Saudi Arabia
1.3. Research Design
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
1.3. Research Design
1.3.1. Research Questions
My main concern in this study is: why there was the difference on the level and form of political movement/violence between Arab states during the Arab Spring? Or, when focus on more intensive political violence/mobilization, whether civil war happened? The study tries to assure whether certain existed civil war theories can explain the difference. The theories under discussion, as I will mention in Part 1.2.3, include Boix, C-H, and F-L. And, I will use the fifteen cases, specifically speaking, 2011 situation in fifteen Arab states,8 as touchstone for them to judge their ability to explain.
In Chapter 3, I focus on the question: how do the existed theories perform while explaining the difference? And it brings me two further questions: (1) how to effectively and parsimoniously apply the three theories in explaining and (2) what is the key factor presented by the theories in explaining.
In Part 3.4.1, what is the best way to effectively and parsimoniously apply the theories when explaining? With three theories at hand, should I take them all at the same time? Or I should only choose one or two among them? I use the explanatory ability, that is, if most cases fit the theoretical prediction, as a standard to make such a judgment. Specifically speaking, I would ask: within which theory or which combination of theories, the cases really experiencing civil war are expected to be cases with civil war. After assuring the way to apply, I would turn to evaluate each individual causal factors mentioned by the theories.
Among so many factors presented by the theories, which one might be the necessary condition for the outbreak of the civil war, especially under the context of the Arab Spring? To judge whether an independent variable is necessary for the happening of civil war, in Part 3.4.2 I prepare two questions for each independent variables. First, I would ask if most cases experiencing civil war have value prone to civil war in a specific independent variable? If yes, then I can include it into the possible group of necessary condition. Furthermore, I would ask if most cases without
8 These Arab states includes Algeria, Bahrain, Egypt, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Qatar, Saudi Arabia, Syria, UAE, and Yemen.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
experiencing civil war present no value prone to civil war in the same independent variable? If yes, then this variable maybe powerful in explaining, because with it or not does effect our prediction of civil war.9
In Part 3.5, I will ask if “peace duration” and “anocracy” hypotheses can be held in two cases fitting the theoretical predication: Iraq and Yemen. Specifically, whether there is old hatred and exclusive political institution? After answering above questions, I temporarily identify two variables “peace duration” and “anocracy” as some key factors behind the difference between case with civil war and case without civil war.
In Chapter 4, I focus on the impacts of “peace duration” and “anocracy” on civil war in the two outliers, Algeria and Libya, which cannot be explained by the existed theories, and I ask the following questions. Concerning “peace duration”, whether the experience of conflicts unavoidably leads to new civil war? Do only previous conflicts accommodating to the definition of civil war cause the happening of latter civil war? As to “anocracy”, whether the used indicators a suitable indicator to capture elements about political institution leading to civil war? Does praetorian polity, the theoretical base of “anocracy” hypothesis, predict the happening of civil war? By discussing in this way, I will identify the weakness of existed theories and make some policy advice to lower the possibility of the civil war.
1.3.2. Research Methods
The main task of my study is to assure the causal mechanisms10 behind the onset of civil war during the Arab Spring. As mentioned in Section 2.3, there are several deniable hypotheses proposed in different theories, and each hypothesis represents a potential causal factor to the happening of civil war. And two of methods mentioned by George and Bennet: congruence method and process tracing are helpful to evaluate predictive power of theories and posit the casual mechanism behind the difference between civil war and no civil war.
Starting from the theories for civil war built by other scholars, the first task of this research is to assess if the situation goes as the theories predicted. Congruence
9 For more details about the necessity of the independent variable, see George and Bennett, Case Studies and Theory Development 189-192.
10 For the definition of “casual mechanism”, see Bennett and George, "Case Studies and Process Tracing" 139-140.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
method can ascertain the performance of each case on independent variables and dependent variable; by doing this, we could understand if these cases under discussion are congruent with the prediction of the hypotheses (George and Bennet, Case Studies
and Theory Development ch. 9). After that, I move my attention to two deviate cases
that is against the prediction of theories.The second task is to trace the causal mechanism behind the two abnormal cases.
The congruence method only tells the observation on the cause end and effect end, but there is still gap between them: how independent variables precisely influence dependent variable? To fill the gap, I plan to take process tracing in my research (George and Bennet, Case Studies and Theory Development ch. 10).
1.3.3. Research Limitations
The limitations on my study can be divided into three kinds: (1) limitations relevant to the research question; (2) limitations relevant to the research method; (3) limitations relevant to the data.
First, my focus is the onset of civil war rather than the type, form, or duration of civil war, which are also the hotspots for scholars. Although it is not easy to distinguish the potential causes behind the onset of civil war from the counterparts behind the type, form, or duration of civil war.
Second, my study is based on small number of case in short temporal span, and this characteristic brings some limitations. For example, my study might not serve as a disproof against any of three theories above, because all of them were developed from larger number of cases in longer temporal span, in other words, the three theories are closer to “covering law,” but my study is relatively similar to “contingent generalization”.11 Also, my study is not qualified to take part in the greater debates on civil war, such as: Which one is the main reason for civil war? Is it greed or grievance? Is it motivation or opportunity? Or this study cannot resolve the controversies as mentioned in Section 2.3, such as: If factors about ethnicity really have no impact on civil war? Both hatred from past experience and cheap weapons available result from shorter “peace duration”, and which one’s effect is more important during civil war? If having “noncontiguous territory” has significant impact
11 For detailed discussion on “covering law” and “contingent generalization”, see Bennett and George,
"Case Studies and Process Tracing" 138 and 146.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
on civil war? What is the meaning of the connection between the huge “population”
and the outbreak of civil war? How to interpret the effect of education level on civil war? Is there casual relation between dependency on primary commodities and civil war? In sum, cases discussed in my research only represent the tendency in the MENA during the Arab Spring rather than a general tendency in a long period.
Third, without doing field research, almost all of evident used in this research comes from second-handed material, such as the information collected by other scholars and the news media. It might cause two risks to my study: (1) limited information might be not enough to effectively eliminate the number of potential causal paths, because as we all know process tracing needs a larger amount of data;
(2) the topic of my research is about domestic riots; under such a confrontational situation during 2011, information, such as the number of army and the number of casualties, might be seriously distorted.
1.3.4. Research Frameworks
The rest of my research will be composed of following four chapters. The logics to organize Chapter 2 to 4 can be visualized in Appendix I. Chapter 2 lists several potential causal factors by the three statistical models about civil war. In Chapter 3, I apply congruence method, the first research method, to assesses each case’s performance on dependent variable and independent variables, and I evaluate the explanatory power of theories and identify key factors resulting in civil war or not. In Chapter 4, I shift my attention to find the reasons behind two theoretical exceptions, Algeria and Libya by sketching the causal mechanisms behind the two exceptions.
Finally, I summarize the research results and make some recommendation about theoretical development and policy in Chapter 5.