flashcards Lect3 – case studies: goals, tools and trade-offs
overarching goals vs intermediate goals vs tools
overarching goals = embedded in specific methodologies
- positivism = descriptive and causal inferences
*important also: refining theory - interpretivism = investigate how reality is socially constructed
intermediate/immediate goals =
- particular objective that contributes to achieving overarching goal
- involve trade-offs: you can’t achieve everything at the same time
- e.g. test hypotheses or develop hypothesis
tools/methods = practices and procedures for achieving intermediate goals
- also trade-offs: not one tool can achieve everything
qualitative vs quantitative on no variance designs
KKV = focus on something that varies, don’t analyze just one case study
this implies that methods guide questions rather than the other way around -> overlook relevant research without variance (e.g. what caused WW1)
- horse behind the car
-> this improves inference, but not innovation
quali + no-variance can be innovative
-> let your question guide your RQ, not the other way around
qualitative vs quantitative on conceptualization and measurement
KKV = focus on measurement validity and reliability + avoid data organizing with typologies
- validity = conceptualization + operationalization
- reliability = operationalization + measurement
BUT: this ignores important concepts that are hard to conceptualize and measure (e.g. political culture, legitimacy)
+ typologies can be good for causal inference (e.g. by identifying causal heterogeneity)
qualitative vs quantitative on selection bias
KKV avoid selecting cases that don’t represent the population (-> large random sample)
!representativeness depends on how you define a universe or population of cases: are countries comparable? are they over time?
!representativeness is not the only possible immediate research goal:
- hypothesis formation, theory refinement
- concept formation through few cases
- analyzing cases that are actually comparable without stretching comparability through a large sample
- comparing cases that are not obviously cases (changing unit of analysis, e.g. historical regions)
set relational causation
= correlational causation vs set-relational causation
diff set-relational causal relationships:
- sufficiency
- necessity
- necessity and sufficiency
- equifinality
- conjunctural causation
- INUS
sufficient condition
X -> Y
all cases of the cause are within the outcome
i.e. all cases with X are Y
e.g. all democratic dyads (X) are within peaceful dyads (Y)
necessity
X <- Y
all the cases of the outcome (Y) are within the cases with the cause (X)
i.e. all cases with Y must have X, there is no Y outside of X
necessity and sufficiency
X = Y
all cases with the outcome are equal to the cases with the cause
equifinality
more than one path (cause) to the same outcome
either X1 or X2 is individually sufficient for Y, but neither is necessary
X1 or X2 -> Y
(within Y lie 2 different X)
conjunctural causation
two conditions produce the outcome but only if they are together
neither of the causes individually sufficient for the outcome
e.g. for welfare state entrenchment you need both conservative government and economic crisis
INUS conditions
= when there is more than one causal path to the same outcome
there are multiple conjuctural causes
- condition in each conjuctural cause is insufficient but necessary for the conjuncture (IN)
- conjunctures are not necessary but sufficient for the outcome (US)
different case studies
based on their objectives:
- ideographic: describe/explain/understand a case as an end in itself (not focused on generalizing findings)
- can be inductive or theory-guided - hypothesis generation: proposing or refining theory
- e.g. by specifying a causal mechanism - hypothesis testing
- use process tracing, structured comparisons, crucial cases etc. - plausibility probe = exploratory case study that highlights elements that will be useful before engaging in broader research endeavour
- provide initial propositions and questions
- beware of unsystematic cases that use this label to legitimize a poor job
case selection criteria
case studies = always a purposeful sample
- crucial = most likely and least likely
- deviant = anomalies of the theory
-> refine the theory: explain under what conditions/scope it works - extreme: show clearly IV and DV of interest
- variation: show relevant variation in IV and DV
!!selection bias risk: need a sense of scope (what type of case it is) to not over- or under-estimate our results
-> what type of case it is (selection bias) does not matter when you’re e.g. testing necessary and sufficient causes OR when you’re doing a within-case analysis
case comparison strategies
- structured comparisons
most similar systems design (JS Mill’s method of difference)
- context = similar
- hypothesized condition = different
- outcome = different
- since alternative conditions (context) are similar, they can’t explain the difference in outcome
Most different systems design (JS Mill’s method of similarity)
- context = different
- hypothesized condition = similar
- outcome = similar
- since alternative conditions (context) are different across cases, they can’t explain outcome convergence
what is a case?
!not historical study that looks at all aspects of reality, it focuses on specific aspects of reality that relate to a theoretical discussion
case = often instance of a class of events (e.g. WW2 is an instance of a war)
-> explanations may be generalizable to other cases
case vs observation: you study one or a few cases, but use many types of observations/data
