Process Tracing

Process tracing is for causal questions where timing and sequence matter. It asks how a cause produced an outcome inside a case and tests the proposed chain against evidence.

What Is Process Tracing?

Process tracing examines the events and decisions that connect a proposed cause to an observed outcome. Its central object is the causal mechanism, the sequence of steps through which the cause produces the effect. Collier (2011) defines the method as “the systematic examination of diagnostic evidence selected and analyzed in light of research questions and hypotheses posed by the investigator.” That makes it more demanding than narrative. You test whether the evidence fits the proposed chain and whether it weakens the alternatives a reader might raise.

Within-case vs. cross-case. Process tracing stays inside one case, while comparative designs such as MSSD or MDSD work across cases. The two often work together. Comparison identifies the variable of interest, and process tracing tests how it worked in a case.

Key Concepts

Causal mechanism. A mechanism is a theorized sequence of steps, or entities and activities, that transmits a causal force from X to Y. The running example on this page is a hypothesis that economic sanctions cause policy change. Sanctions cut export revenue, firms linked to the ruling coalition lose profits, business elites pressure the government, and the government concedes. Each step needs evidence.

Temporal structure. Seawright (2016) describes a mechanism as a sequence running from an initial condition at T₁ through intermediate steps to the outcome at Tₙ. At each point the mechanism implies a state of affairs you should be able to observe. In the sanctions example, export revenue falls at T₁, connected firms report losses at T₂, and so on until the policy changes. The sequence has to hold, and each step has to leave evidence you can check at that moment. If elites pressured the government before sanctions caused any losses, the mechanism fails however plausible each step looks on its own. Treat the mechanism as a chain of linked claims, each of which passes or fails its own evidence test. One broken link puts the whole chain in doubt. When you specify your mechanism (see Step 1), give each step a place in the sequence and say what you expect to observe at that point.

Diagnostic evidence. Process tracing relies on evidence that discriminates between competing explanations by confirming or disconfirming a specific step. Beach and Pedersen (2019) distinguish the kinds of observable manifestations that can serve this purpose.

Variants of process tracing. Beach and Pedersen (2019) distinguish three variants. Student theses usually use the theory-testing variant, which derives a mechanism from the literature and tests whether it operated in the case. If theory-building or explaining-outcome process tracing fits better, read Beach and Pedersen (2019, chs. 3–5) and discuss the choice with your supervisor.

Equifinality. Different causal paths can produce the same outcome. Process tracing can show that your mechanism operated in the case, but it cannot prove that it is the only explanation, so you must address the alternative mechanisms.

Keep the chain visible. Tie every piece of evidence to a specific step. A paragraph you cannot map to a step in the mechanism is probably background narrative.

Types of Evidence Tests

Mahoney (2012) and Collier (2011) set out four classic tests, each with different implications for confirming or eliminating a hypothesis.

Test	Passing confirms?	Failing eliminates?	Logic
Straw-in-the-wind	Slightly	No	Consistent with the hypothesis, but weak
Hoop	No	Yes	A necessary condition
Smoking gun	Yes	No	Strong confirmation if found. Absence does not eliminate
Doubly decisive	Yes	Yes	Both necessary and sufficient. Rare

In the sanctions example, an official mentioning sanctions in a speech is straw-in-the-wind evidence. It fits the mechanism without showing that it operated. A hoop test sets a minimum threshold. If sanctions hurt elites, there must be evidence of actual economic losses, and without it the mechanism fails, although finding losses does not confirm the rest of the chain. A leaked memo in which business elites press the government to change course because of sanctions would be a smoking gun, but not finding one does not mean the pressure never happened. Doubly decisive tests are the strongest of the four and are very rare in social science.

When to Use It

Process tracing fits a research question that asks how or why an outcome occurred in a specific case, provided you have a mechanism to test or develop and can obtain detailed evidence on the sequence of events. It is a poor fit when the source base is too thin. For prevalence across many cases, use comparison or quantitative analysis. For meaning-making without a causal mechanism, consider framing analysis or discourse analysis.

