that may confirm or disconfirm these ideas, and identify the tests
appropriate for evaluating this evidence.
Finally, along with the value per se of refining process tracing,
this discussion is important in wider debates on political meth-
odology. Political science is in a period of major innovation in
refining tools for quantitative analysis, and in particular, quanti-
tative tools for causal inference. This trend has produced some
worries among qualitative researchers about the adequacy of their
own tools, and perhaps it has intensified the skepticism of some
quantitative researchers about causal inference in qualitative stud-
ies. This skepticism led the eminent statistician David Freedman
(2010a) to counter with the argument that the kind of qualitative
analysis involved in process tracing is indeed a type of scientific
inquiry in its own right. In that spirit, the goal here is to take
steps toward placing this form of inquiry on a more rigorous
foundation.
䡲
NOTES
Among the several colleagues who provided valuable comments on this article, Maria
Gould and three anonymous reviewers for PS deserve special thanks.
1. The approach discussed here differs from other research traditions that can be
linked to the idea of process tracing—for example, the work on mechanisms of
Tilly (2001) and McAdam, Tarrow, and Tilly (2001).
2. Within-case analysis can become multi-case analysis if different facets of the
initial “case” are analyzed. The key idea here is that the point of departure is a
single case, when viewed from the perspective of a wider comparative analysis
focused on a larger N.
3. George (1979); George and McKeown (1985); George and Bennett (2005);
Bennett (2008, 2010). On “soaking and poking,” see Fenno (1977, 884; 1978,
xiv; 1998, v). Process tracing has much in common with Lazarsfeld’s (1940,
preface) procedure of “discerning”; Campbell’s (1975, 181–82) “pattern match-
ing,” which is also advocated by Yin (1984/2008); Sewell’s (1996, 261) “causal
narrative”; Bates et al.’s (1998) “analytic narratives”; and Hall’s (2003, 391–95)
“systematic process analysis.”
4. See ^tinyurl.com/DavidCollier&.
5. Ideas about these designs based on a matching of cases are often drawn from
J.S. Mill (1974) and Przeworski and Teune (1970). For a comment on the
weakness of these designs for causal inference, see Collier, Brady, and Sea-
wright (2010a, 10).
6. Addressing this question raises issues about the logic of inquiry and the form
of social scientific knowledge that are well beyond the scope of this discus-
sion. Only a few basic points are addressed here that are salient for the accom-
panying exercises.
7. Obviously, such prior knowledge is essential in all research, both qualitative
and quantitative.
8. Waltz calls claims about these regularities “law-like statements” (p. 1). We
prefer the alternative label used here.
9. The expressions “descriptive inference” and “causal inference” are employed
here in the sense of King, Keohane, and Verba (1994, 7–8, chaps. 2–3). Their
usage can be seen as approximating an ordinary language meaning of “de-
scription” and “causation”; and by “inference” they mean that researchers
have “the goal of making inferences that go beyond the particular observa-
tions collected,” that is, they are analyzed within the larger framework used
by the investigator. This usage contrasts with ideas of “descriptive inference”
and “statistical inference” that are standard in the work of statisticians (e.g.,
Berk 2004, chap. 11).
10. Achieving good description in this sense, and developing fruitful ideas about
the unfolding of the process, may of course interact in an iterative manner.
11. Tannenwald’s study is also discussed in Collier, Brady, and Seawright (2010a,
189–90; 2010b, 509).
12. Lerner’s analysis—which is the focus of one of the exercises—is closely tied to
modernization theory, which might concern some readers; and at certain
points the presentation seems condescending. Further, the female interviewer
is presented in a sexist way (although in survey research, selecting interview-
ers in light of characteristics such as these is widely recognized as important).
However, these drawbacks are outweighed by the opportunity presented by
the chapter to illustrate the practice of making careful observations, and also
to see how they can be integrated into a complex picture of social change.
13. For a framing of qualitative vis-à-vis quantitative in terms of four dimensions,
see Collier, Brady, and Seawright (2010a, 177–82).
14. Another puzzle is explaining the disappearance of the horse, but as Holmes
himself emphasizes, that is a secondary issue (see p. 11 in the accompanying
online version of the story).
15. See Levy’s (2008) excellent discussion of counterfactuals and case studies.
16. There is a parallel here to the idea in statistical work that the test does not
stand on its own, but rather is shaped by prior assumptions. In quantitative
analysis, the construction of the statistical model depends heavily on such
assumptions, and in general the statistical test does not directly evaluate these
assumptions. Rather, it estimates the relationship based on the supposition
that the model assumptions, as well as the underlying assumption of causal-
ity, are true. See, for example, Freedman (2010b).
