SLIDE 1
Review Mechanisms Process Tracing
Causal Mechanisms and Process Tracing
Department of Government London School of Economics and Political Science
Review Mechanisms Process Tracing
Causal Mechanisms and Process Tracing Department of Government - - PowerPoint PPT Presentation
Causal Mechanisms and Process Tracing Department of Government - - PowerPoint PPT Presentation
Review Mechanisms Process Tracing Causal Mechanisms and Process Tracing Department of Government London School of Economics and Political Science Review Mechanisms Process Tracing 1 Review 2 Mechanisms 3 Process Tracing Review Mechanisms
SLIDE 2
SLIDE 3
Review Mechanisms Process Tracing
1 Review 2 Mechanisms 3 Process Tracing
SLIDE 4
Review Mechanisms Process Tracing
Review Case Studies
Many uses of case studies In case comparisons (last week), we focused on scoring cases on variables to test theories between cases
SLIDE 5
Review Mechanisms Process Tracing
Theory testing involves: Between-case comparisons, or Across-time comparisons, or Between-case & across-time comparisons Within-case comparisons at a lower level
- f analysis
SLIDE 6
Review Mechanisms Process Tracing
Theory testing involves: Between-case comparisons, or Across-time comparisons, or Between-case & across-time comparisons Within-case comparisons at a lower level
- f analysis
SLIDE 7
Review Mechanisms Process Tracing
SLIDE 8
Review Mechanisms Process Tracing
1 Review 2 Mechanisms 3 Process Tracing
SLIDE 9
Review Mechanisms Process Tracing
Four (or five) principles of causality1
1 Correlation 2 Nonconfounding 3 Direction (“temporal precedence”) 4 Mechanism 5 (Appropriate level of analysis)
1From Kellstedt and Whitten
SLIDE 10
Review Mechanisms Process Tracing
D Y
SLIDE 11
Review Mechanisms Process Tracing
D Y A
SLIDE 12
Review Mechanisms Process Tracing
D Y A B
SLIDE 13
Review Mechanisms Process Tracing
D Y A B Z
SLIDE 14
Review Mechanisms Process Tracing
D Y A B Z Confounding
SLIDE 15
Review Mechanisms Process Tracing
D Y A B Z
SLIDE 16
Review Mechanisms Process Tracing
Four (or five) principles of causality1
1 Correlation 2 Nonconfounding 3 Direction (“temporal precedence”) 4 Mechanism 5 (Appropriate level of analysis)
1From Kellstedt and Whitten
SLIDE 17
Review Mechanisms Process Tracing
Four (or five) principles of causality1
1 Correlation 2 Nonconfounding 3 Direction (“temporal precedence”) 4 Mechanism 5 (Appropriate level of analysis)
1From Kellstedt and Whitten
SLIDE 18
Review Mechanisms Process Tracing
Mediators/Mechanisms
Definition: “the generative mechanism through which the focal independent variable is able to influence the dependent variable of interest” 2 Dropping the tautology, “the pathway(s) or process(es) by which an effect is produced” Allows us to distinguish: Direct effects Indirect effects
- 2p. 1173 from Baron, R.M., and Kenny, D.A. 1986. “The Moderator-Mediator
Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations.” Journal of Personality and Social Psychology 51(6): 1173–1182.
SLIDE 19
Review Mechanisms Process Tracing
D Y A B Z
SLIDE 20
Review Mechanisms Process Tracing
D Y A B Z M
SLIDE 21
Review Mechanisms Process Tracing
D Y A B Z M
SLIDE 22
Review Mechanisms Process Tracing
D Y A B Z M
SLIDE 23
Review Mechanisms Process Tracing
D Y A B Z M
SLIDE 24
Review Mechanisms Process Tracing
Two Uses of Studying Mechanisms
1 Determine how a causal effect comes
about
2 Establish seemingly disconnected cause
and outcome through a chain of smaller causal effects
SLIDE 25
Review Mechanisms Process Tracing
- 1. The how of the why
A causal effect is an explanation of why something occurs Mechanisms explain how that effect
- ccurs
SLIDE 26
Review Mechanisms Process Tracing
Example: Smoking
We know that smoking kills. How does this effect occur?
