GV249 Seminar WT1: Case Studies

💼 Comparative case studies

📖

Geddes - Paradigms and Sand Castles - Chapter 4

We should seek to test theories on different cases to those the theory was developed with.
Difficult to retain generalizability. We should not seek to over-explain each case.
- When the “the variables outnumber the cases, and explanation degenerates into description” (p. 151)

💼 Comparative case studies

📖

Skarbek - Covenants without the Sword?

Great example of hypothesis testing using qualitative case study analysis.

🧫 Mill’s Methods

Method of Agreement (Most Different Systems Design)

If two or more instances of the phenomenon under investigation have only one circumstance in common, the circumstance in which alone all the instances agree, is the cause (or effect) of the given phenomenon.

All experience the DV; but different IVs, apart from one (the cause).

Method of Difference (Most Similar Systems Design)

If an instance in which the phenomenon… occurs, and an instance in which it does not occur, have every circumstance in common save one, that one occurring only in the former; … is the cause … of the phenomenon.

Some experience DV, others not. Between these groups, all IVs are the same apart from one (the cause).

Mill: MoA is more useful for thinking about the effects of causes, and MoD is more useful for thinking about the causes of effects.

📈 Synthetic Control Method

Useful for looking at the effects of very large shocks on “macro” outcomes.
We find the most similar cases where the case did not happen to the case where the cause of interest did happen.
We then apply statistical procedures to data from the most similar cases to create a single weighted (by pre-treatment similarity) average → the “synthetic control”.
Increasingly used in the private sector to evaluate causal effects (e.g., Netflix estimating the effect of releasing a new series on subscriptions).

📈 Synthetic Control Example: Effect of Brexit

Born et al (2019) want to know the effects of Brexit on the UK’s economic performance.¹

Challenges

The treatment (leaving the EU) was not random.
We don’t have similar enough individual cases for MoD.

📈 Synthetic Control Example: Effect of Brexit

🗣️ Group Exercise

Come up with a phenomenon that you want to explain or something that you want to understand the cause of.
Design a comparative case study using either (1) the Method of Agreement with Most Different Systems Design or (2) the Method of Difference with the Most Similar Systems Design.
1. Which cases would you select?
2. What key variables would you want to include in the comparison?
How might we use the Synthetic Control Method to study the same question (in the case of MoD in MSS design)? What would be the opportunities and constraints in using comparative case studies vs. Synthetic Control?

🧫 Mill’s Methods

Method of Agreement (Most Different Systems Design)

If two or more instances of the phenomenon under investigation have only one circumstance in common, the circumstance in which alone all the instances agree, is the cause (or effect) of the given phenomenon.

All experience the DV; but different IVs, apart from one (the cause).

Method of Difference (Most Similar Systems Design)

If an instance in which the phenomenon… occurs, and an instance in which it does not occur, have every circumstance in common save one, that one occurring only in the former; … is the cause … of the phenomenon.

Some experience DV, others not. Between these groups, all IVs are the same apart from one (the cause).

Mill: MoA is more useful for thinking about the effects of causes, and MoD is more useful for thinking about the causes of effects.