Unleashing the potential of Election Data (National Democratic Institute)

Page 1

102



UNLEASHING THE POTENTIAL OF ELECTION DATA


Š 2015 National Democratic Institute. Licensed under Creative Commons Attribution Attribution 4.0 International ISBN 978-0-9910142-2-4 Original version: English


TABLE OF CONTENTS INTRODUCTION AND TARGET AUDIENCES

5

SECTION 1 ELECTORAL INTEGRITY

9

WHY DO WE CARE ABOUT ELECTIONS?

10

WHAT ARE CREDIBLE ELECTIONS?

11

ELECTIONS ARE A PROCESS

13

PUBLIC CONFIDENCE, TRANSPARENCY, AND OPEN DATA

15

THE ROLE OF ELECTION OBSERVATION

17

SECTION 2 OPEN ELECTION DATA PRINCIPLES

19

INTRODUCTION

20

PRINCIPLE 1: TIMELY

22

PRINCIPLE 2: GRANULAR

24

PRINCIPLE 3: AVAILABLE FOR FREE ON THE INTERNET

26

PRINCIPLE 4: COMPLETE AND IN BULK

27

PRINCIPLE 5: ANALYZABLE

29

PRINCIPLE 6: NON-PROPRIETARY

32

PRINCIPLE 7: NON-DISCRIMINATORY

33

PRINCIPLE 8: LICENSE-FREE

34

PRINCIPLE 9: PERMANENTLY AVAILABLE

36

SECTION 3 KEY ELECTION PROCESS CATEGORIES

37

INTRODUCTION

38

LEGAL FRAMEWORK FOR ELECTIONS

39


ELECTORAL BOUNDARIES (DISTRICTS)

41

ELECTION MANAGEMENT BODY AND ADMINISTRATION

43

ELECTION MANAGEMENT BODY PROCESSES

46

ELECTION SECURITY

50

POLITICAL PARTY REGISTRATION

53

BALLOT QUALIFICATION

55

ELECTION CAMPAIGNS

57

CAMPAIGN FINANCE

59

VOTER REGISTRATION

65

VOTER LISTS

67

VOTER EDUCATION

69

POLLING STATIONS

71

ELECTION RESULTS

73

ELECTRONIC VOTING AND COUNTING

77

ELECTORAL COMPLAINTS, DISPUTES AND RESOLUTION

79

SECTION 4 OPEN ELECTION DATA IN PRACTICE: EXAMPLES FROM LATIN AMERICA INTRODUCTION

84

ARGENTINA

85

BRAZIL

88

CHILE

91

COSTA RICA

93

MEXICO

95

PERU

99

ACKNOWLEDGMENTS

4

83

100


INTRODUCTION AND TARGET AUDIENCES Citizens, technologists and civil society movements across the globe are demanding more openness in government, including government data, so that they can ensure public institutions are more transparent, responsive and accountable. In addition, electoral management bodies (EMBs) around the world are making public more electoral data. The National Democratic Institute (NDI or the Institute) has a keen interest in ensuring that key electoral data is available and in helping stakeholders use that data to enhance the integrity of elections and hold electoral management bodies (EMBs) and other electoral actors to account. NDI plans to leverage its capacity, experience and long-standing relationships with citizen observer groups around the globe to help them use this document to access electoral data, evaluate the integrity of elections and advance political and governmental accountability. NDI is the leading organization worldwide in providing technical assistance to nonpartisan election monitoring organizations, assisting over 400 groups in more than 95 countries to monitor more than 300 elections and to mobilize more than three million observers. The first section of the guide, Electoral Integrity, lays out the four guiding elements – transparency, accountability, inclusiveness and competitiveness – necessary for elections to have integrity. The second section, Open Election Data Principles, outlines the nine principles that make electoral data open (i.e., timely, finely grained, available for free on the internet, complete and in bulk, analyzable, non-proprietary, non-discriminatory, license-free, and permanently available. The third section outlines the major phases or Key Categories in Elections (e.g., Election Management Body Administration, Ballot Qualification, Campaign Finance, Voter Registration, Election Results, and Complaints and Disputes Resolution) as well as examples of the type of data that may be released for each category. The fourth section, Open Election Data in Practice: Examples from Latin America, highlights six countries as examples of how relevant, granular data is being provided in open, analyzable and bulk formats – in other words, they meet many of the open election data principles. The countries are Argentina, Brazil, Chile, Costa Rica, Mexico and Peru. In many cases, it is not the voters who are directly using the electoral data but rather election monitoring organizations, civic groups, political parties, and the media who are (re)using or repurposing the data. These types of organizations are sometimes called “information intermediaries.” These organizations may use their domain area knowledge to analyze the data and present the findings in a more digestible format for citizens. Or they may (re)mix the data and present it in a form that is easier for an individual to use. For example, the Voter Information Project1 in the United States re-uses data on the coordinates of polling place locations and combines it with data on electoral boundaries to provide voters with a way to find their specific polling place by entering their home address. 1  www.votinginfoproject.org

