Annual Report 2018

Administration VISUALIZING PARTY POLITICS: ELECTIONS AND POLITICAL CAREERS IN BRAZIL Compared to changes occurred in the analyses of public policies, the study of the functioning of the Brazilian political system lags behind. In this area, most works do not present a solid empirical base, and have used selected cases only to illustrate the main point. This mismatch brings considerable difficulties to those dedicated to improving these policies. It limits the comprehension of the restrictions imposed by politics on the formulation of innovative proposals and reduces their chances of implementation. The objective of this project is to eliminate this mismatch, by creating conditions for the discussion of proposals for innovation also based on detailed political data and empirical evi- dence. For this, it is necessary for data to be obtained rapidly, at low cost, with dissemination to a broad public. This project contributes to filling this gap by creating systems to visualize electoral data and information on political careers on the internet and with applications for tablets and smartphones. OBJECTIVE • To popularize access to electoral data and information on the career paths of Brazilian politicians, as well as to analyze and visualize these data by creating systems on the in- ternet and applications for tablets and smartphones. RESEARCH METHOD • • Use of Data Warehousing (DW) techniques and development of Business Intelligence (BI) projects to assemble databases, from loading to final consultation. First, address the problem the so-called "vertical big data" or "big noise". It is necessary to deal with extensive databases (the results of a candidate in each electoral precinct, of all members of a party, etc.), which present noise because the data are not perfectly "clean". The second methodological challenge is called "big data", also called "horizontal big data" to differentiate it from "vertical big data". In this case, the problem is not necessar- ily a substantial number of records (or lines), but instead a large number of fields (col- umns). At first, it is unknown which fields should compose a database, and this statistical problem (with a Bayesian appeal) requires an answer. Big data techniques are necessary because there will be gaps in the careers of politicians. The process of matching the po- litical party membership data with data from other sources, such as the RAIS (Annual List of Social Information), will be facilitated with the use of these techniques. Finally, the project includes an effort to georeference polling places. The traditional geo- referencing applications (Google Maps and Here) are used as inputs in a more powerful system that allows performing this task more effectively. The construction of this tool results in a much higher potential to locate information about public policies in general. It is possible to associate sociodemographic data with voting places, something not pro- vided by any other database. RESULTS • The system on the internet has already been prepared and opened to the public at large due to our decision to realize the "CepespData Challenge", which attracted many start- ups of information technology to suggest improvements in the CepespData. Based on the data collected, several academic articles have been published or are under analysis for publication. Each article generated by analysis of the electoral data and infor- mation on politicians' careers will have a broader public version for the general press. A book will also be published with analysis of the composition of state's cabinets. CONTRIBUTIONS OF THE STUDY • Creation of the section called "Electoral Maps" at the site (http://spatial2.cepesp.io/), enabling the public at large to visualize the electoral maps of all candidates for elective office in statewide races (governor, senator, federal deputy, and state deputy). At the end of the project, users will have simple and fast internet access to the electoral pattern over time of all relevant candidates and elective offices since 1998. The system developed for the internet permits other systems to incorporate this informa- tion into their applications or search sites, for greater dissemination of knowledge. APPLICATIONS OF THE RESULTS AND POSSIBLE EXTENSIONS OF THE STUDY The CEPESP received an invitation from the news portal JOTA (https://jota.info/) to have a permanent column. The texts will cover so-called "Evidence-Based Politics", in an anal- ogy to "Evidence-Based Public Policies".. Likewise, the group of researchers of the CEPESP involved in the CepespData project has reached an agreement with the newspaper Valor Econômico to publish 18 weekly columns on the elections using the data available in the CepespData system. These researchers also have been undertaking intense work to disclose the site among journalists covering the 2018 elections. In the same sense, a seminar will be held each year gathering journalists and academics to disseminate the CepespData system. Interest in CepespData and the analyses it enables has been growing. A discussion is underway for an arrangement with Brown University for continuing work on the site. Sim- ilarly, Fundação Brava will finance studies to be conducted by the CEPESP team on the financing of election campaigns in Brazil. Finally, researchers of the Sao Paulo School of Law (Direito SP) are developing an applied research project on "Democracy and Repre- sentation in the 2018 Elections", involving the association with CEPESP researchers and intensive use of the data from CepespData. From the standpoint of teaching, this project will serve as an embryo for a new course aimed at professionals involved in political campaigns and other interested people. AUTHORS: Ciro Biderman and George Avelino. RESEARCHERS: Jonathan Philips and Natália S. Bueno. ORGANIZATION: Center for Public Sector Policy and Economics (CEPESP) of the Sao Paulo School of Law (Direito SP), Sao Paulo School of Business Administration (EAESP) and Sao Paulo School of Economics (EESP), all of Fundação Getulio Vargas. SUPPORT: Applied Research Fund (FPA FGV) of Fundação Getulio Vargas, São Paulo State Research Support Foundation (FAPESP), Superior Electoral Tribunal (TSE) and Brown University. 50 Annual Report 2018 51 RESEARCH

View entire presentation