ME.19:11 - SoTA-Echoing
| Source | Retained contribution | Use boundary |
|---|---|---|
| Stacey et al., Methods as a form of engineering knowledge | Attention to Method knowledge, variation, loss, and engineering context. | Conceptual history supplies no project causal verdict or current architecture authority. |
| Collier, Understanding Process Tracing, with 2019 addendum | The 2011 foundation supplies within-case sequencing and diagnostic observations; the addendum develops relations among rivals and moves beyond rigid necessity/sufficiency test classes. | This historical foundation is supplemented by the later inferential treatment below; it does not replace FPF causal-use assessment. |
| Fairfield and Charman, Explicit Bayesian Analysis for Process Tracing and A Dialogue with the Data | Compare specified alternatives, condition on prior evidence, retain source dependence, and reconsider existing evidence when hypotheses develop. | These foundations discipline qualitative comparison; numerical probabilities require warranted assumptions. They supply no complete causal-use verdict or architecture decision. |
| Fully specified Bayes factors for hypothesis testing and sensitivity analysis in process tracing, v2, 2026 preprint | Model-based evidence weighting and sensitivity analysis sharpen the question of how a conclusion depends on coding, selection and weight. | Its binomial observation model and one-rival comparisons require substantive justification; clustered source dependence limits the claimed guarantees. This optional numerical development is not a default historical protocol. |
| Mahoney, The Logic of Process Tracing Tests | Weak, necessary-condition, and strongly discriminating test logic. | Grades observations; it does not make the whole chain certain. |
Current FPF C.28 | Causal-use question, rung, support components, common threat screen, verdict, use boundary, limits, and reopen. | C.28 states evidence support only; downstream choice and authority remain separate. |
Reopen when a current source changes the diagnostic logic or causal-use rules, when a representative history cannot keep differentiation and causal support separate, or when a downstream use repeatedly needs a missing result field.