How to compare Week 5,
Sem 1, 2015/2016
ROHANA ABDUL HAMID
TECHNIQUES
• 1. CASE STUDIES
• 2. FOCUSED COMPARISON
• 3. STATISTICAL ANALYSIS
1. Case studies
What is a case study ?
Intensive scrutiny of one case
Micro level and macro level analysis : the selection of cases. Thus, it is a case oriented , not variable oriented. Thick description. Multi-method : reading academic lit,
secondary documents, primary materials, interviews
What are cases ?
Individuals, organisations, processes, programs, institutions, events
RQ : How and why questions:
Examples of case studies
• 1) Graham Allison’s study of a single case (1971) about 1962 Cuban Missile Crises. He explained the crises by using three competing theories – U.S and Soviet Union performed as a) rationale actors, b) complex bureaucracies and c) politically motivated groups or persons.
• Single case study
Examples
2) A book about major strategies for improving social conditions by illustrating four policy topics : welfare reform, strengthening the child education, education reform and transforming neighborhoods. The book refers to specific cases of successful programs.
• Results : The writer develops generalization based on the case studies of the need for the successful programs to be results oriented
• Multiple case cross analysis
Example [case studies with multiple cases]
3) A book with 8 social programs . The commonalities in the program : the effectiveness of the programs [although many these policy programs often meets failure].
• However, the programs varied widely in their focus such as : education, nutrition, drug prevention, preschool education, delinquent youths, drug treatment, homelessness, mental health.
• Findings : can identify programs that are low in cost , programs that have convincing evidence of long term effects; enable the government to replicate the programs
Single case study : comparative?
1) Single case study can be subsumed under comparative method if it is theory-driven. How ?
a) Theory infirming and theory confirming case study : analysis of a single case within the framework of established generalizations. The case is the test of the proposition , which in turn may confirmed or infirmed by it.
….
b) Deviant case : studies of single case that are known to deviate from established generalizations.
So , why do we select the case ?
We want to uncover relevant variables that are not considered previously.
What is the theoretical value?
It weakens the original proposition , able to suggest modified proposition that may be stronger.
…..
• C) Representative case : Typical and standard example of a wider category. It is common, useful and undramatic.
• A collection of representation of cases (representative) will provide raw materials for later distillation/ subsequent generalizations.
Example : Coalition government – choose to study the phenomena in your country in detail. Home country is your research site, but you hope that the results would contribute to broader comparative understanding. HOW ?
Disadvantage of single case study and who do we do about it ?
• 1) What works in country A may not work in country B.
• Why ?
Contextual differences. But by drawinganalogies , case studies can encourage thesearch for more general knowledge .
Asking question : What differences betweencountries explain why the same policy led tocontrasting result ?
Other types of case study
4) Prototypical case study = The case is chosen because it is expected to become so.
Example : US has been used as a prototype case for democracy. Why ? This is because it is pioneering in democracy. Thus, when we study the case, it helps us to understand the phenomenon occurred elsewhere.
• 5) Archetypical case study = creates the category. • Example : French Revolution. Refer to French
revolution when you want to look at the study of revolution.
….
• 6) Critical type of case study : if it works here, it would work anywhere.
• Example: Imposing democracy in Iraq (case). Iraq – a country with little experience of democracy (a case). If democracy survives in Iraq, it would work anywhere such as Libya, Syria.
2. FOCUSED COMPARISON
• What is it ?
• A small N studies – concentrate on the intensive comparison ( of a political phenomenon)
• Choose a study because of the case
• The number of countries – either 2 ( paired , binary comparison) or 3 (triangular comparison)
Works well when few countries are compared over time.
How do you select the countries ?
• Apply two approaches : Most similar design and Most Different Design
Advantage of most similar design : the more similar units being compared, it is possible to isolate the factors responsible for differences between them .
Example : Social security programs in Canada and USA
….
• Advantage of Most Different design : Can test the relationship by finding out whether it can be observed in a range of different countries.
• If it is proven to be positive, our confidence that the relationship is real (the relationship is not on the basis of third factors, unmeasured variable)
• Example: If we find that plurality method of election of is associated with two-party system in the diverse group of countries, the robustness(strong, sturdy) in the relationship would increase.
Statistical Analysis
• Variable-based, not country specific.
• Large number of countries/cases. The data are used for generalisation.
• It permits us to explore covariation between variables.
• The variables – we can measure quantitatively.
• In statistical analysis, the variables are –dependent and independent.
Example of studies using statistical analysis :
1. The higher a person’s social status, the greater his/her participation in politics.
2. The more affluent (wealthy) a country, the more likely it is to be an established democracy.
• - Correlation : Positive or negative ?
• - Can you identify the dependent and independent variables?
Statistical approach
• Correlation : the association between variables
Correlation score is zero 0 : zero linear relationship
Correlation score , 0.5- 1 = strong correlation
Weak correlation = 0.1- 0.4
Graph : page 93-94
Risks to SA : 1) Spurious correlation
• When you have strong correlation between two variables, but the relationship of the variables (both or one) depend on third factor (unmeasured variable).
• How ? Consider correlation between income and political participation. The third unmeasured variable is likely to be education.
• Thus, we have to consider all variables (relevant) in our analysis (research)
2. The direction of causation
• Correlation is not able to determine the direction of causation. For instance ,
• A) x causes / promotes y or;
• B) y causes / promotes x
• Why ? This is because, correlation shows the association not the direction.