Comparative Research

1
Comparative Research
2
Comparative Research

• General
• All research is comparative
• Terms of comparative social science
• Comparative Research Proper (Ragin)
• Goals/Advantages
• Characteristics
• Comparative Methods
• Method of Agreement (Mill)
• Method of Difference (Mill)
• Qualitative Comparative Analysis (QCA) (Ragin)

3
All research is comparative !

• Implicit vs explicit comparison
• Synchronic vs diachronic comparison
• Case studies (n 1) and comparison
• Implicit comparison (description, concepts)
• Within-case comparison
• Case process (implicit diachronic comparison

4
Types of Case Studies

• Extreme Case
• Typical Case
• Crucial Case
• Deviant Case
• Counterfactual Case
• What makes your case significant? By assigning
  your case study to one of these types, you are
  relating your case to the universe of other
  cases, your are setting it into context. Even
  though you might be doing a case study (n1), you
  are comparing (going beyond that single case).

5
Terms of comparative social research

• Observation
• One characteristic of one case
• Description
• representation of all relevant features/aspects
  of a case
• Classification
• Grouping of cases along one (and only one )
  feature. Each case can be attributed to one class
  only.
• Ex Classification of voters according to their
  party preference Labour voters, Con Voters,
  LibDem Voters
• Typology
• Grouping of cases along two or more aspects.
• Real type aspects that can be found together
  quite often empirically.
• Ideal type (Max Weber) - one-sided accentuation
  of some logically connected characteristics. Real
  cases do not necessarily fit into these types.
• Ex Aristotles forms of government (number of
  people involved normative judgement)
• Lijpharts types of democracy Consensus v.
  Majoritarian Democracy
• Note neither of the four does offer an
  explanation of reality, they are meant to bring
  order/perspective to social life and to the
  (potentially) infinite number of possible
  observations

6
The comparative method (proper)
7
The comparative method (proper)

• Small/medium number of cases Small/medium
  number of aspects/variables
• Why use such a research strategy?
• Answer to problem in social science many
  variables, few cases (not enough for quantitative
  analysis)
• But also (according to Ragin) it is a deliberate
  choice in order to study patterns of diversity.
  Remember
• qualitative approach commonality
• quantitative approach diversity with regard to
  one variable

8
Research Goals

• Exploring Diversity
• Interpreting Cultural or Historical Significance
• Advancing Theory
• Typical goal unravel different causal
  conditions connected to different outcomes
  (Ragin 1994 108)

9
Characteristics

• Unites case orientation with variable orientation
• Detailed knowledge of cases important for
  analysis, but also part of the research goal
• Studies cases as configurations (no isolation of
  variable)
• Analytic frame not as flexible/fluid as in
  qualitative research, but much more flexible than
  in quantitative studies
• Importance of case selection (form of theoretical
  sampling, rather than random selection), because
  of quasi-experimental character

10
3 comparative methods

• Method of agreement
• Method of difference
• QCA (Qualitative Comparative Analysis)

11
Method of agreement (MDSO)

• Based on John Stuart Mill
• Basic idea
• In order to find explanations for a certain
  phenomenon (dependent variable), we look at the
  most different cases which all share this
  particular feature
• Also known as most different with same outcome
  (MDSO)

12
Method of agreement (II)

• Select dependent variable (what is to be
  explained?) Ex 2-party system
• Select cases select cases as different as
  possible, which all/both exhibit the dependent
  variable (USA, UK, 19th century, 20th century)
• Analysis look for common features in all these
  cases plurality electoral system (FPTP)

13
Method of Agreement
14
Analysis

• Case 1 and 2 differ drastically, yet in both
  cases the outcome X occurs
• The only common variable is a
• Reasoning a is the cause of X (sufficient
  condition a leads always to X)

15
Problems

• Presumption variable has always the same
  consequences (no matter of context) (peasant
  discontent in agrarian and industrial society,
  quite different consequences)
• Presumption dependent variable is caused by the
  same cause (what if different factors produce the
  same outcome?)
• Selection bias by only looking at positive cases
  (with outcome x) we cannot know, whether there
  are cases were a exists yet X does not follow. We
  need test case(s), where X did not occur (hope
  that a was not present)

16
Method of Difference (MSDO)

• Based on John Stuart Mill
• Basic idea Find the most similar cases which
  nevertheless differ with regard to dependent
  variable (phenomenon to be explained). The
  feature(s) these different cases do NOT have in
  common explains whether the result is x or not
• Also known as most similar with different
  outcome (MSDO)

17
Method of Difference (II)

• Classify cases according to dependent variable
• Select cases that differ with regard to dependent
  variable yet are most similar with regard to
  potential causes
• Look for existing differences with regard to
  independent variables (potential causes)

18
Method of Difference
19
Analysis

• Case 1 2 are very similar with respect to
  context (broad historical situation,
  institutional setting etc.) and potential causes
• Yet in case 1 X occurs, in case 2 it does not
  occur
• Case 1 and 2 differ only in one independent
  variable (b)
• Therefore b is the cause of X (necessary
  condition without b there is no X)

20
Problems

• b is necessary, but is it sufficient (perhaps a
  b are the cause)?
• How to control for context (when is it the same)?
• How to distinguish between context and
  independent variable?

21
General Problems of Mills Methods

• Direction of causality can be contested (what is
  cause, what is effect)
• Isolating variables. Patterns of variables
  (specific configurations) can not be detected as
  cause
• Do not account for multiple causation
• Uses dichotomous variables (a, non-a), what about
  matters of degree?
• No relationships of a probabilistic nature can be
  detected (relationships that are not perfect, but
  work only most of the time)

22
Purpose of Mills Methods

• Help to explore diversity
• Help to eliminate wrong hypotheses
• Help to getting closer to a causal explanation
• But should not be pursued to mechanistically
• Need to be grounded in theoretical arguments

23
Qualitative Comparative Research

• Ragins answer to (some) problems of Mills
  methods
• Allows for and detects configurations of
  variables as causes and allows for multiple
  causation
• Problem still
• Dichotomous variables (not always possible)
• Non probabilistic, only deterministic

24
You should know

• All research is comparative
• Some technical terms
• Classification v. Typology
• Dependent v. Independent variable
• Characteristics and goals of comparative research
• Basic idea of Method of Agreement/Difference
• Problems of these methods