This week's analysis focuses on data I collected in a survey assignment that my undergraduate students had to complete as part of the course “Theories of IR”. The main goal of the assignment is to replicate existing research in the relevant topics we address in class and demonstrate to students how we build research design and find similar results.
The Assignment: Democracies in International Conflict
The students answered a survey in which they read two scenarios about an international security situation, and had to offer their level of support for the US president's decision to attack a state that is alleged to develop nuclear weapons.
The main factor tested in both scenarios is the adversary – the state that develops nuclear weapons is either a democracy or a non-democracy. The respondents state their level of support using a 5-point scale item (from "Favor attack" to "Oppose attack").
In the figure below, I present the proportion of support for the president's decision in the first scenario. The break-down of responses is separated by the regime of the adversary (democracy or non-democracy).
The findings provide preliminary support for the original study – when the adversary is non-democracy, respondents favor the decision to attack. However, they display more opposition when the adversary is a democratic state.
I run an additional analysis with the objective of offering clearer results. For this analysis, I use the data collected in both scenarios/experiments and implement two steps. First, I aggregate the results into two categories ("Favor" or "Oppose") by combining the results of "Somewhat" and "Strong" for each opinion. Second, I remove the middle category of “Neither”.
Those data manipulation steps should clarify the effects of the variation in the main factor – the regime of the adversary. The figure below displays the proportions for "Favor" or "Oppose" attack, separated by regime of the adversary, and in both experiments.
The results of both scenarios/experiments show the successful replication of the work by Tomz and Weeks (2013) and offer support for the theoretical argument – citizens are less supportive for military attack when the adversary is a democratic state.
If you are interested in the full R code and the associated document that describe the experiment procedure and analysis described here, check out my Github page.
Comentarios