JOHN GERRING - CASE STUDY RESEARCH
SUMMARY
Chapter 1: Surveys
In political science, case study research is highly influential (more so then in other fields)
- IR: conflicts, crises, international agreements
- CP: nations, regions, localities. Political parties, interest groups, events.
- PA: agencies, programs, decisions.
Four categories of research:
1. Small-C (Case study)
2. Large-C (Quantitative)
3. Mixed-method (including both small- and large-C)
4. Other (non-empirical, no clear unit of analysis)
Chapter 2: Definitions
Case: a spatially and temporally delimited phenomenon of theoretical significance
- Must comprise the phenomena that an argument attempts to describe or explain
- Equivalent to unit with the added implication of a temporal boundary
Case Study (small-C study): intensive study of a single case or small number of cases which draws
on observational data and promises to shed light on a larger population of cases
- Highly focused, intensively studied
- Observational: if concerning a causal case, the cause is not intentionally manipulated
- Holistic: variety of styles of evidence are employed
- Multilevel inference: evidence is drawn from different levels of analysis (cross-case vs. within-
case, qualitative vs. quantitative).
- Must be possible to put the study into a larger context, even though there is a large amount of
uncertainty about representativeness
Argument: central point of a study, what it is attempting to demonstrate or prove. Either a formal
theory or specific propositions or hypotheses.
Observation (N): unit of analysis in a particular analysis. Different from a case: does not embody
units of theoretical interest.
Variable: measurements of various dimensions of an observation. Any sort of scale: nominal,
ordinal, interval, ratio.
- Y: Outcome variable, dependent variable
- X: causal factor of theoretical interest
- Z: background factors of no theoretical interest which may affect X and Y and may therefore
serve as confounders
Sample: cases or observations subjected to analysis.
Population: the sample of cases rests within a population of cases to which a given proposition
refers. The population of an inference is thus equivalent to the breadth or scope of an argument.
Chapter 3: Overview of case selection
Descriptive cases (to describe)
- Typical: mean, mode or median
- Diverse: typical sub-types
Causal cases (to explain Y)
- Exploratory (to identify Hx, hypothesis of interest)
- Extreme: maximize variation in X or Y
- Index, first instance of Y
- Deviant, poorly explained by Z
, - Most-similar, similar on Z, different on Y
- Most-different, different on Z, similar on Y
- Diverse, all possible configurations of Z
- Estimating (to estimate Hx)
- Longitudinal, X changes, Z constant or biased against Hx
- Most-similar, similar on Z, different on X
- Diagnostic (to assess Hx)
- Influential, greatest impact on P (probability of Hx)
- Pathway, X —> Y strong, Z constant or biased against Hx
- Most-similar, similar on Z, different on X
Italics concern case-selection strategies that require a minimum of two cases
Goals of case selection
- Intrinsic importance: perceived importance of a case, e.g. historical significance or personal
association
- Case independence: chosen cases should be independent of each other and others in the
population: they should be bounded units
- Within-case evidence: the main value-added, a case must provide new evidence
- Logistics: it must be possible to examine a case, convenience sampling (choosing a case based on
accessibility) is not necessarily a bad thing, cherry picking (choosing cases that prove theory) is
- Representativeness: of a larger population in whatever ways are relevant for the larger argument
Chapter 4: Descriptive Case Studies
Descriptive: Not organized around a central, overarching causal hypothesis or theory. The fields
where case study research is most dominant are most primarily concerned with descriptive
research. Descriptive case studies ar less problematic as they are not subject to problems of causal
inference. However, they are less falsifiable and therefore less generalizable.
Typical: identify a case that exemplifies a common pattern. Chosen by the virtue of
representing features that are common within a larger population, the central tendency of a
distribution. Typical is not representative, because the central tendency is not the same for
the entire distribution.
Diverse: identify cases that exemplify common patterns. Several cases that intend to
capture the diversity of a subject: typical cases of each envisioned type.
Gering - Chapter 5: Causal Case Studies
Causal: to shed light on a causal argument.
- Oriented around a central hypothesis about how X affects Y.
- Do not, as opposed to large-C research, estimate a precise causal effect and an accompanying
confidence interval.
- Case does not necessarily provide the sole basis for the estimation of the causal effect, it might
rely on large-C research to do so. Instead: focus on other aspects of the relationship.
- Ought to exemplify quasi-experimental properties: replicate the virtues of true experiment even
while lacking a manipulated treatment.
- Case selection criteria are cross-sectionally and/or (preferably) longitudinally.
Exploratory: identifying a new hypothesis. Researcher works backwards from a known outcome to
its possible causes. Goes against social-science wisdom: it selects on the dependent variable. But,
they are usually selected on the change it the outcome. Moreover, cross-case variation is often
included. Therefore, selecting on Y is not as problematic.
Extreme: maximizes variation in the variable of interest, either X or Y (most of the time).
Concepts are often defined by their extremes. Three versions: 1. Extreme values on X or Y.
