic and social development (e.g., Pampel
and Williamson 1989). Due both to their data
limitations and to their focus on statistical regu-
larities and theory testing, cross-national studies
typically slight historical processes and homogen-
ize cultural differences across cases. The under-
standing of cases as real social units deserving of
explanation in their own right consequently is
sometimes lost (Tilly 1984; Skocpol 1984b; Ragin
1987; Stryker 1996). In the 1990s, however, some
cross-national statistical analysts, though continu-
ing their search for general patterns, began to