5


Potential Audiences and Uses This guide has several potential audiences and uses, for: • those in the media who may be new to elections and who are looking to get a better understanding of the election process and what kinds of data are produced during each phase of the election process; • election monitoring organizations, civil society groups, or political parties who want to learn more about open data principles such as analyzable, finely-grained, or license-free and/or how they apply to the data released during elections; • civil society organizations or media that want a framework for evaluating how “open” their country’s election data truly is; • civil society organizations that may want to advocate for making election data more open and want examples of how data has already been made more open; and • for electoral management bodies (EMBs) responsible for publishing election data that want clear, concrete definitions of open data. Resources on the process of opening data (“how”) This guide focuses on defining the key phases in an election process and outlines principles for how electoral data should be open to the public. Throughout the guide there are examples where EMBs have provided electoral data in varying degrees of openness. However, this guide does not cover the exact steps in the process of making data open because resources already exist on the topic of implementation. For guidance on the ‘how’ of opening data, please refer to the “How to Open Up Data”2 section of the Open Knowledge Foundation’s Open Data Handbook.3 In terms of election data, legislative bodies and EMBs that are making policy, laws, regulations or other guidance should engage with the public, including actual and potential users of electoral data, as early and as fully as possible. For more information, see the following sources. • The UK’s Open Data White Paper4 contains information about designing an engagement strategy and the use of community boards. • The World Bank’s Open Government Data Working Group developed an Open Data Readiness Assessment (ODRA) tool5 that can be used to conduct “an action-oriented assessment of the readiness of a government or individual agency to evaluate, design and implement an Open Data initiative.” • The Open Data Institute (ODI) has a short guide on engaging with reusers.6 • Making Data “License-Free”7 is a website that has example language that federal agencies in the United States can use. The site provides suggested language to make it clear that the publications are “license-free.” 2  opendatahandbook.org/en/how-to-open-up-data/index.html 3  opendatahandbook.org 4    gov.uk/government/uploads/system/uploads/attachment_data/file/78946/CM8353_acc.pdf 5    toolkit.dev.zognet.net/en/odra.html 6  theodi.org/guides/engaging-reusers 7  theunitedstates.io/licensing

6


Timeline of Major Developments in Open Government and Open Data • 2006: Open Knowledge8 (then called the Open Knowledge Foundation or OKF) created their ‘Open Definition’9 where they defined “open” in terms of “open data” and “open content.” They summarize their longer 11-point definition by stating that “a piece of data or content is open if anyone is free to use, reuse, and redistribute it – subject only, at most, to the requirement to attribute and/or share-alike.” It is worth noting that the when the Open Definition was written it was not specifically referring to government-held data; it was an attempt to define “openness” more broadly. • 2007: Tim O’Reilly of O’Reilly Media and Carl Malamud of Public.Resource.Org hosted a meeting of thirty people interested in advocating for openness in government. The gathering was sponsored by Google, the Sunlight Foundation, and Yahoo. The working group attendees created and published “8 Principles of Open Government Data”,10 marking one of the first attempts to apply open data principles to the area of government accountability. • 2011: The Open Government Partnership (OGP)11 was launched. The OGP is a multilateral initiative that includes both government and civil society participation so that governments become sustainably more transparent, accountable and responsive to their own citizens, with the ultimate goal of improving the quality of governance. Transparency & Accountability Initiative published the Open Government Guide, which was updated for the 2013 London OGP Summit; the Guide includes a chapter on elections. • 2012: Open Government Data: The Book12 was published by Joshua Tauberer, a civic hacker who founded govtrack.us, and co-organizes Open Data Day DC.13 • 2012: The Open Data Institute14 was founded by Sir Tim Berners-Lee15 and Professor Nigel Shadbolt16 to examine how open data creates larger economic, environmental, and social value. • 2012: The Sunlight Foundation published the Open Data Policy Guidelines17 to address what data should be public, how to make data public, and how to implement policy. • 2013: Leaders from Canada, France, Germany, Italy, Japan, Russia, USA and UK met for the G8 Summit. Transparency was one of three main issues taken up by the governments. As a result, the G8 governments issued and signed the Open Data Charter18 where they agreed that “open data are an untapped resource with huge potential to encourage the building of stronger, more interconnected societies that better meet the needs of our citizens and allow innovation 8    okfn.org 9    opendefinition.org 10    public.resource.org/8_principles.html 11  opengovpartnership.org 12  opengovdata.io 13  dc.opendataday.org 14  opendatainstitute.org 15    www.w3.org/People/Berners-Lee 16  users.ecs.soton.ac.uk/nrs 17  sunlightfoundation.com/opendataguidelines 18    gov.uk/government/publications/open-data-charter

