Chapter 2 Why Do We Compare?

2.1 Causal effect and Counterfactuals

We compare to make causal inference (or argument). In other words, we want to say, X causes Y. However, it is not easy to make causal arguments. Causal inference should state causal effect, which is the difference between the states of an individual unit when it is subjected and not subjected to the cause.

However, can we observe the “identical” unit with/without treatment? No, we cannot because we are not able to observe the identical unit with/without treatment simultaneously. Thus, we need counterfactual cases. Since counterfactual cannot be observed, assumptions about counterfactuals cannot be directly tested. Therefore, we construct counterfactuals using comparable cases in Comparative Politics.

Suppose that the case we are interested in = A. Also, the case we use as counterfactual = A'. A and A' are expected to be similar with all aspects without a difference; cause. Comparing A and A', we want to draw causal inference:

Holding other conditions constant, X causes Y.

2.2 Comparative method

Comparative method is a way to justify A' is[are] relevant counterfactual(s) to compare with A. Causal inference consists of three parts: cause, outcome, and conditions.

An outcome: An outcome is the phenomena, which we are interested in (want to know).
A Cause: A cause is something that produces its effect whenever it occurs.
Conditions: Conditions are accidental happenings that help lead to specific events.

In sum, when we can translate the expression that “[H]olding other conditions constant, X causes Y.” into plain words as follows:

“When we consider accidental happenings expected to affect the outcome are hold constant (do not vary), we can say the change in X systematically leads the change in Y.”

2.3 Lijhpart (1971)

Lijphart (1971) explains three methods–experimental, statistical, and comparative method. According to him, all of these methods seek to produce scientific explanations, which consist of (1) the establishment of general empirical relationships among two or more variables, while (2) all other variables are controlled. Furthermore, the comparative method is the same as the logic of the experiment/statistical method. The only difference is the number of cases. Thus, Lijphart (1971) argues that experimental/statistical/comparative methods share the same rationale of controlling. Social science inherently has the problem of many variables-small number of cases, which makes it challenging to establish credible controls difficult.

2.4 Przeworski and Teune (1982)

In Chapter 2, Przeworski and Teune (1982) compare the two different comparative strategies: Most Similar System Designs vs. Most Different System Designs.

2.4.1 The Most Similar System Design

Similarities are constant and differences are explained. Based on the belief that systems as similar as possible with respect to as many features as possible constitute the optimal samples for comparative inquiry. The initial assumption is that individuals were drawn from the same population = systemic factors do not play any role in explaining the observed behavior. This assumption produces intrasystemic level explanations.

2.4.2 The Most Different System Design

Differences are ignored by intention, and similarities are explained. The question of at which level the relevant factors operate remains open throughout the process of inquiry. MDSD finds the population of units at the lowest level observed in the study, most often individuals. Eliminates factors differentiating social systems by formulating statements that are valid regardless of the systems within which observations are made.

2.5 Concepts you should know

Argument
Causal effect
Condition
Control
Counterfactual
Deductive approach
Falsificationism
Individual treatment effect
Inductive approach
Most different system design
Most similar system design
Problem of many variables-small number of cases
Science
Scientific explanations
Scientific method
Tautology
Theory
Treatment
Valid argument