PROCESS TRACING, Bennet & Checkel (2015) & Beach (2020)
DEFINITION: “Process tracing is the analysis of evidence on processes, sequences, and
conjunctures of events within a case for the purpose of developing/ testing hypotheses about
causal mechanisms that might causally explain the case. […]
The deductive theory-testing side of process tracing examines the observable implications of
hypothesized causal mechanisms within a case to test whether a theory on these mechanisms
explains the case. […] The inductive theory development side of process tracing uses evidence
from within a case to develop hypotheses that might explain the case.”
A buzz-word problem has arisen, where process tracing is mentioned, but often with little thought
or explication of how it works in practice.
!!! Independent variable = cause & dependent variable = effect
Techniques falling under the label of process tracing are particularly well suited for measuring and
testing hypothesized causal mechanisms.
The goal of this book is to explain the philosophical foundations, fundamental techniques,
common evidentiary sources, and best practices of process tracing to reduce the risks of making
inferential errors in the analysis of historical and contemporary cases.
Origins of the term actually date back to the late 1960s/70s, in the US, and was initially used in
cognitive psychology, and it referred to the examination of intermediate steps in a process to
make inferences about hypotheses on how that process took place and whether and how it
generated the outcome of interest.
- The DOMINO metaphor: If one had a row of fifty dominoes lying on the table after they had
previously been standing, how could one make inferences about whether the first domino
caused the last to fall through a domino process, or whether wind, a bump of the table, or
some other force caused the dominoes to fall?
o Answer given by George & Bennet use evidence on the intervening processes posited by
each of the alternative explanations
o Weakness of the definition: possible confusion about what “intervening variable” means,
because social scientists are accustomed to thinking of variables as either causal
(independent) or caused (dependent), but in this case the intervening variable is both fully
caused by the independent variable(s) that preceded it, and it transmits this causal force
force, without adding to it, subtracting from it, or altering it, to subsequent intervening
variables and ultimately through them to the dependent variable.
o When the observable events that intercede between hypothesized causes and observed
effects have this character, they constitute “diagnostic evidence”
Different types of variables:
- Exogenous variables: excluded from a model because they are either not powerful or frequent
enough, or too complex/unpredictable, to be brought into the theory
- Complementary variables: add to or subtract from the effects of the main variables of interest,
but do so independently, or without interaction effects related to the main variables.
1
- Additional variables: interact with the independent variables of interest in more complex ways
need to be either brought into the model (endogenized) or identified but set aside from the
model (exogenized) for the sake of simplicity.
In sum, process tracing is a key technique for capturing causal mechanisms in action, and there´re
metrics and best practices that allow one to distinguish good process tracing from bad.
3 issues for the definition of causal mechanisms:
1. Are causal mechanisms in some sense unobservable?
We make hypotheses or theories about how such ontological entities as mechanisms might
work, and we test the observable implications of these hypotheses, but we do not observe
causal mechanisms directly.
We do not get to observe causality – we make inferences about it.
2. Does explanation via reference to causal mechanisms involve a commitment to
methodological individualism, or beyond that, to explaining human behavior by
neuroscience and ultimately by sub-atomic physics?
Explanations need not always go to the individual level of analysis (or beyond); it is possible
to do process tracing on hypothesized causal mechanisms at the macro level
3. Do causal mechanisms involve relations of sufficiency or probabilism?
Thorny issue, stochastic relations have some aspects of causal explanation, but lack others.
The core problem is that even if the world is deterministic, we observe it as probabilistic because
of measurement error and specification error, including the omission of important variables from
our models.
Causal mechanisms are ontological entities and processes in the world, and theories or
hypotheses are in our heads; we theorize about mechanisms. Such mechanisms are ultimately
unobservable, but our hypotheses about them generate observable and testable implications.
Main issues with process tracing:
a) Generalizability
b) Constructivism/interpretivism
c) Bayesianism
a) Generalization can be a problem
Causal mechanisms are operationalized in specific cases, and process tracing is a within-case
method of analysis, generalization can be problematic.
The use of process tracing to test and refine hypotheses about causal mechanisms can clarify
the scope conditions under which a hypothesis is generalizable.
