Lecture 3 PDF

Preview

Summary

Lecture 3...

Description

LECTURE 3 CHOOSING CASES Cases - Cases: Countries, regions, groups, organisations, individuals - More cases, more credible results are - You are allowed to use historical cases like Skocpol in States and Social Revolutions Single case study can be comparative if: - case carefully selected - contrasted with previous knowledge and theory - Michels: German Social Democratic Party the ‘least likely case’ to turn into oligarchy Avoid Selection Bias - A source of selection bias is random selection in comparative research - Selection bias comes from choosing cases that you know best, are in the news, that easily prove your hypothesis, that share a positive outcome, “selecting on the dependent variable.” Mill - Laid the foundations for the comparative method - Method of difference: suggests comparing instances in which a phenomenon occurs with those in which it does not (otherwise similar). - Method of agreement: comparing different instances where a phenomenon occurs The Most Similar Systems Design - The method of differences has been developed into the most similar systems design (MSSD). It is based on the method of difference. - Cases that have a lot in common but not the outcome we want to explain. - Most widely used of the two methods and is dominant in political science today. - If you have two competiing hypotheses the solutions are stronger theory and process tracking: steps IV -> DV. Bring in a third case. - All countries still very different! - Solution: add variables! - Problem: cannot have more variables than cases! - Solution stronger theory Two competitng hypotheses? - Solutions! - Stronger theory! - Process tracing - trace the development of something Number of cases - The bigger number of cases (Big-N) gives more confidence in results MSSD disadvantages - Data collection very time consuming - Issues with measuring and concepts - Some maths skills required - Inappropriate for some topics

The Most Different Systems Design - Based on the method of agreement - Method of agreement is comparing different instances where a phenomenon occurred. - Skocpol - why did social revolution happen in France, Russia and China? - MDSD - cases that have little in common but the outcome that we want to explain

- Newton, K.! & J. W. van Deth (2010). Postscript How and What to Compare? In K. Newton & J. W. van Deth, Foundations of Comparative Politics, pp. 400-15. What can be gained from comparison? - It widens and deepens our knowledge of the political world. We can learn about a wider range of countries, and what might play a role in causing democracy. These similarities and differences then can be used to develop and to test explanations which go beyond specific cases. How to compare? - In order to explain and generalise our comparisons, comparison is distinguished in 2 ways - One is distinguishing the comparisons by the number of countries considered. Whether large number of countries to find common patterns or focusing on one or a few countries to understand how they work. - Distinguish comparisons by the strategy they use in their approach to comparison. Whether they compare similar countries or compare countries which are very different. The article looks at four kinds of comparison to outline the most important problems of designing research in comparative politics. These kinds of comparison are: - Comparing many or a few countries - Selecting comparable countries - How many countries is enough? - Comparing apples and oranges Comparing many or a few countries? - Can do cross-national comparison such as voting patterns of different social groups. - A key example is Putnam’s study of the development of democracy and civil society in Italy which examines the differences between the 20 administrative regions of the country. - The logic behind systematic comparisons of state is identical with the logic of comparing other entities. - However, comparative politics usually refers to a comparison of countries. Comparative politics involves two basic approaches Comparing many cases - Comparing many countries in called large-n comparison. - It allows us to develop or test broad generalisations across a wide variety of different conditions. - Also allows us to identify unexpected or deviant cases that are exceptions to the general rule e.g. that all of the ex-colonies of the British Empire are undemocratic when they aren’t - Pakistan - Large-n study is usually comparing 20 or 30 countries - Information of the countries must be quantified (numerical measures of population density, per capita income) and standardised (these measures are collected the same way in every country and so are directly comparable). - Large-n comparisons are often called statistical comparison as information is analysed with statistical techniques. - Large-n comparisons are best carried out on large, standardised data-sets made by the OECD and World Bank. Also through other specialised data archives - World Values Studies, Eurobarometer and the European Social Survey. Some data sets are cross-sectional (same measure in different countries at a given point in time) or are time-series (cover same events in a country or countries at different time points). - The main goal of large-n comparisons are to study relationships between various factors or variables in a set of countries. Also called variable-oriented approaches. - Problems! - Information that is important can sometimes not be easily quantified.! - Standardised quantitative data may make data available dictate research questions. ! - Every important measure may not be adequately standardised due to the way it was collected as countries have their own ways of collecting it. ! - Many statistical techniques are based on correlations which do not tell us whether one variable

