Lecture 4: Case Studies and Case Selection Flashcards
Classification
Uni-dimensional concept (concepts with one attribute)
- dichotomies (either/or)
- purpose = systematizing cases, determining core attributes of a concept
Typologies (same level) and Taxonomy (different levels)
Two- or multidimensional concept (combination of two or more classifications)
- purpose = systematizing cases, determining core attributes of a concept, explanation
Important characteristics of both classifications and taxonomies
- mutual exclusiveness: each case belongs to one class or type only
- exhaustiveness: each case must belong to one class or type
Systematied concepts
Have a structure that determines how attributes are linked to each other
Necessary conditions concept
Case must exhibit ALL attributes in order to be subsumed under the concept
Family resemblance concept
Case must exhibit ONE attribute in order to be subsumed under the concept
Population
Universe of cases, studied and unstudied
Sample
Studied cases
Case-centered case studies (ideographic)
Aim is to describe, explain, interpret and/or understand a single case as an end in itself rather than as a vehicle for developing broader theoretical generalizations
- should still be theory-guided and explicitly structured by a well-developed conceptual framework
- interrogates mainly within-case evidence
- usually Y-centered (explanation for outcome)
- leans towards a deterministic way of understanding causal relations
- often focusses on rare events
Theory-centered case studies (nomothetic)
A case study is theory centered when it contributes to the advancement of general theory
- exploratory / hypothesis generating or modifying case study
- confirmatory / hypothesis testing case studies
Different research goals of case studies
- case-centered case studies (ideographic)
- theory-centered case studies (nomothetic)
Selection on the dependent variable
- often levied as a carnal sin by quantitative scholars
- can lead to jumping to conclusions that any characteristic that the selected cases share is a cause
- useful for the development of new theories or the identification of plausible causal variables
General objectives of case selection
- useful variation on the dimensions of theoretical interest ( X and Y )
- representativeness (external validity)
- puzzle (theoretical / empirical)
Strategies of case selection
- Theory based (cross-case vs. within-case) (theoretical prominence of a case)
- Distribution-based = statistical techniques (frequentist or causal inference)
Theory based case selection strategies
- most similar (different outcome) case study design
- most different (same outcome) case study design
- crucial case study design –> most likely / least likely
Distribution based case selection stategies
- typical case study design
- diverse case study design
- extreme case study design
- deviant case study design (partly theory-based as well)
- influential case study design
Most similar (different outcome) case studies
The chosen pair of cases is similar on all the measured independent variables, EXCEPT the independent variable of interest
- objective for case selection = variation
- selection of at least 2 cases
- used to identify and test causal effects between X and Y
- identification of most similar cases:
Small-N = cross-tabulation, Large-N = matching
strategies vs. control variable approach
Most different (similar outcome) case studies
Reverse image of the most similar (different outcome) –> The chosen pair of cases is different no all the measured independent variables, EXCEPT the independent variable of interest.
- objective for case selection = variation
- selection of a set of cases, at least 2
- used to identify and test causal effects between X and Y
- identification of most similar cases:
Small-N = cross-tabulation, Large-N = matching
strategies vs. control variable approach
Crucial case studies
Crucial case studies, based on most-likely or least-likely designs, can be useful for the purpose of testing certain types of theoretical arguments.
- objective for case selection = (un)representativeness
- useful for confirmatory research designs (theory testing)
- based on the assumptions that some cases are more important for testing a theory than others
- best to test hypotheses of the necessary/sufficient condition type
- most-likely = to confirm a theory
- least likely = to disconfirm a theory
Identifying most-likely / least-likely cases
- based on prior theoretical expectations as how X and Y should relate
- based on assumptions and scope conditions the original theory states
Typical case study
The typical case study focuses on a case that exemplifies a stable, cross-case relationship
- objective for case selection = representativeness
- puzzle of interest lies on the within-case level (causal mechanism - identification, validation, disconfirmation)
- Small-N = case needs to display the theoretically expected score on X and Y have a high representativeness of other cases in the population
- Large-N = selection of cases based on smallest possible residual
Diverse case study
The diverse case method has as its primary objective the achievement of maximum variance along relevant dimensions
- objective for case selection = variation (cases represent the full range of values characterizing X, Y, or the X/Y relationship
- selection of a set of cases (at least 2)
- can be used for both exploratory and confirmatory research goals
- for categorical variables: choose one case from each category
- for continuous variables: chose both extreme values and the mean/median as well
- for vector variables: use cross-tabulation to choose diverse cases
- for causal paths: select cases which exemplify different causal mechanisms linking X to Y
Extreme case study
The extreme case method selects a scase because of its extreme value on the independent or dependent variable of interest
- objective for case selection = rareness of the value on X or Y
- aim = maximizing variance between case and population
- used for exploratory research aims (generating hypotheses)
- Large-N = look at the standard devotions from the sample mean
Deviant case study
The deviant case methods selects that case that demonstrates a surprising value.
- objective for case selection = atypicality (not representative)
- goal = identify factors that pull back the deviant case
- useful for investigations of theoretical anomalies
- identification of new or modified hypotheses
- disconfirming deterministic propositions
- identification: by reference to some general cross-case relationship
- Large-N = logically the opposite of typical case selection, now we are looking for outlier cases that deviate from the regression line
Influential case study
These studies look at cases that do
influence overall findings of a model
- objective for case selection = checking theoretical assumptions
- if these cases would be excluded from an analysis, the conclusions of the analysis would change
- used to address concerns that the results of a study are driven by one or a few influential cases
- Large-N: techniques to identify cases with high leverage