In short, we may uncover hypothesized mechanisms through process tracing that may be
either very generalizable or unique to one or a few cases, but it is almost impossible to know
prior to researching a case the degree to which any inductively derived explanations will be
one or the other.
Insufficient attention has been given to the issues of generalizability and theory development
raised by process tracing
b) Relation between process tracing and interpretivism
2
Most constructivists agree that structures or institutions are social as well as material,
and that agents and structures are mutually constitutive; but differ on some issues.
o Three types of constructivist approaches: conventional, interpretive and
radical/post-modern
In sum, while there are philosophical hurdles to be bracketed –there´re possibilities for
developing a richer understanding of process tracing by drawing upon the various ideas
of the different approaches.
c) Bayesianism
Bayesian logic tries to use probabilities to statistical problems, to update prior beliefs in
light of the evidence of new data. The probability expresses a degree of BELIEF in a specific
event.
Bayesianism is the most developed logic in the context of process tracing, and the two are
in agreement in key respects.
o Both use evidence to affirm/reject a hypothesis;
o Both put importance on the probative value of evidence relative to competing
explanations;
o Both allow for the possibility that a few pieces or even one piece of evidence with
high probative value can help observers to converge in their views on the proper
explanation of the case;
o Both warn against becoming 100% confident in any theory/explanation due to the
limits on observational evidence and the possibility that undiscovered theories may
yet prove superior to existing ones;
o Central to Bayesianism and process tracing is the idea that some pieces of
evidence provide higher inferential power than others
Techniques and BEST PRACTICES of p.t.
Process tracing usually proceeds through a combination of induction and deduction The
particular mix depends on the prior state of knowledge and theorizing about the case selected for
study.
- Little prior knowledge: inductive process (involves analyzing events back in time)
- Some prior knowledge already exist, e.g. theories that appear to offer potential
explanations of the case, already exist: proceed + deductively
In this phase it´s important to seek to identify “new facts” (i.e. additional observable implications)
to avoid confirmation bias (the tendency to search x, interpret, favor, and recall information in a
way that confirms/supports one´s prior beliefs/values).
ADVANTAGES of process tracing
o It helps to address the limits of Mill´s methods of comparison
Mill himself recognized that the possible presence of equifinality – i.e. multiple paths to
the same outcome – could threaten inferences based on comparisons of small numbers of
cases; and omitted variables can undermine case comparisons.
o Process tracing can also be combined with QN techniques, in a mixed-method design
o Variation: (quasi-)quantitative techniques such as agent-based modeling can be
used to check the plausibility of inferences about causal mechanisms derived from
process tracing.
How is a good process tracing recognizable? 3-part standard
3
1. Meta-theoretically: grounded in a philosophical base, ontologically consistent with
mechanism-based understandings of social reality and methodologically plural.
2. Contextually: will utilize pluralism to reconstruct carefully hypothesized causal processes
and keep sight of broader structural discursive contexts.
3. Methodologically: will take equifinality seriously and consider the alternative causal
pathways through which the outcome of interest might have occurred.
Not all criteria may be relevant for any given study. However, they should serve as a starting point
and checklist, thus maximizing the likelihood of conducting good process tracing.
The ten best practices can thus often be addressed sequentially, over time, and not all at once.
4
Actors actually do make calculations and choices through rational processes to maximize their
preferences. This raises several challenges for process tracing.
a. Revealed preference problem
b. How can we infer actors’ real preferences, given that they are often engaged in strategic
contexts that provide incentives to misrepresent them?
c. How can we avoid circularity or tautology by inferring preferences separately from the
behavioral choices that these preferences are supposed to explain?
1) Process tracing is the analysis of processes, sequences, and conjunctures of events within a case
for the purposes of developing and/or testing hypotheses about causal mechanisms that might
causally explain the case. One of the issues with process tracing is generalization, while an
advantage is that it has a high internal validity.
2) Process tracing is the examination of intermediate steps in a process to make inferences about
hypotheses on how that process took place and whether and how it generated the outcome of
interest. His usage can lead to generalization, that can be problematic. But it also helps on the
methodological scope, by taking equifinality seriously and considering the alternative causal
pathways through which the outcome of interest might have occurred.