The Comparative Method
Why is theory needed in comparative politics? What are some examples of theoretical approaches within comparative politics?
- Without theory, comparative politics would just be collection of info → need for analysis -Research questions are to be guided by theory or itself constitute a potential answer to an existing theoretical argument -Structural functionalism: identifies necessary activities of all political systems and then compares the manner in which these functions are performed -Corporatism: stresses the central role of state and society interactions in governing -Institutionalism: focus on central role of structures in shaping politics and individual behavior
What is conceptual stretching?
-Distortion that happens when a concept developed for one set of cases is extended to additional cases where the features of the conept do not apply in the same matter
What is Sartori's ladder of generality?
-Has to do with conceptual stretching -Utilizing for wider use of a theoretical concept by extension (of its initial meaning, moving from point A to B) involves a loss of intension (observations reflect the original features). Intension reduces applicability of a concept in comparative research across more cases but enhances internal validity
Purposes of comparison
-How political problems are addressed in different contexts aids in policy learning -Exposure to new ideas and perspectives -Examine if a political phenomenon is a local issue or a broader trend -Most importantly, it is one of the primary means for establishing social scientific generalizations
Why does Shapiro criticize Large N studies?
-It's focused on quantifiable data that is measured numerically --> can lead to variables being chosen for their usefulness in statistical analysis rather than a priori relevance to the research problem -'method-driven' rather than 'problem-driven' research -equate it to a drunk looking for his keys under a lamppost because that's where the light is
What are Large N case studies?
-Lieberman: a mode of analysis in which the primary causal inferences are derived from statistical analyses which ultimately lead to quantitative estimates of the robustness of a theoretical model -Tries to assess the relationships between a given dependent variable and one or more independent variables across a large number of cases, -Context is not so much 'controlled for' as dismissed from the analysis, on the basis that if a given relationship holds across a variety of contexts, then context cannot be so important: irrelevant variables will 'wash out'.
What are Przeworski and Teune's 'most similar systems' and 'most different systems' research designs?
-MSSD: choose cases with many similar features so that variables can be held constant. Narrows that number of potential explanatory variable. However, there will almost always be enough differences between cases to overdetermine the dependent variable. Draws from Mill's Method of difference -MDSD: draws from Mill's method of agreement. seeks out similarities between cases in spite of the potentially confounding differences between them. The understanding behind this approach is that if a hypothesized relationship between two or more variables is replicated across a wide variety of different settings, than there are stronger grounds for arguing that there is a causal link between the variables. Unit of analysis should be at lowest possible level → individuals. Approach implies a preference for large-N rather than small-N research, and for quantitative over qualitative data.
Why in the comparative method used?
-Most obvious route for political scientists as controlled experiments are usually impossible in social sciences -However some controlled experiments have been done. Wantchekon had a controlled experiment on the effects of clientelism and public policy platforms on voters in Benin. -Hopkin: "We may not be able to engineer the presence of X and Z in order to observe whether they result in Y or not, but we can look around to see where we observe X, Y and Z, and analyze how variations in the independent and dependent variables relate to each other in different cases"
What is Galton's problem and how can it be countered?
-Observed differences and similarities may be caused by exogenous factors that are common to all the cases selected for comparison -Example: choice of Westminster-style parliamentary governance in former British colonies). -Example: Globalisation → diffusion creates similar influence in each case -Can be countered by triangulation or by applying comparative historical analysis
What are small-N studies? What are its two significant risks? What is one advantage?
-Qualitative research of a smaller number of cases -Risks: cases studied can be unrepresentative and findings specific to these particular cases. Don't leave room for hypotheses to be falsified. Heavy reliance on interpretation of unique events and processes exposes the analysis to the observer's bias. -Qualitative-comparative research can be distinguished from quantitative-comparative research by its 'holism'. Whereas quantitative studies are strongly analytic and produce probabilistic predictions, tending to abstract particular phenomena from their context in order to compare them across cases, qualitative studies look at the phenomena within their contexts, looking at the cases as 'wholes' (complex combinations or configurations of variables) and generating deterministic predictions
US/UK overdetermination example
-UK and US are both Anglo-Saxon and share liberal political tradition, two party system, FPTP electoral laws, less regulated markets. However these cannot explain differences between the two, for example why a strong working-class party emerged in UK but not US. There are so many potentially relevant differences though: parliamentary versus presidential government, monarchy versus republic, medium-small population versus large population, relative ethnic homogeneity versus ethnic heterogeneity. Makes impossible to pinpoint cause. -Problem also cant be helped by adding more cases because there are rarely enough cases to find right combinations of similarities and differences → Lijphart's 'too many variables, too few countries' problem
What is triangulation?
-notion of using small-N and large-N analysis as a coordinated joint research strategy -case studies can be used to generate theory, by examining a variety of possible causal mechanisms, and to test theoretical propositions in greater detail and complexity than is possible in large-N studies. The findings from case studies can be used to inform subsequent quantitative analysis, thus generating more robust generalizations than would be possible from the case study alone
What is Lieberman's nested analysis?
-quantitative analysis is the starting point of hypothesis testing, followed by small-N analysis which will have different objectives depending on the reliability of the results of the large-N analysis: if the results are weak, small-N study will be used to generate better theoretical propositions which can then be tested using quantitative methods, if the results are strong, the small-N study will serve the purpose of testing out the theory with qualitative analysis
Why can't a single method be used across all of comparative politics?