Source: c Wellcome Trust
SLIDE 27
Review Mechanisms Process Tracing
- 2. Sum of small effects
We may be able to establish a number
- f small linkages
The product (multiplication) of these effects is the total effect
SLIDE 28
Review Mechanisms Process Tracing
- 2. Sum of small effects
We may be able to establish a number
- f small linkages
The product (multiplication) of these effects is the total effect Two ways to conceptualize this:
Deterministic causality Probabilistic causality
SLIDE 29
Review Mechanisms Process Tracing
Pearl’s Front Door Criterion
Same rules for understanding mechanisms as causes generally Mechanisms must be: exhaustive isolated
D Y M
SLIDE 30
Review Mechanisms Process Tracing
Pearl’s Front Door Criterion
Same rules for understanding mechanisms as causes generally Mechanisms must be: exhaustive isolated
D Y M1 M2
SLIDE 31
Review Mechanisms Process Tracing
Pearl’s Front Door Criterion
Same rules for understanding mechanisms as causes generally Mechanisms must be: exhaustive isolated
D Y M1 M2 M3
SLIDE 32
Review Mechanisms Process Tracing
Pearl’s Front Door Criterion
Same rules for understanding mechanisms as causes generally Mechanisms must be: exhaustive isolated
D Y M1 M2 Z
SLIDE 33
Review Mechanisms Process Tracing
Pearl’s Front Door Criterion
Same rules for understanding mechanisms as causes generally Mechanisms must be: exhaustive isolated
D Y M2
SLIDE 34
Review Mechanisms Process Tracing
SLIDE 35
Review Mechanisms Process Tracing
Do We Care About Mechanisms?
Write for two minutes Is understanding a mechanism necessary for causal inference? When should we be satisfied that we have “bottomed out” a causal process?
SLIDE 36
Review Mechanisms Process Tracing
1 Review 2 Mechanisms 3 Process Tracing
SLIDE 37
Review Mechanisms Process Tracing
Process Tracing
Definition: “analysis of processes of change that seeks to uncover causal mechanisms and causal sequences”3 Single-case method Focused on gathering CPOs Sequence of counterfactuals
3p.300 from Brady, H.E., and Collier, D. 2004. Rethinking Social Inquiry. Rowman
& Littlefield.
SLIDE 38
Review Mechanisms Process Tracing
Causal Process Observations
Definition: “An insight or piece of data that provides information about the context, process, or mechanism, and that contributes distinctive leverage in causal inference”4 Might be used to:
Inductively generate hypotheses Deductively test a chain of causal relationships
4Brady and Collier 2004, p.277
SLIDE 39
Review Mechanisms Process Tracing
Inductive Process Tracing
Broad search for sequential steps necessary for an event to occur No a priori expectations to test Analogous to detective work
Source: Public Domain
SLIDE 40
Review Mechanisms Process Tracing
Deductive Process Tracing
Sequence of within-case hypothesis tests Theory or extant evidence guide chosen comparisons
May iterate if there is no or very weak evidence for one’s hypothesis(es)
SLIDE 41
Review Mechanisms Process Tracing
Four Process Tracing Tests5
Broadly consistent with Neyman-Pearson hypothesis testing.
1 Straw-in-the-wind test 2 Hoop test 3 Smoking gun test 4 Doubly decisive test
5Note: I am not a fan of this typology.
SLIDE 42
Review Mechanisms Process Tracing
Major Caveat: Uncertainty
Our uncertainty is a function of n
SLIDE 43
Review Mechanisms Process Tracing
Major Caveat: Uncertainty
Our uncertainty is a function of n Process-tracing is a single-case design
Reduce uncertainty by finding within-case variation Accept only high certainty about specific case
SLIDE 44
Review Mechanisms Process Tracing
Major Caveat: Uncertainty
Our uncertainty is a function of n Process-tracing is a single-case design
Reduce uncertainty by finding within-case variation Accept only high certainty about specific case
Can we gather within-case DSOs at a lower level of analysis to better understand causality?
SLIDE 45
Review Mechanisms Process Tracing
Major Caveat: Uncertainty
Our uncertainty is a function of n Process-tracing is a single-case design
Reduce uncertainty by finding within-case variation Accept only high certainty about specific case
Can we gather within-case DSOs at a lower level of analysis to better understand causality?
Local-level geographical variation Across-time variation
SLIDE 46