7


and prosperity to flourish.” In the Charter, the governments committed to developing open data action plans by the end of 2014 with the aim to implement the charter by the end of 2015 at the latest. • 2014: The second edition of Open Government Data: The Book19 was published. • 2014: The Open Knowledge Foundation (OKF) published the Global Open Data Index20, which attempts to score a government’s openness of data in areas such as budgeting, spending, national statistics, and legislation.

19    opengovdata.io 20  index.okfn.org

8


SECTION 1 ELECTORAL INTEGRITY

9


WHY DO WE CARE ABOUT ELECTIONS?

Democracy requires more than elections, but a country cannot be a democracy without holding genuine elections. They are the basis for democratic legitimacy. Elections give citizens a means to hold their leaders accountable by voting incumbents out of office or promising to hold to account those leaders successfully elected. They are a mechanism to peacefully resolve the transfer of political power. Elections are also central to making democracy deliver a better quality of life by linking voters’ interests to those in government and allowing citizens to select representatives who reflect their will. Genuine elections contribute to longer-term development goals, laying the foundation for responsive governance. “When the electorate believes that elections have been free and fair, they can be a powerful catalyst for better governance, greater security and human development. But in the absence of credible elections, citizens have no recourse to peaceful political change.” Kofi Annan Deepening Democracy: A Strategy for Improving the Integrity of Elections Worldwide21, a report by the Global Commission on Elections, Democracy and Security.22

Citizens have a fundamental right to take part in government and the conduct of public affairs of their countries. They have a right to vote and to be elected in periodic elections, without discrimination. The will of the people, expressed through their ballots, provides the basis of authority for democratic government. These concepts are enshrined in Article 21 of the Universal Declaration of Human Rights23 (Universal Declaration), Article 25 of the International Covenant on Civil and Political Rights 24(ICCPR) and other international human rights instruments. They are central to the principle that elections belong to the people and are key features of any genuine democracy. 21  idea.int/news/deepening-democracy-a-strategy-for-improving-the-integrity-of-elections-worldwide.cfm 22    kofiannanfoundation.org/our-work/global-commission-elections-democracy-security 23  un.org/en/documents/udhr/index.shtml 24  ohchr.org/en/professionalinterest/pages/ccpr.aspx

10


WHAT ARE CREDIBLE ELECTIONS?

Credible elections are characterized by inclusiveness, transparency, accountability, and competitiveness.25 Inclusive elections provide equal opportunities for all eligible citizens to participate as voters in selecting their representatives and as candidates for election to government. This right to participate is a broad concept and can only be subject to reasonable restrictions that are provided by law. Authorities must take steps to assess and remove any barriers to the participation of all citizens. This includes removing barriers for traditionally marginalized populations such as ethnic, religious or linguistic minorities; women, youth or elderly citizens; and persons with disabilities. Reasonable restrictions on participation should be narrow, like a minimum age for voting and standing as an electoral contestant. Elections are transparent when each step is open to scrutiny, and stakeholders can independently verify whether the process is conducted honestly and accurately. The principle of transparency is linked to the fundamental right of citizens to seek, receive and impart information (which are elements of the freedom of expression), as well as the right to take part in government and public affairs.26 Decision making processes must be open to scrutiny, and reasonable opportunities for public input should be provided. Information relating to all stages of the electoral cycle must be made available and accessible to citizens, including voters and candidates. Nonpartisan and partisan observers should be accredited to observe all phases of the election process and be permitted to comment publicly on the process free from unreasonable restriction. Accountability in elections refers to the rights of citizens with respect to the conduct of other electoral stakeholders, including the government, election management bodies (EMBs), 25    For a more detailed discussion of these principles, see Promoting Legal Frameworks for Democratic Elections: An NDI Guide for Developing Laws and Law Commentaries, Section Two (2008), also published as Chapter One in International Election Principles: Democracy & the Rule of Law (JH Young, ed., American Bar Association (ABA), 2009). 26    Merloe, Patrick. “Human Rights – The Basis for Inclusiveness, Transparency, Accountability and Public Confidence in Elections,” in International Principles: Democracy & the Rule of Law (JH Young, ed., ABA, 2009), p. 3, 18-20.

