BSSn4495: Qualitative research in security studies Nuts and bolts of process tracing March 29, 2022 Miriam Matejova, PhD Agenda • Process-tracing: issues, practice How to process trace? Examining a single instance in which the outcome did or did not occur and trying to explain why. 1. We wonder: Does C cause E? 2. We see that E is present and that C is present in a case. – But this doesn’t tell us that C caused E in that case. How could we figure that out? 3. Think about the causal logic through which C would have caused E if C did cause E. 4. Now investigate the case to see whether that causal logic in fact unfolded within the case. What caused DRC civil war? • Ethnic tensions? Causal logic(s): Historical ethnic hatreds → desire by each group to dominate or extinguish the other → increase in violence by each side designed to dominate/extinguish other group → spirals into all-out war What caused DRC civil war? • Ethnic tensions? Causal logic(s): Historical ethnic hatreds → desire by each group to dominate or extinguish the other → increase in violence by each side designed to dominate/extinguish other group → spirals into all-out war What caused DRC civil war? • Ethnic tensions? Causal logic(s): Historical ethnic hatreds → desire by each group to dominate or extinguish the other → increase in violence by each side designed to dominate/extinguish other group → spirals into all-out war Process tracing tests Hoop test • A test that a hypothesis has to pass for us to believe it (a “hoop” the theory has to jump through) – If hoop test failed: the hypothesis is greatly weakened – If hoop test passed: the hypothesis survives, but doesn’t mean it’s true Smoking gun test • A test that can point strongly to the correctness of a hypothesis – If smoking gun test failed: the hypothesis survives – If smoking gun test passed: the hypothesis is very likely true Process tracing tests (cont.) • Straw in the wind test – Passing = hypothesis is relevant but not confirmed – Failing = hypothesis is not eliminated but slightly weakened • Doubly decisive test – Passing = hypothesis is confirmed and others are eliminated – Failing = hypothesis is eliminated Process tracing tests: issues • Deterministic vs probabilistic conclusions from PT tests – Hoop test as necessary for the validity of hypothesis? – Smoking gun test as sufficient for the validity of hypothesis? • Easy vs difficult PT tests • Triviality of conditions • Ideational “clues” Process-tracing: practice • Choose a causal question • Think of a cause that produces a somewhat lengthy causal chain to the effect • Think of a causal logic/causal story that connects the cause to the effect – Causal logic: X→p→q→r→Y • What evidence would a researcher look for to support this causal logic in your case?