is causally related to another ! - Using statistical techniques requires statistical and mathematical understanding Comparing a few cases - Small-n comparison is comparison of a few countries. It studies a few, but in a more detailed manner which allows us to understand the complexity of relations. - Allows historical and cultural understanding of countries. - Small n-studies usually looks at 5 or 6 countries. - They focus on the exact processes and relationships in a few countries they can include qualitative evidence and methods. - it uses methods to explore complex processes and relationships based on trial and error and an intuitive understanding of politics and society. E.g. could explore and understand the different nature of the Protestant religion in Denmark and the Hindu, Sikh and Buddhist religions in India, we might come to an understanding of their relationship with democracy. Cannot gain such a subjective understanding of religion by examining statistical data. - Small-n studies can handle a mass of country-specific information of a qualitative nature without any need to standardise. - The main goal of small-n approach is to understand actual processes in a few countries. It is also called case-oriented approach. - Problems! - Knowing every detail about countries does not help us reach general conclusions about democracy across the globe. ! - Hard to have in-depth knowledge and familiarity with the countries considered.! - Generalisations cannot be tested rigorously. Ragin proposed Qualitative comparative analysis (QCA) which is the systematic comparison of a few cases using specific techniques to develop and test generalisations. Selecting comparable countries - Efficiency: it can be difficult to compare even a small number of countries as it might take many years to master the politics and government of them. - Relevance - Generalisation: four or five countries may be too small to make generalisations for something. - What we compare should be determined by what we want to know in the first place. - The number of countries we select depends on the research question we ask. Relevance determines which countries should be included in our comparisons and what is relevant depends on what theory we want to test or what generalisation we are interested in developing. - John Stuart Mill distinguished between two basic types of comparisons! - Method of Agreement: if a phenomenon happens in two or more situations then the phenomenon must be in the common features of those situations. Therefore the presence or absence of a common crucial factor or factors must account for the phenomenon we wish to explain. ! - Method of Difference: if two or more situations are similar, but the phenomenon exists in only one of them, its cause must be related to the different features of its situation. For example, India and China are large countries but that Indian is a democracy cannot be the cause of its size since China is large but undemocratic - The Method of Difference provides stronger evidence of casual explanations. This method is used for regional or area studies. - Mill applied his methods mainly to the natural sciences. - Przeworski has developed 2 methods for comparisons! - Most Different Systems Design (MSDS) Follows Mill’s method of agreement. It says to select countries that are very different from each other, but which share the characteristic we want to study. ! - Most Similar Systems Design (MSSD) suggests looking at countries that have a lot in common but do not show the same phenomenon or outcome we want to explain.

- To use the methods you need to have a lot of information about the countries you select for -

comparison. And you need to pick out factors that are presumed to be crucial to our casual explanation, and to distinguish them from irrelevant ones. You need to have these requirement fulfilled before deciding between an MDSD or MSSD strategy.

