lecture 7 - comparative/historical and case selection Flashcards
comparative method
= the rules, standards and procedures for identifying and explaining differences and similarities between cases, using concepts that are applicable in more than one case or country
- comparison must serve a theoretically justified purpose: test theory, develop new theory/hypotheses, apply existing theory to new cases (micro replication)
(Rose: single-case studies often not enough)
- helps to avoid false uniqueness
- helps to avoid false universalism
single case study
- components
- purpose
- case selection
- advantages and disadvantages
two components
- detailed/thick description of a case -> internal validity
- engage wider discussion/literature -> external validity (challenging)
purpose:
- provide descriptive contextualization (thick description)
- apply existing theory to new contexts
- explain exceptions to the rule
- generate new theory
(single) case selection:
- critical: to testing a theory (if it holds in this case, than it is likely to hold in all cases)
- revelatory: case reveals relationship which cannot be studied by other means (e.g. more info is available on this case)
- unusual: throw light on deviant or extreme cases/outliers
- crucial (Eckstein): confirm or disconfirm a theory (kind of the same as critical cases)
advantages =
- rich/thick description
- good match of theory and evidence
- high internal validity
disadvantages =
- low external validity
- lack of context
small-N case study / comparison
= analysis of a limited number of cases (2 or more)
advantages:
- detailed in-depth analysis of cases still possible
- better ability to contextualize
disadvantages:
- high risk of selection bias (misleading inferences)
- causality tends to be deterministic, not probabilistic
case selection: MSSD or MDSD
Qualitative Comparative Analysis
QCA = comparative intermidiate N method
compromise between qualitative approaches and formalized statistical analyses
truth table: list with relevant conditions and outcomes
- crisp set = codes: absent (0) or present (1)
- fuzzy set = codes: interval scale from 0.0 to 1.0
analysis: process of paired comparison to generate or test summaries/typologies/theories
- compare / define relationships of all possible combinations of factors and conditions, usually done with software
quantitative analysis large-N comparison
advantages =
- lower selection bias -> more external validity
- can account/test for many explanatory factors simultaneously
disadvantages =
- limited ability to capture causal processes
- thin concepts and theories (e.g. simplistic indicators)
- equivalance of meaning across cases : danger of concept stretching
case selection = representative sample
historical research
focus on temporality: context and timing
- Comparative Historical analysis
- narrative case studies
- process tracing
- Event Structure Analysis
- historical institutionalism = focus on path dependency, temporal order, cirtical junctures, positive feedback etc.
data = primary sources (original historical documents) vs secondary sources (interpretation, commentary, analysis)
*key task = establish authenticity, reliability and accuracy of information
historical vs comparative research
- historical events research = single case study
- historical process research = single case study (longitudinal)
- cross-sectional comparative research = small-N comparison
- comparative historical analysis (CHA) = small-N longitudinal comparison
Event Structure Analysis
= analytical procedure to ‘unpack’ an event into intermediary causal steps or constituent parts - why and how did it happen?
- construct narrative account of what happened
- break narrative into series of short statements (reflecting key decisions, events, and dev.)
- order statements into a diagram that reflects causal sequence or relations (sufficient and necessary conditions)
crucial steps/elements case select ion
- defining full set of data units: universe/population of cases
- selecting a subset/sample of data units from this universe
!case selection = purposive
!sampling = probablistic, random
selection on the dependent or independent variable
!!!never select on the outcome / dependent variable
selecting on the independent variable is better: easier to see if the cause always goes together with a certain outcome
challenges of case selection
- selection bias
- heterogeneity = non-equal size
- historical contingency = joint-history (cases can’t be seen as independent anymore, e.g. EU member states makes them bound, behavior is shaped)
- path dependency: stable trends (earlier decisions can lead to path dependency, e.g. certain institutions, if you then want to look at the influence of a policy, you can not study it separately from the stable trend)
- outliers: including these impedes the results
case selection affects outcomes (e.g.)
e.g. Ebbinghaus 2005: relationship between social expenditure and openness: OECD countries decent relationship, control for population size leads to stronger relationship, looking at European OECD countries weakens relationship
case selection techniques
typical
- pick cases close to the regression line (average / correlation)
- is represenative
- use: theory testing
diverse
- divide data/graph into quadrants and select from each
- can be representative
- use: theory testing and generating
deviant
- take regression line and pick cases that deviate from the average
- maybe representative
- use: theory generating
extreme
- pick cases that score really high or really low in IV or DV
- not representative
- use: theory generating
influential cases
- pick cases that have unique combo of factors that have strong effect on relationship between IV and DV
- use: theory testing
- not representative
most similar and most different
- representative? depends
- ! you select on IV/cause!
crucial and pathway
- crucial = extremely important cases to test a theory
- pathway = very similar to revelatory cases: allows researchers to look into things that can’t be investigated in other cases
(when theory of case selection hits reality)
often selection according to convenience rather than ‘‘correct ways’’
- familiarity with cases
- language barrier
- availability of literature
- accessibility of data
- population of cases unknown
case selection = trade-off between what is theoretically desirable and practically feasible