11


political parties, candidates and security forces. Elections are a key mechanism through which citizens hold their governments accountable, but there must also be accountability within election processes themselves. There must be effective remedies in place for violations of citizens’ election-related rights. There must also be administrative accountability for those organizing elections and those conducting governmental activities related to elections. In addition, there must be timely procedures to bring to account those who conduct criminal acts that affect electoral-related rights. Elections are competitive when citizens have reasonable and equitable opportunities to compete to be elected to governmental offices. Political competition is a central component of elections that truly reflect the will of the people. The principle of competitiveness relates to aspects of the election throughout the electoral cycle. The legal framework must allow citizens to come together and register political parties to represent their interests, and provide access to spots on the ballot for parties and/or candidates. Parties and candidates must be able to campaign and voters to cast their ballots free from illegal influence, intimidation or violence.

12


ELECTIONS ARE A PROCESS

Elections are a process, not an event. Every election comprises numerous elements and involves multiple institutions and actors throughout the pre-election, election day and post-election periods, all of which affect the transparency, inclusiveness, accountability and competitiveness of the election. In the pre-election period, legislatures and state institutions develop or refine the legal and procedural framework in which the elections are conducted. EMBs conduct a range of activities in preparation for election day and determining the electoral outcome. Depending on the type of election and electoral system, authorities may define electoral boundaries. EMBs or other state institutions register political parties and determine which parties and/ or candidates qualify for a place on the ballot. They also accredit international and citizen (domestic) observers, candidate and party agents, and media to witness electoral procedures. EMBs or other responsible institutions typically undertake a voter registration process to make sure all eligible citizens are on the voter list. Voters may need to actively take steps to register, and in some cases have opportunities to correct or update information on the list. Meanwhile, political parties and candidates campaign to attract support for their political platforms. EMBs and civil society organizations conduct voter education so that citizens understand election procedures and can make informed choices in the voting booth. EMBs also conduct the administrative and logistical preparations for polling, including procuring and delivering materials, as well as recruiting and training pollworkers. On election day, eligible voters cast their ballots at polling stations. Election officials facilitate voting and, when voting ends, conduct the ballot counting process. Accredited observers from political parties and candidates, nonpartisan civil society groups and international organizations, and media often monitor the voting and counting processes. The vote counts are sent to tabulation centers and ultimately, to a final vote tabulation center, where initial or preliminary results are processed. Electoral disputes and complaints concerning various

13


processes before and on election day are to be resolved through peaceful, legal means by using EMB complaint procedures and the courts. It often takes several days or sometimes weeks for EMBs to release final, verified results. The post-election period provides an opportunity for stakeholders to evaluate the process and undertake necessary reforms. The EMB and civic groups evaluate the election process and define ways to reform and strengthen elements within each phase. This critical time between elections is also used to prepare for future elections, including updating voter lists and addressing issues with the legal framework or institutional management.

