Process Tracing

The Epistemology and Methodology of Process Tracing

As outlined by Alexander George and others, process tracing can be used to develop and test historical explanations of a single case. George and other proponents of process tracing often use the metaphor of detective work to illustrate the technique. Just as a detective assesses alternative suspects' potential means, motives, and opportunities to have committed the crime under investigation, a social science researcher can use evidence to assess the alternative explanations proposed for the outcome of a historical case. This involves close attention to the hypothesized causal mechanisms, processes, and sequences that alternative theories predict must be evident in the history of a case if these theories are to explain the outcome of the case.

The emphasis that process tracing places on causal processes and mechanisms makes its epistemological assumptions similar to those of the scientific-realist school of thought in the philosophy of science. As Fred Chernoff has noted, there are many variants of scientific realism. Most self-identified scientific realists agree, however, that there are entities in the world with causal powers and that these entities act through causal mechanisms to generate the phenomena that we observe. In this view, although causal mechanisms are ultimately not directly observable, our hypotheses about how these mechanisms operate generate observable implications that we can test against actual observations to make (potentially fallible) inferences on the validity and scope conditions of our hypotheses. The observable implications of hypothesized causal mechanisms and processes may concern predictions on population-wide patterns, which can be assessed through statistical means. According to Henry Brady and David Collier, this analysis involves inferences from “data set observations” or from measurements of variables across many different cases. The observable implications of hypothesized causal mechanisms may also involve detailed predictions about processes and sequences of events within individual cases. It is these within-case observable implications that process tracing assesses through what Brady and Collier term causal process observations, or evidence on sequences and events within a single case.

Process tracing proceeds through a combination of deduction and induction, with the mix varying depending on the state of development of theories on the phenomenon under study and the availability of relevant evidence. Deductively, the researcher uses extant theories to generate observable implications within a case, which can then be tested against the evidence. In this mode, the researcher asks questions such as “Who should have known or who did what, where, when, and how if this explanation of the case is true?” Inductively, the researcher develops detailed knowledge of the sequence of events in a case and uses this knowledge to generate additional potential explanations of the case. These inductive explanations, once translated into more rigorous and deductive theories, can generate additional testable implications within the case or within other cases, thereby helping the researcher guard against confirmation bias. Even when the outcome of a case is already known, there are usually many details about processes and sequences within the case that are unknown to the investigator prior to undertaking process tracing. These previously unknown details can provide an independent check on the investigator's theoretical assumptions.

...