COMPARATIVE-HISTORICAL SOCIOLOGYincorporate a more thorough appreciation of his-
tory and cultural difference into their analyses
(e.g., Western 1994).
Formal qualitative comparison, by way of con-
trast, views cases ‘‘holistically,’’ as qualitatively dis-
tinct and independent units that cannot (or should
not) be decomposed into scores on quantitative
variables as in statistical analysis. This strategy
adopts a ‘‘case-oriented’’ approach that pervades
the entire research process (Ragin 1987). The
explanation of a few carefully chosen cases, for
example, generally is the rationale for and product
of the analysis, and explicit strategies have been
designed to select proper instances to be com-
pared and analyzed (Przeworski and Teune 1970;
Frendreis 1983; Lijphart 1971; Stryker 1996). De-
tection of causal regularities is often inferred
through the application of John Stuart Mill’s ([1843]
1967) inductive canons or ‘‘method of agreement’’
and ‘‘method of difference’’ to comparative data
(Skocpol 1984b; Ragin 1987). Using the method of
agreement, analysts select cases that have positive
outcomes on the phenomenon under study but
that differ on putative explanatory conditions.
These cases are then compared to see what causal
factor they share. Alternative explanations are
eliminated if antecedent conditions representing
those claims do not occur in all cases with positive
outcomes. The ‘‘method of difference’’ is used to
guard against false inferences adduced from the
method of agreement. Here analysis is conducted
with cases instancing both positive and negative
outcomes but that are as similar as possible on the
putative causal factors. The objective of Mill’s
methods is to find the one condition that is pres-
ent in all positive cases and absent in all negative
cases (Skocpol 1984b; Ragin 1987).
Mill’s methods are marred because they pre-
suppose that one causal factor or configuration
holds for all cases with positive instances. Histori-
cal patterns displaying ‘‘causal heterogeneity’’ or
‘‘multiple causal conjunctures’’—two or more dis-
tinct combinations of causal forces generating the
same outcome (Ragin 1987)—are therefore logi-
cally ruled out. This shortcoming results from the
excessive weight Mill’s canons give negative cases:
Because exceptions exist to almost any general
process, the inability to find causal universals that
are doubly confirmed by the twin logics of agree-
ment and difference can rule out virtually any
nontrivial explanation (Lijphart 1971; Ragin 1987;
Mahoney 1999). However, Charles Ragin’s (1987)
alternative comparative algorithm—’’qualitative
comparative analysis’’ (QCA)—allows for the de-
tection of causal heterogeneity in a large number
of cases. QCA uses a data reduction strategy root-
ed in Boolean algebra to search for similarities
among positive instances and to exploit the infer-
ential utility of negative cases. Negative or ‘‘devi-
ant’’ cases may be explained by, or give rise to, an
alternative causal process, but they are not allowed
to invalidate all causal generalizations. Ragin ar-
gues that QCA’s Boolean logic most closely ap-
proximates the mode of reasoning—including the
use of logically possible ‘‘historical hypotheticals’’
and ‘‘historical counterfactuals’’—employed by We-
ber (1949) and Barrington Moore (1966) in their
powerful but less formalized comparative-histori-
cal studies.Formal qualitative comparison can provide
both historically grounded explanation and theo-
retical generalization. QCA is especially useful
because it permits multiple causal configurations
to emerge from comparative historical data and
largely removes the inferential problems induced
by the analysis of a small number of cases of
unknown representativeness. Nonetheless, detrac-
tors—who otherwise disagree about preferred re-
search strategies—believe that formal comparison
via either Mill’s methods or QCA often is fraught
with hidden substantive assumptions, unable to
exert sufficient analytical control over the multi-
tude of competing explanations, and compromised
by its roots in inductive logic (Burawoy 1989; Kiser
and Hechter 1991). Moreover, formal compari-
sons rest on a key assumption that is seriously
challenged by proponents of holistic interpretive
comparison—that the historical cases themselves
are not systematically interrelated (Wallerstein
1974; McMichael 1990).Interpretive comparisons.. Interpretive com-
parisons are most concerned with developing a
meaningful understanding of broad cultural or
historical patterns (Skocpol 1984b). Two very dif-
ferent comparative logics, ‘‘holistic’’ and ‘‘indi-
vidualizing,’’ are used to construct historical inter-
pretations. Although neither logic relies extensively
on formal analytic procedures, nor is geared to-
ward testing theory, interpretive comparisons are
not necessarily atheoretical or lacking in rigor.
Indeed, concepts and theories, often of sweeping
scope and grandeur, are extensively developed