ELECTIONS ARE A CYCLE

BA L

SEC U

VO

C

E

V

P

POLI TIC

CTION ELE MA

LE

EL EC

TION MAN EC A EL

COMPLA IN

EL

E

GN FINAN PAI C AM

N

ES CESS PRO N TIO EC EL

S

ONIC VOT TR IN EC

EGISTRAT RR IO TE

OF ELECTIO TY N RI

G MPAI NING CA

TER LISTS VO

EDUCATI ON ER OT

Y DA

ION RESUL CT T LE

G

14

N O

G STATIO LIN NS OL

UALIFICA TQ TI LO

TY REGI ST PAR AL

N TIO RA

TION LU SO

POS T-E LEC TIO N

A

MENT BO GE DY NA SS OCE ES PR

BODY &

TRATIO INIS N DM

NT ME GE

ISPUTES, R ,D E TS

BOUND RAL AR TO S IE

S SE

RAMEWO LF RK GA

S

PR OC ES

ELECTION PROCESSES PRE-


PUBLIC CONFIDENCE, TRANSPARENCY, AND OPEN DATA

Public confidence in each step of an election process is critical to the integrity of the election. Citizens not only have a right to participate in elections, they have a right to know for themselves whether the electoral process is valid. Access to information about each phase of the election process is fundamental to creating and reinforcing public confidence in elections. That knowledge is the basis for public confidence in elections and their resulting governments. The concepts of transparency and trust are intrinsically linked when it comes to public institutions. Public institutions collect vast amounts of data. The data collected provides a foundation for decision making processes at both the policy and project levels. Trust in decision makers requires openness about the factual evidence on which the decisions were made. In addition, most of the data collected by governments should by law be public data –­ and “public data” should be open to, and shared with, the public. Indeed, just as elections and government belong to the people, public data belongs to the people. And, when public institutions such as election management bodies (EMBs) open up data, they demonstrate transparency, which can significantly increase public confidence in the election process. In some contexts, other bodies, such as legislatures, might be involved in decision-making about government data openness. For example, in 2008, the Parliament of Georgia in consultation with the Central Election Commission (CEC) passed a law27 requiring the immediate publication of preliminary results for all polling stations on the CEC website. Other institutions, such as agencies holding census data or other information that is important for assessing the credibility of election processes, may affect public confidence in elections through­ their open data practices. In order for citizens to participate in, understand, evaluate and, ultimately, accept an election process and its outcome as representing their will, election data must be open to citizens. As described in the section on open data principles, election data is open when it is released in 27

Article 76, paragraph 8 of the Unified Election Code of Georgia.

15


a manner that is timely, granular, available for free on the internet, complete and in bulk, analyzable, non-proprietary, non-discriminatory and available to anyone, license-free and permanent. When these conditions are met, citizens can, among other things, use election data to participate more fully as voters or contestants, make informed choices about whom to elect, understand and accept how winners are determined, recommend reforms to improve the next election and hold electoral stakeholders accountable. In these ways, open election data enhances the inclusiveness, transparency, accountability and competitiveness of elections. As information communications technologies (ICTs) continue to rapidly advance, the ability to make election data publicly available improves substantially, while the burdens for doing so diminish. This changes the public policy – making calculus in favor of open electoral data practices. Increasingly, countries are releasing different types of election data. Data relating to electoral boundaries, voter registration, polling station lists and polling station results are among the most common. In Latin America, most countries make significant amounts of election data available to the public, often in open formats (see Section 4: Open Election Data in Practice: Examples from Latin America). EMBs around the world should take steps to provide more open election data as a meaningful way to make their elections more inclusive, transparent, accountable and competitive.

16


THE ROLE OF ELECTION OBSERVATION

Over the last three decades, a global movement of nonpartisan citizen election monitoring has collectively mobilized millions of citizens around the world to participate in their countries’ democratic processes as election observers. When citizens come together to observe elections, they may have a number of goals. Depending on the context, civil society might mobilize to observe an election in order to: engage citizens in the election process; deter fraud; expose problems and irregularities; provide an accurate measure of the quality of the election; promote confidence in the process and outcomes; and provide recommendations for improving the process for the next election. In the longer term, citizen election observation helps ensure electoral integrity, and strengthens civil society’s capacity to promote citizen participation, engage in policy advocacy and foster governmental accountability within and well beyond the election cycle. More than 200 citizen election observation organizations in over 80 countries make up the Global Network of Domestic Election Monitors (GNDEM)28, formed in 2009. GNDEM is united by the Declaration of Global Principles for Nonpartisan Election Observation and Monitoring by Citizen Organizations29, which was launched at the United Nations in 2012 and has been endorsed by more than 245 citizen observer groups. The Declaration of Global Principles defines the rationale, rights and responsibilities of nonpartisan citizen election observers.30 GNDEM unites regional networks of citizen election monitors from five continents as well as national organizations and coalitions. The global and regional networks are a consequence of cross-border relationships developed over more than 25 years with the assistance of NDI and others. These networks overlap with other participation, accountability, anti-corruption and open government efforts, which can advance open data principles and practice. 28    gndem.org 29    gndem.org/declaration-of-global-principles 30    NDI has published numerous handbooks and manuals on various aspects and methodologies of citizen election observation, available at https://www.ndi.org/elections.

