Summary: How the cases you choose affect the answers you get; selection bias in comparative politics by Barbara Geddes PDF

Preview

Summary

Summary for the course Research Design...

Description

Barbara Geddes. 1990. “How the Cases You Choose Affect the Answers You Get: Selection Bias in Comparative Politics.” Political Analysis 2: 131-150. Summary: When you choose your cases that will determine the answers you get. The main idea of this article is about selection criteria. There is a taboo that states that Selection on the dependent variable is forbidden. The article shows the consequences of violating that taboo. Using some countries and not others can led the research through. They talk about some cases: - Why some countries develop faster than other? They use Taiwan, South Korea, Brazil and Mexico. However, they got some conclusions like government extensive controls over labor provoked most expressions of work discontent, but maybe there are other that doing all of that they do not reach any successful industrialisation. - State and Social revolutions: French, Russian and Chinese. The use of cases selected from both ends of the dependent variable make more this a more sophisticated design than the studies of the NICs - Selection of the endpoints of a time series. Study case: secular decline in the terms of trade for primary products. -

-

-

-

What is a dependent variable? That is something that is caused by other factors. Its the outcome. The thing that we will focus on understanding. What is an independent variable? The cause, it is a variable that can be manipulated or controlled. It has a relationship with the dependent variable. Its the one we want to use, in an implicit way. On the line, we want to use the independent variable to explain the dependent variable. The thing we observe needs to change. What are the three problems with “selecting cases on the dependent variable” according to Geddes? - ⅔. 3 problems: 1. Relationship that exists between cause and effect. For example, when you talk about revolutions, can you talk about the Haitian Revolutions if this one does not give you any importance to your studies related to revolution, or with the cases you are comparing? Cause and effect problem. 2. Similarity between the cases you have selected, are they only a similarity by accident? Is the similarity enough? Therefore, you don't have an outcome to explain. 3. What time pointing us, what period you observe. How far do you go back in your historical analysis? Depending on the time, you will get different results. What alternatives to selecting on the dependent variable does she propose? (Hint: Geddes gives three examples wherein she tests existing hypotheses with different samples). 1. Turn from the dependent variable to the independent variable than the first. 2. To say we start on selecting and focusing on the independent variable. 3. You should not select the dependent variable. o Is selecting on the dependent variable a problem for inductive research?

Yes, selecting on the dependent variable means restricting set of observations to cases in which some phenomenon of interest has been observed and excluding from the set cases in which the phenomenon was not observed. Thus, it will lead to bias and incorrect conclusions. It is still a problem because they they canʻ t test the theories theyʻre proposing, but they can identify plausible variables and highlight anomalies that defy established theories Inductive approach is about generating theories, not about testing them. If you provide an indication that the dependent variable is the one that led you to your choice of your participants or the places or the behaviours that you observe, you need to say so, you need to say why you chose that. Selection bias are more avoidable in quantitative studies, while in qualitative studies is more difficult to avoid. - What are her principal conclusions? The fact that the choice of the timeframe that you look at the selection of the dependant variable use and the fact that this relates to a link between your proposed cause and effect that might disappear with a new case those three problems are key principle conclusion....