Project Summary
This project addresses the relationship between electoral strategies and distributive politics, in particular, between electoral strategies and clientelism. The authors distinguish clientelism from other forms of distributive politics in two ways. First, clientelist resources are targeted towards particular groups of voters, and are not distributed indiscriminately. Second, receiving clientelist resources is contingent upon supporting the party that is distributing the resources. Thus, electoral support from voters who receive clientelist goods is coerced. Beginning with these two properties, which are located within a broader taxonomy of distributive politics, the authors address following questions: How does non-programmatic politics, and especially clientelism, work? What causes shifts away from clientelism and toward other, non-broker-mediated distributive strategies? Which kinds of distributive politics are consistent with the norms of democracy, which are inconsistent, and why? The research design utilizes a multi-layer mix of strategies and relies on a variety of data sources. The authors conduct sample surveys of voters in Argentina, Venezuela, and India, and use publicly available individual data from Mexico to make inferences about the kinds of voters whom political machines target. The authors draw upon a vast and rigorous secondary literature and offer what they believe to be the broadest empirical review of ecological studies of distributive politics yet produced. They use secondary historical materials to make arguments about the demise of at least some forms of clientelist politics in several of today’s advanced democracies. Generally then, the authors test different aspects of the theories developed in the book using different research designs, unique datasets on voters, brokers, and leaders, and qualitative fieldwork. They seek to overcome the limitations associated with each individual strategy by triangulating evidence from a variety of sources.
Data Abstract
The full project data consist of recorded surveys of city council members and brokers who work for the city council members. The surveys contain both open- and closed-ended questions and were collected in person, with the help of research assistants between 2008 and 2013. The respondents were selected using a clustered random sampling design. Four Argentinian provinces – Buenos Aires, Córdoba, Misiones, and San Luis – were selected non-randomly. Within each province municipalities were randomly selected; within each selected municipality, city council members were randomly selected and surveyed. At the end of each of these surveys, the city council members were asked to provide a list of brokers whom they employ and whom they know by name, and then asked if a proportion of the brokers could be surveyed. One-third or one-fifth of these brokers were randomly selected from these lists. Interviews took place in city council offices and in any location that was convenient for the respondent. These data were consulted because the project hinges upon organizational challenges for political machines. In particular, the theory derives analytic leverage by focusing upon a principal-agent problem that arises between party leaders and brokers: brokers target more core voters than is ideal from the party leaders’ perspective, and brokers privately consume resources that could earn votes for the party. Testing this theory requires data that measure the behavior and decisions of brokers, which is what the survey provides.
Files Description
The original organization of the data reflected the sampling strategy used to select the survey respondents into the sample. It was organized first in municipal clusters and then in clusters around a city council member. The latter clusters consist of a city council member and her brokers.
Due to the large number of interviews only a random sample of these interviews could be transcribed and shared. These data are shared because they provide the bulk of the qualitative data for the project, and because they provide a rare collection of surveys from a probability sample of brokers. These data contain anonymized transcriptions of survey interviews that were conducted with political brokers from 6 municipalities in the broader metropolitan region of Buenos Aires, Argentina. The respondents consist of city council members and brokers who work for them. The survey interviews consist of closed and open-ended questions, which ask the respondents about their political careers, the politics of their localities, and their political parties. The survey also asks the brokers to respond to hypothetical situations regarding resource distribution and their abilities to draw inferences about voters. In total the respondents were asked to respond to 43 questions. Although many questions were closed-ended, the transcriptions contain many details that were not captured with the quantitative data generated from the survey. These transcriptions come from the larger project, in which 800 political brokers were drawn as a probability sample and surveyed from four Argentine provinces: Buenos Aires, Córdoba, Misiones, and San Luis. In Buenos Aires, 255 respondents were surveyed from 10 municipalities. These transcriptions consist of interviews with 110 of the 255 respondents and come from 6 municipalities that were randomly selected from the original 10 in the sample.
To share these data we randomly selected recorded survey interviews to be transcribed, and we anonymized the transcriptions. The original survey was conducted in 10 municipalities that were randomly selected from the 24 municipalities that form the broader metropolitan region of Bueno Aires. The municipalities that were selected for this project consist of 6 municipalities that were randomly selected from the 10 municipalities in the survey sample. We transcribed all of the recorded interviews in 5 of the municipalities. In the 6th municipality, 17 randomly selected interviews were transcribed. We did not transcribe all of the interviews in the 6th municipality due to budgetary limitations. In total, this deposit contains 110 interviews of the 255 respondents who were surveyed. After we completed the transcriptions we anonymized the surveys. The decisions that guided the anonymization process are recorded in a protocol that is included in the deposit.
Originally, we were going to cluster the transcriptions by municipality. But, while anonymizing the surveys, we determined that this would create too much risk of revealing the identities of the respondents. So now the files just have a randomly assigned number and they are not organized by municipality. The B or C after the number in each file name designates whether the respondent was a Broker or Council member, respectively.
The other data from the project consist of open-ended interviews, quantitative data sets, and historical analysis. Many of these data were used to test the theories of the project with different populations that consist of voters in Argentina and populations in other countries and time periods. The open-ended interviews provided context for the analysis. All of the quantitative data are already publicly available and the historical data are available through secondary sources. The open-ended interviews could potentially be shared, but they consist of a convenience sample, are fewer in number, and are not comparable across interviews, which may limit their analytic value for other researchers.