17


no stollab rieht tsac ,etov ot retsiger ot woh tuoba sretov etacude pleh ot noitamrofni noitcele .stnatsetnoc gnoma seciohc demrofni ekam dna yad noitcele

Citizen monitors are well-positioned to evaluate a range of aspects of the election process if relevant data is effectively available. Equipped with open election data, observer groups can better promote confidence in the process, deter fraud, provide an accurate assessment of the election or achieve other goals. With access to polling station data, for example, some groups use statistics to deploy observers to a random, representative sample of polling stations, giving them a highly accurate assessment of election day procedures. These efforts can contribute significantly to promoting public confidence in an election and calming tensions in the post-election period. Other groups, with access to the voter list, conduct analysis to verify accuracy of the list and highlight ways that voter registration – and the list itself – could be improved for the upcoming and/or next election. Observers also often use election information to help educate voters about how to register to vote, cast their ballots on election day and make informed choices among contestants.

International election observation can assess whether an election is in line with international standards, highlight specific recommendations for improvement, and demonstrate the support of the international community for genuine democratic elections. Where deserved, international election observation can also promote public confidence in an election. The legal and ethical bases as well as basic methods of international observation are defined in the Declaration of Principles for International Election Observation31, developed in a multiorganization process initiated by the United National Electoral Assistance Division, NDI and The Carter Center. The Declaration of Principles, launched at the United Nations in 2005, has been formally endorsed by more than 45 of the leading international observation organizations and recognized with appreciation by the UN General Assembly for contributing to harmonizing the practice of international observation. International election observers can play an important role in promoting open electoral data principles and practices as a means of advancing public confidence in genuine democratic elections. While the varying conditions of specific countries may be more or less conducive, international election observers can comment on the transparency of electoral practices and would benefit from access to electoral data and analysis or open electoral data by various credible sources.

31  ndi.org/declaration_monitoring_principles

18


SECTION 2 OPEN ELECTION DATA PRINCIPLES

19


INTRODUCTION Many organizations collect data to perform their functions or to provide a service. Public institutions, such as election management bodies (EMBs), produce and collect large amounts of data. In many countries, the data collected by these public institutions is, by law, public data. It makes sense that such public data should effectively be shared with the public. In addition, data collected is often the basis for decision making as well as implementation. By sharing that data, a public institution is demonstrating its intent to be transparent about its decision making and implementation processes. Citizens and organizations can also use that data to hold public institutions accountable. Making data ‘open’ refers to sharing the data in ways that make the data freely and easily used, reused, and redistributed by members of the general public. As described in this section, ‘open’ election data is: • timely, • granular, • available for free on the internet, • complete and in bulk, • analyzable, • non-proprietary, • non-discriminatory, • license-free, and • permanently available. Open data, especially open election data, is an immense resource that has only recently been tapped. Even though the ‘open data’ field is young, there are already numerous examples of the ways in which it has already created both social and economic value around the world. When election data is “opened up,” it can: • Increase the transparency of individual election processes when, for example, citizen election monitoring organizations combine it with their monitoring efforts; • Improve the effectiveness of an election management body when, for example, it allows voters to check those registered at their address or when the images of results forms are released and citizens can check the transcription of results; • Increase voter participation in elections when, for example, citizen organizations use registration rates to help historically marginalized populations register to vote or when political parties use the information in voter outreach; • Increase citizen engagement when, for example, the results of individual polling stations are released and voters can check the results of their station and compare it to nearby stations; • Improve the inclusiveness of traditionally marginalized groups when, for example, citizens organizations use information to advocate for the accessibility of polling station or for ballots in local languages;


• Reduce tension when, for example, election monitoring organizations combine results data with their assessment of the election day process and make early projections of the results; • Generate new insights when, for example, citizen organizations combine results data with information on the location of political violence. There is still much untapped potential in making election data more open. The following open election data principles provide a way of unleashing that potential.

