Causality and causal inference -

Causality and causal inference -

Causality and causal inference 4th session, reading group in qualitative methods 11/7-2008 What is causality Ontology What is causality? The interpretations of Hume: constant conjunction or skepticism Regularity views and counterfactual accounts One definition (David Lewis); A causes B if A ->B ~A ->~B Applies to singular events ontologically, but can be

generalized. Start with the idiosyncratic, and generalizations as descriptions. How can we infer causality Epistemology Causation and covariation Different causal structures that might lead to covariation The role of time Epi-phenomenality What can we infer from spatial variation? Unit homogeneity and constant effect assumptions. In general: Ceteris Paribus clauses (Gerring, ch6) What if there is no variation temporally or spatially? Counterfactual claims The intermingling of empirical evidence, theories and models, design structure and sometimes common sense

Causal inferences are always made with a degree of uncertainty. Insight into the social world, and non-observability (Hume again) Some general problems: Measurement error, omitted variable bias (KKV) or other potential mechanisms at work (Elster), endogeneity problems, multi-colinearity Inference continued Experiments and quasi-experiments The long and tedious (but enlightening) text from Campbell and Stanley dwells on the potential pit-falls with different experimental set-ups. Some key-words: learning effects, selection effects, history and time trends, unobserved variables Statistics Econometrics, pit-falls and remedies

Unit homogeneity? Fixed effects Endogeneity? Lagging variables and 2SLS Linearity? Functional form, matching Conjunctural causality? Interaction terms..(Ragin as alternative) Comparative method Case-studies The role of additional evidence, knowledge of phenomena, triangulation of different sorts and pieces of evidence (arriving at empirical implications) Causal effects KKV and counterfactuals..systematic and unsystematic components. Average causal effect and variation in causal effect. Contingencies of causal effects.

Equifinality and conjunctural causality: does not require different metaphysics, but puts different demands on design for inference Causal effect is prior to mechanisms. The latter gives us more fine grained knowledge about processes, but we need to start with something to explain. A B..through which mechanisms? A x y B z Causal mechanisms

Mayntz: Mahoney found 24 different def by 21 diff authors! Should be used on linked activities. Linking events (Elster) Sequences as important, but looping and feedback: Paul David and QWERTY Causal mechanisms and the plausibility of our inferences. Methodological individualism and the role of actors Mayntz: specify level of reality, degree of conceptual abstraction and assumed scope of application. Specification and classification: Cumulative knowledge One example from Mayntz: macro-micro, micro-micro and micro-macro. Mechanisms based explanation and causal effect-based explanation are complementary rather than contradictory

Some tools: social psychology, cognitive psychology, game theory What if multiple equilibria and many potential cognitive mechanism at work? Empirical evidence. Be aware of and report plausible alternatives and uncertainty of inf. Elster and mechanisms The dynamic aspect of the social sciences and the quest for ever more fine-grained explanations. The variety of mechanisms..pre-emptive mechanisms..Find the one that is operating Explanation and prediction Explanation and narratives..evidence Explanation and functionalism Gerring and process tracing Triangulating bits and pieces of different types of evidence

Often hard to quantify, because of the different nature of the evidence. Still try to qualitatively assess uncertainty.. Process tracing and detective work. Theories and predictions of different phenomena..various levels of analysis and aggregation, various spheres of social life Often leans on general assumptions about the social world: Stringency, inference and the need to explicate these assumptions: folk psychology and Keynes in economic history. An interesting one: people believe what they report and report truthfully! Advice from Gerring: 1) Clarify the argument (visual diagrams?), 2) verify each stage of the model, evaluate uncertainty. Focus on dubious parts.

Recently Viewed Presentations