17. In one story, Sherlock Holmes takes a strong stand on coincidences (“Adven-
ture of the Second Stain”; in Doyle 1960, this is on p. 655). Watson refers to
the juxtaposition of two key events as “an amazing coincidence.” Holmes
replies: “A coincidence! The odds are enormous against its being a coinci-
dence. No figures can express them. No, my dear Watson, the two events are
connected—must be connected. It is for us to find the connection.” Ironically,
it turns out that these two events are only tangentially connected, so Watson’s
statement was closer to the truth than Holmes’s, and the weaker assumption
was more appropriate. In another story, Holmes is initially more cautious
about inferences and coincidences, but then he backtracks and insists on the
certitude of his inferences (“The Sign of Four,” chap. 1; in Doyle 1960, this is
on p. 93).
18. “The Sign of Four,” chap. 6 (in Doyle 1960, this is on p. 111).
19. A further perspective on unusual or bizarre explanations (see again table 6,
H8, inference d) derives from William James’s famous dictum that “every
difference must make a difference.” To put this in a less extreme form, it
might be said that some differences make a difference. In this instance, the
form of the murder was so distinctive that it called for a distinctive
explanation—which turned out to be the kick of a horse. On William James,
see Copi (1953, 331–32).
20. Andrew Bennett (personal communication) has underscored the parallel here
with diagnostic tests in medicine.
21. Waldner (2011) offers an interesting discussion of such issues.
REFERENCES
Bates, Robert H., Avner Greif, Margaret Levi, Jean-Laurent Rosenthal, and Barry
Weingast. 1998. Analytic Narratives. Princeton, NJ: Princeton University Press.
Bennett, Andrew. 2008. “Process Tracing: A Bayesian Perspective.” In The Oxford
Handbook of Political Methodology, ed. Janet M. Box-Steffensmeier, Henry E.
Brady, and David Collier, 702–21. New York: Oxford University Press.
Bennett, Andrew. 2010. “Process Tracing and Causal Inference.” In Rethinking
Social Inquiry: Diverse Tools, Shared Standards, 2nd ed., ed. Henry E. Brady and
David Collier, 207–19. Lanham, MD. Rowman and Littlefield.
Berk, Richard A. 2004. Regression Analysis: A Constructive Critique. Newbury Park,
CA: Sage.
Brady, Henry E. 2010. “Data-Set Observations versus Causal-Process Observations:
The 2000 U.S. Presidential Election.” In Rethinking Social Inquiry: Diverse Tools,
Shared Standards, 2nd ed., ed. Henry E. Brady and David Collier, 237–42. Lan-
ham, MD: Rowman and Littlefield.
Brady, Henry E., David Collier, and Jason Seawright. 2010. “Refocusing the Discus-
sion of Methodology.” In Rethinking Social Inquiry: Diverse Tools, Shared Stan-
dards, 2nd ed., ed. Henry E. Brady and David Collier, 15–31. Lanham, MD:
Rowman and Littlefield.
Campbell, Donald T. 1975. “‘Degrees of Freedom’ and the Case Study.” Compara-
tive Political Studies 8 (2) : 178–93.
Collier, David, Henry E. Brady, and Jason Seawright. 2010a. “Sources of Leverage
in Causal Inference: Toward an Alternative View of Methodology.” In Rethink-
ing Social Inquiry: Diverse Tools, Shared Standards, 2nd ed., ed. Henry E. Brady
and David Collier, 161–99. Lanham, MD: Rowman and Littlefield.
Collier, David, Henry E. Brady, and Jason Seawright. 2010b. “Outdated Views of
Qualitative Methods: Time to Move On.” Political Analysis 18 (4) : 506–13.
Copi, Irving M. 1953. Introduction to Logic. New York: Macmillan.
Doyle, Arthur Conan. 1960/1986. The Complete Sherlock Holmes, Vol. 1 and 2. New
York: Doubleday. The 1986 edition (New York: Bantam Classics) is available in
paperback, and all the stories are readily available online.
Fenno, Richard F., Jr. 1977. “U.S. House Members in Their Constituencies: An
Exploration.” American Political Science Review 71 (3): 883–917.
Fenno, Richard F., Jr. 1978. Home Style: House Members in their Districts. Boston:
Little, Brown.
Fenno, Richard F., Jr. 1998. Senators on the Campaign Trail: The Politics of Represen-
tation. Norman: University of Oklahoma Press.
.............................................................................................................................................................................................................................................................
PS • October 2011 829