1. Research method depends on the research question (number of cases, type of data [qualitative or quantitative], time period. 2. Dimension include: spatial, functional [cross-organizational or cross-process], longitudinal comparison [across time] 3. Units of analysis can be diverse 4. Research design can focus on similarities or differences
What did Mill identify as experimental, error, and extraneous variance?
Experimental variance: observed differences or changes in the dependent variable of the research question Error variance: occurrence of the random effects of unmeasured variables. Effects are almost impossible to avoid as social science is quasi-experimental Extraneous variance: controlling for other possible influences
What are potential solutions to avoid conceptual stretching?
Family resemblance: Extends the initial concept by adding features that share some of the attributes of the original concept (AC, AB, BC) Radial categories: Each step of extension is defined by a hierarchy of attributes belonging to the initial concept → defining A as the essential attribute whereas B and C are secondary. (A+B, A+C)
What are three issues in interpreting results?
Galton's problem: Observed differences and similarities may be caused by exogenous factors that are common to all the cases selected for comparison Fallacies (individual and ecological): An ecological fallacy is when data measured on an aggregated level (such as state level) are used to make inferences about individual or group level behaviour. An individual fallacy is when the interpretation of data measured at the individual or group level is used as if they represent the whole Over-determination and selection bias: comes from case selection. When MSSD is used, chances are high that the dependent variable is overdetermined by another difference that is not catered for in the research design
What are the five I's?
Institutions, interests, ideas, individuals, international environment
What is internal and external validity?
Internal: degree to which descriptive or casual inferences from given set of cases are correct for most if not all the cases under inspection External: extent to which results of research can be considered to be valid for other more or less similar cases not included in the research Research design helps link empirical evidence to theoretical relationships by means of comparative method and enhances the internal and external validity of the results
What are four limitations to large-n studies?
Limited cases: Lack of available cases. Individual level data especially hard to obtain for pertinent issues in political science such as the behavior of individual politicians in high-level or sensitive decision-making is often hidden from public view Data availability: Some quantitative 'large-n' analyses are not that large at all (15-20 cases). Comparing advanced economies is justifiable on practical grounds as these countries have most reliable data. However, these states are also similar in many ways, so using them to make casual statements is difficult. Data reliability: Even data produced by reputable sources can be unreliable. Democracy and development literature frequently cites per capita GDP as measure of economic performance even though it's flawed (doesn't include non-monetarized economic production, calculated in USD, measures wealth but not general welfare). Issue of using qualitative data coded into numerical form. Data relies on coding a piece of legislation as more or less restrictive on the basis of the researcher's opinion. Careless conceptualization: conceptual vagueness and inconsistency pose a serious threat to the validity of empirical generalizations about the relationship between these two variables → example of economic development and democracy --> could lead to conceptual stretching
What are the Methods of Agreement and Difference?
Method of difference: -Comparing cases that differ in respect to either the dependent variable or the independent variable but do not differ across comparable cases with respect to other variables. Most similar systems design -Mill's example: if we were interested in establishing the beneficial effect of commercial protectionism on national prosperity, then we would have to find two cases similar in all respects except that one was rich and had protective tariffs, and the other was poor and espoused free trade Method of agreement: -Comparing cases in order to find those relationships between the independent and dependent variables that remain similar nothwithstanding the differences in other features of the cases Most different systems design -Mill: the two cases should be completely different in every circumstance, except that they are both protectionist and both rich, in order to confirm the theory
What can be units of analysis in comparative politics?
National political systems are most important political unit in comparative politics, but sub-national regional systems, supranational units (Western Europe), and supranational organisations (EU).
What are the options for case selection?
Single case study: only implicitly comparative. External validity is low or absent. Useful for deviant cases, use as a pilot for generating hypotheses or confirming/invalidating theory, post-hoc validation of general theory Time series: compare a specific configuration within a few cases to inspect change. Democratization as example Closed universe: looks at a few cases for comparison at different points of time based on external events (critical junctures) like war or economic crisis. European democracies maintained and democracies that switched to dictatorship during interwar period Cross section: several cases compared simultaneously. Cases resemble each other more than differ Pooled analysis: looks at cases across time and systems. Requires advanced skills in statistical methods
Fisman and Miguel's In/dependent variable in political economy literature on corruption example
Usually a focus on institutional variables (type of legal system, structure of the civil service) or on role of entrenched traditions and norms → Adjudicating between these rival views is difficult because countries with a 'culture of corruption' also tend to have inappropriately designed institutions. Fisman and Miguel studied this by examining the diplomats in New York City where parking regulations are relatively strict so the institutional design favors citizens that behave in the law-abiding way. But diplomats enjoy immunity from these fines, so the researchers examined whether the cultural norms held by diplomats made any difference to their compliance with New York's parking regulations (the institutional design is held constant, the cultural norms are allowed to vary). find that diplomats from countries which have high levels of corruption incur more parking fines.
What is a variable?
Variable is a concept that can be systematically observed and measured in various situations
What is the goal of comparative politics?
to describe differences and similarities between political systems and their features; to explain these differences; to predict which factors may cause specific outcomes