BA vs. MA expectations. BA projects should state the mechanism in advance and assess the evidence step by step. MA projects should also justify the chosen variant and apply the evidence tests explicitly, with a fuller treatment of alternatives and equifinality. See Assessment Standards for how methodology is evaluated.

How to Apply It

Step 1: Specify Your Causal Mechanism in Advance

Before examining your case, break your explanation into discrete, sequential steps, each saying who does what and why. Be specific. “Economic pressure leads to policy change” is too vague to count as a mechanism. “Sanctions reduce export revenue, firms linked to the ruling coalition lose profits, business elites lobby the president for concessions, and the president concludes that holding out costs more than conceding” is a mechanism.

Draw it out. Diagram the mechanism with each step in its own box and arrows between them. The exercise forces you to make the causal logic explicit and shows where the chain might break.

Step 2: Derive Observable Implications for Each Step

For each step in the chain, look for the kinds of evidence that would confirm or disconfirm that the step happened. If you claim that business elites pressured the government, look for evidence that they had actually been pressing for change, such as meeting records, public statements by business leaders, lobbying disclosures, media reports of business opposition, or changes in their political contributions. Note what kind of test each piece of evidence is (hoop, smoking gun, and so on) so that you can prioritize hoop and smoking-gun evidence, which has the most inferential power.

Step 3: Collect and Evaluate Evidence Step by Step

Gather evidence for each step, usually from primary sources (government documents, speeches, archival records, interview transcripts, media coverage) supplemented by secondary scholarship. For each piece of evidence, ask four questions.

Which step does it address? Map every piece of evidence to a specific step.
What kind of test is it? A hoop test, a smoking gun, or a straw-in-the-wind.
How reliable is the source? Consider its biases, the context in which it was produced, and whether it can be corroborated.
Does it support or weaken the mechanism? If evidence contradicts a step, report it and assess what it means for the argument.

Step 4: Assess the Overall Chain

Every step must pass its hoop test, since one failure breaks the chain, and the body of evidence should fit your mechanism better than plausible alternatives. A smoking gun for every step is welcome but not necessary. Note which steps still have weak or missing evidence.

Step 5: Test Alternative Explanations

Consider at least one, and preferably two, other mechanisms that could explain the outcome, and subject them to the same tests. If those mechanisms explain the evidence equally well, you cannot make a strong case for yours. This step is required, although students often give it only passing attention in their theses. Bennett and Checkel (2015) stress its importance for the inferential power of process tracing.

Exercise: Draft your causal mechanism

State your research question and the outcome you want to explain.
Write the proposed mechanism as a step-by-step chain of three to five steps.
For each step, list at least two pieces of evidence you would expect to find if the step occurred, and note which could serve as a hoop test or a smoking gun.
Name the strongest alternative explanation and how you would distinguish it from your mechanism.
Bring this to your supervisor before you begin detailed evidence collection.

Structuring Your Thesis

Introduction

Present your how or why question about a specific case, motivate the puzzle, and preview the mechanism.

Literature Review

The competing explanations in the literature become the alternative mechanisms you test.

Analytical Framework

Justify process tracing for your question, citing the methodological literature (e.g., Beach and Pedersen 2019 and Bennett and Checkel 2015). Then lay out the mechanism step by step, derived from theory before you turn to the data, with the evidence you expect at each step and the tests you will apply. Do the same for the alternative mechanisms.

The mechanism must appear before the findings. If the mechanism first appears after the evidence, the analysis reads as a post hoc narrative instead of a test. Specify it in the analytical framework so the reader knows what you are testing before seeing the results.

Findings

Organize the findings around the steps of the mechanism. For each step, present the evidence, say what kind of test it represents, and state whether the step passes. Three to five steps fit in one chapter with a subsection each. A longer mechanism can be split across two chapters. Address alternative explanations within the findings or in a dedicated section.

Conclusion

Did the mechanism hold, and where was the evidence weakest? Relate the result to the debate from your literature review, discuss limitations (especially thinly supported steps), and suggest directions for future research.

Example structure for a process tracing thesis.