How many countries is enough? - There is the small-n/large-V problem: with each additional explanatory variable (V) the number of cases (n) required for comparisons grows exponentially. Therefore, only a few explanatory variables are often too many for the relatively small number of cases available, in which case an empirical test is not possible. - If we want to test explanations statistically, the number of cases much be much larger than the number of variables. - There are 3 strategies to fix the small-n/large-V problem. - Drop the goal of testing general statements. By opting for MSSD strategy this will help concentrate on a more limited number of factors. The problem however is that we arrive at limited conclusions that are valid only for the group of countries with ‘similar systems’. For variable oriented approaches this strategy is unattractive. - A second strategy is to reduce the number of explanatory variables by specifying our theories more rigourously. - Another strategy is to increase the number of cases - use different regional units of governments in the same country and count these as being different political systems. By doing this we might avoid problems of small-n/large-V problem. However, despite different regional units of governments in the same country, they all share national institutions, common history and culture. However, we can only treat them as different cases if we stress the unique aspects of each of these states. Therefore they are not separate and don’t count as different political systems. This is the Galton problem. - Therefore we can only deal with the small-n/large-V problem by developing a specific theory stating precisely the casual variables involved and their relationship with the effect we are trying to explain. - In this situation relying on MSSD strategies seems the best way which is restricting comparisons to groups of similar countries. Comparing apples and oranges - No 2 things in the world are identical, and therefore comparative politics struggles to compare the same phenomena or similar objects in different countries. - For example the what the French call their president does not exist in Indiaa and that is Indian’s prime minister. - Therefore, a fundamental problem of comparative research is finding concepts that can be used in different countries when each is unique. Two strategies to deal with this is looking for more abstract concepts or looking for equivalent concepts. Looking for more abstract concepts - Use a more general term to put different sets of objects into one category - For example the French President and the Indian prime minister can be compared by treating them both as government leaders - political executives. - Sartori argues that searching for general concepts loosens information and covers up differences between things. - Using more general and abstract terms for comparisons implies conceptual stretching, which is broadening the meaning of a concept, this in turn, makes it unclear what is to be included and excluded. But the advantage is it makes comparisons of different phenomena possible. - Comparists need to create general concepts for comparisons which are precise, clearly defined and well grounded so we do not lose info. Sartori says this can be done by using the ladder of abstraction.

Ladder of abstraction - High level categories: universal concepts that can be used for comparisons across the world no matter how different the circumstances. - Medium level categories: General concepts are used to compare reasonably similar things. Like how left-wing party covers everything from communist to centrist democratic socialist parties in established democracies. - Low level categories: for country by country comparisons. such as Indian prime minister or the French president. Looking for equivalent concepts - Very similar phenomena or objects are not identical. However, does not mean that we can’t compare them. - To see whether the exchange rate of different currencies is over valued or under valued, you can use the Big Mac Index. The difference between the actual price of the hamburger and official exchange rate price tells us how much a foreign currency is over-valued or under-valued. - This means we can compare apparently very different phenomena in different countries if we know that they have similar positions in those countries or perform similar functions. Instead of identity, comparisons are based on the idea of equivalence (two objects or phenomena are equivalent if they have the same value, importance, use, function or result). What matters is that they operate in similar ways or perform similar functions in different situations. It is possible to compare different institutions if they preform the same function in their respective countries. - By using info about specific situations in specific countries we can identify similar phenomena without running the risk of loosing info by stretching our concepts unacceptably.

Landman, T. (2003). Issues and Methods in Comparative Politics, Routledge, pp. 24-30, 40-42, 68-75, 81-82. Introduction - There are three different strategies of comparative research: comparing many countries, comparing few countries and single country studies. Methods of Comparison - Different comparative methods need to be used depending on the research question, the time and resources of the researcher, the method the researcher likes and the epistemological position they adopt. Researchers who adhere to deductive theory may use different methods to those adhering to inductive theory. - The central distinction between different comparative methods depends on the trade off between the level of abstraction and the scope of countries being studied. - The higher the level of conceptual abstraction, the more potential there is for the inclusion of a large numbers of countries in a study. E.g. in the study of democratic institutions, a comparison of many countries may use dichotomy between presidential or parliamentary political systems. However, a comparison of Latin American political systems would have to adopt more refined categories of presidentialism since all countries in the region are presidential.

- Comparing many countries is referred to as ‘large-n’ comparison, and comparing few countries is ‘small-n’ comparison, where n is the number of countries. Comparing many countries - Many studies that compare many countries use quantitative methods. - As this method compares many countries at once, it requires a higher level of abstraction - This method came from the behaviourist revolution in the social sciences - Advantages of this is: ability to use statistics to rule out rival explanations and control for surprising factors, coverage of countries over time and space, its ability to make strong inferences that hold for more cases than not, - Disadvantages of this method of comparison is limited availability of data for many countries and many times periods, having to know mathematical and computing skills to analyse increasingly complicated data sets - Many see this method of comparing many countries as not good for analysing many topics involving complex causal mechanisms, historical processes, and deeper meanings and understanding that are highly dependent on the contextual specificities of discrete country cases....