2. Input or output is conceptualized in a binary fashion, one value (positive) is especially
rare. 3. Choose cases lying at both tails of the distribution, i.e. polar cases. Any case with an
SUMMARY
Chapter 1: Surveys
In political science, case study research is highly influential (more so then in other fields)
- IR: conflicts, crises, international agreements
- CP: nations, regions, localities. Political parties, interest groups, events.
- PA: agencies, programs, decisions.
Four categories of research:
1. Small-C (Case study)
2. Large-C (Quantitative)
3. Mixed-method (including both small- and large-C)
4. Other (non-empirical, no clear unit of analysis)
Chapter 2: Definitions
Case: a spatially and temporally delimited phenomenon of theoretical significance
- Must comprise the phenomena that an argument attempts to describe or explain
- Equivalent to unit with the added implication of a temporal boundary
Case Study (small-C study): intensive study of a single case or small number of cases which draws
on observational data and promises to shed light on a larger population of cases
- Highly focused, intensively studied
- Observational: if concerning a causal case, the cause is not intentionally manipulated
- Holistic: variety of styles of evidence are employed
- Multilevel inference: evidence is drawn from different levels of analysis (cross-case vs. within-
case, qualitative vs. quantitative).
- Must be possible to put the study into a larger context, even though there is a large amount of
uncertainty about representativeness
Argument: central point of a study, what it is attempting to demonstrate or prove. Either a formal
theory or specific propositions or hypotheses.
Observation (N): unit of analysis in a particular analysis. Different from a case: does not embody
units of theoretical interest.
Variable: measurements of various dimensions of an observation. Any sort of scale: nominal,
ordinal, interval, ratio.
- Y: Outcome variable, dependent variable
- X: causal factor of theoretical interest
- Z: background factors of no theoretical interest which may affect X and Y and may therefore
serve as confounders
Sample: cases or observations subjected to analysis.
Population: the sample of cases rests within a population of cases to which a given proposition
refers. The population of an inference is thus equivalent to the breadth or scope of an argument.
Chapter 3: Overview of case selection
Descriptive cases (to describe)
- Typical: mean, mode or median
- Diverse: typical sub-types
Causal cases (to explain Y)
- Exploratory (to identify Hx, hypothesis of interest)
- Extreme: maximize variation in X or Y
- Index, first instance of Y
- Deviant, poorly explained by Z
, - Most-similar, similar on Z, different on Y
- Most-different, different on Z, similar on Y
- Diverse, all possible configurations of Z
- Estimating (to estimate Hx)
- Longitudinal, X changes, Z constant or biased against Hx
- Most-similar, similar on Z, different on X
- Diagnostic (to assess Hx)
- Influential, greatest impact on P (probability of Hx)
- Pathway, X —> Y strong, Z constant or biased against Hx
- Most-similar, similar on Z, different on X
Italics concern case-selection strategies that require a minimum of two cases
Goals of case selection
- Intrinsic importance: perceived importance of a case, e.g. historical significance or personal
association
- Case independence: chosen cases should be independent of each other and others in the
population: they should be bounded units
- Within-case evidence: the main value-added, a case must provide new evidence
- Logistics: it must be possible to examine a case, convenience sampling (choosing a case based on
accessibility) is not necessarily a bad thing, cherry picking (choosing cases that prove theory) is
- Representativeness: of a larger population in whatever ways are relevant for the larger argument
Chapter 4: Descriptive Case Studies
Descriptive: Not organized around a central, overarching causal hypothesis or theory. The fields
where case study research is most dominant are most primarily concerned with descriptive
research. Descriptive case studies ar less problematic as they are not subject to problems of causal
inference. However, they are less falsifiable and therefore less generalizable.
Typical: identify a case that exemplifies a common pattern. Chosen by the virtue of
representing features that are common within a larger population, the central tendency of a
distribution. Typical is not representative, because the central tendency is not the same for
the entire distribution.
Diverse: identify cases that exemplify common patterns. Several cases that intend to
capture the diversity of a subject: typical cases of each envisioned type.
Gering - Chapter 5: Causal Case Studies
Causal: to shed light on a causal argument.
- Oriented around a central hypothesis about how X affects Y.
- Do not, as opposed to large-C research, estimate a precise causal effect and an accompanying
confidence interval.
- Case does not necessarily provide the sole basis for the estimation of the causal effect, it might
rely on large-C research to do so. Instead: focus on other aspects of the relationship.
- Ought to exemplify quasi-experimental properties: replicate the virtues of true experiment even
while lacking a manipulated treatment.
- Case selection criteria are cross-sectionally and/or (preferably) longitudinally.
Exploratory: identifying a new hypothesis. Researcher works backwards from a known outcome to
its possible causes. Goes against social-science wisdom: it selects on the dependent variable. But,
they are usually selected on the change it the outcome. Moreover, cross-case variation is often
included. Therefore, selecting on Y is not as problematic.
Extreme: maximizes variation in the variable of interest, either X or Y (most of the time).
Concepts are often defined by their extremes. Three versions: 1. Extreme values on X or Y.
2. Input or output is conceptualized in a binary fashion, one value (positive) is especially
rare. 3. Choose cases lying at both tails of the distribution, i.e. polar cases. Any case with an