Introduction
Literature Review
Analytical Framework (method justification, full mechanism specification, observable implications, alternative mechanisms)
Findings I: Steps 1-2 of the mechanism
Findings II: Steps 3-4 of the mechanism and alternative explanations
Conclusion and Discussion
Bibliography

Example from the Literature

George and Bennett (2005, ch. 10) use the peaceful end of the Cold War as an illustration. How did Soviet foreign policy change from confrontation to cooperation in the late 1980s? Showing that the advent of Gorbachev and the end of the Cold War are correlated is one thing, but process tracing aims to identify the mechanism through which the former led to the latter. The researcher examines each link in the chain. Did new ideas emerge within a community of Soviet policy intellectuals? Was Gorbachev receptive to them? Did he then make decisions consistent with these ideas, such as accepting asymmetric arms reductions? Did feedback from the West reinforce these policies? At each stage, the researcher checks the link against the documentary and testimonial record. The researcher also considers other mechanisms, such as economic decline forcing Soviet concessions regardless of leadership, and shows why they fit the evidence less well.

The model to copy is the order. State the mechanism first, then test each link against the evidence and the alternatives. Collier (2011) shows the same logic with simpler examples.

Common Pitfalls

Confusing narrative with process tracing. A findings chapter that tells the story in chronological order is narrative. Process tracing needs a mechanism specified in advance and explicit evidence tests for each link in the chain.

Late mechanism specification. Deriving the mechanism from the evidence you use to test it is circular. You can refine it during research, but derive it from theory and state the version you test before the findings.

Cherry-picking evidence. Look actively for disconfirming evidence and report it transparently. If a step fails a hoop test, say so.

Ignoring alternative explanations. Evidence that fits your mechanism may also fit others. At minimum, subject the strongest competing explanation from the literature to the same tests (see Step 5).

Treating all evidence as equal. Say what kind of test each piece of evidence is. A thesis built entirely on straw-in-the-wind evidence stays weak however much of it there is.

Vague or unfalsifiable mechanism steps. Each step must be specific enough that evidence could count against it. “Elites were unhappy” is too vague to test. “Business leaders in the export sector publicly opposed the sanctions policy” is testable.

Ask yourself

For each paragraph of your findings draft, can you name the step it addresses and the type of evidence test it represents? If not, you may be narrating instead of tracing.
Have you presented any evidence that contradicts your mechanism? If everything fits perfectly, ask whether you have been self-critical enough.

Key Readings

Essential

Beach, D., & Pedersen, R. B. (2019). Process-Tracing Methods: Foundations and Guidelines (2nd ed.). University of Michigan Press. The most comprehensive guide, covering all three variants. Start here for method design.
Collier, D. (2011). Understanding process tracing. PS: Political Science & Politics, 44(4), 823–830. (Free PDF). A concise introduction to the method and the four evidence tests.

Advanced

Bennett, A., & Checkel, J. T. (Eds.). (2015). Process Tracing: From Metaphor to Analytic Tool. Cambridge University Press. A multi-author volume on traditions and applications of the method.
Mahoney, J. (2012). The logic of process tracing tests in the social sciences. Sociological Methods & Research, 41(4), 570–597. Formalizes the logic of the four evidence tests.
Seawright, J. (2016). Multi-Method Social Science: Combining Qualitative and Quantitative Tools. Cambridge UP. Develops the temporal logic of causal chains and combines process tracing with cross-case methods.
George, A. L., & Bennett, A. (2005). Case Studies and Theory Development in the Social Sciences. MIT Press. A foundational text on case study methods. Chapter 10 covers process tracing.

Where to start. BA students should begin with Collier (2011), then use Beach and Pedersen (2019) to design the analysis. MA students should read Beach and Pedersen in full and add Bennett and Checkel (2015).

Comparative Case Study. Comparison identifies the variable or case difference that needs explaining, and process tracing tests the mechanism inside one case.
Building a Corpus. Use it when the evidence comes from documents, speeches, policy records, or media coverage. A clean source base makes each step of the mechanism easier to test.