21


PRINCIPLES

PRINCIPLE 1: TIMELY

TIMELY

G

Election data should be made available as quickly as necessary for it to be useful. Timely publication of election data is essential because it feeds into several decision making and participatory processes. Citizens need information on how to register, which parties or candidates are running and where to vote so they can participate in the process and make informed decisions. Potential contestants need information on ballot qualification so they can compete in the election. Contestants need information on media access and allocation so they can get their message out. The timely publication of election data is also an essential NON-PROPRIETARY FORMAT component of transparency. Election monitoring organizations, political parties, and the media need election data to assess the transparency of a particular process and whether it is

NON-DISC AVAILAB USE WITH

accurate and fraud free. Since the phases that make up the election cycle32 are interdependent and build on one another, the timely release of data takes on a greater importance in elections than some other subject areas. The timely release of data often means the data must be frequently updated, as happens during voter registration and when preliminary polling station results are released. Users should be able to quickly identify and locate the updates. For example, if the results for a

CATEGORIES

polling station are changed and a new, corrected form issued, then the data should explicitly

note the change(s), why the changes were made (e.g., small mathematical errors were

corrected), and when they were made (e.g., a timestamp of when last updated). Releasing data in a timely manner is a way the election management body (EMB) demonstrates that the process is trustworthy.

32  idea.int/elections/eea/images/Electoral-cycle.png

ELECTORAL BOUNDARIES

22

ELECTI BODY &


How does data timeliness play out during the election cycle? During the planning and implementation phase, it is important that the EMB release data related to the budget so that its implementation can be properly monitored. Data about who is running the election at which level is necessary so they can be contacted with questions and held accountable for their part in the process. Timely information on procurement, especially if electronic technologies are involved, is needed to evaluate the terms of the contract and whether the contractor can meet the necessary requirements. Often, it is important that the data related to a particular phase is released before that phase is even finished. Voters need timely information about their registration status and where to register so they can take action before the voter registration phase ends. Parties and issue-based civil society organizations need timely information during the voter registration process so they can conduct registration drives and make sure their supporters are registered and can vote on election day. Data about campaign finance and media allocation must be released quickly to make sure the campaign period is fair. Near real-time publication of provisional polling station results helps provide transparency in the counting process. Data on electoral complaints must be released quickly so it is clear how they may affect results or so that parties can appeal. Releasing data on the progress of tabulation helps election monitoring organizations and the media to evaluate the transparency of the process and contributes to public confidence.

23


PRINCIPLES

PRINCIPLE 2: GRANULAR

TIMELY

GRANULAR

AVAILABL THE

Election data should be available at the finest possible level of granularity or detail. Data that is finely-grained, or full resolution, is not in an aggregate or modified form. This principle has also been expressed as making data available at the “primary” level or the level at which the source data is collected. Granularity is necessary for promoting transparency and accountability. In terms of elections, it is clear that the location, address and contact information for each polling station should be released in advance of election day. Citizens need that information NON-PROPRIETARY FORMAT NON-DISCRIMINATORY AND so they can find out where to vote. What is less obvious is that polling station-level is the AVAILABLE TO ANYONE TO WITHOUT LIMITATIONS most granular level for election results data. Since the votes areUSE cast at polling stations

LICE

(sometimes called precincts, polling streams33 or tables34), polling stations are the primary level for election results. If datasets like election results are only made available in an aggregate or modified form, then that dataset is not open. Indeed, much of the ‘fraud’ that has been committed in elections is easy to conceal if data is not released at the most granular, or primary, level. In terms of the voter registration process, the most granular level is the individual voter.

CATEGORIES

While the primary unit of data collection for voter registration is the individual voter, the level at which the data is released varies widely from country to country. EMBs in some countries post a paper copy of the voter list with all of the voter details (first name, last name, address, national ID number) outside of the polling station. Posting a voter list for a specific polling station is granular but the dataset is clearly not complete because it is one part of the entire list for the whole country. In other countries, the full list of voters is 33    In some African countries such as Malawi and Zambia, the lowest level is called the polling stream and one level up is called the polling station. When we say “polling station” in this document, we mean the lowest level at which the results are cast and counted. 34    In many Latin American countries, the lowest level is called the ‘mesa’ or ‘table.’ A voting center (a building) will usually contain several “mesas” (in several rooms within the building). Again, when we use the term “polling station” we mean it is as the lowest level for results and it is interchangeable with the concept of a mesa or table.

ELECTORAL BOUNDARIES

24

ELECTION MANAGEMENT BODY & ADMINISTRATION

ELECTI BO


available on the EMB’s website as multiple files. In that case, the data can be analyzed, for example, to note the number and percentage of women on the list or to see whether other historically marginalized groups are included. Still some other countries state concerns over voter privacy and instead of publishing a voter list, allow voters to lookup their individual details. IfQuestions the legaltoframework places restrictions onisthe of election help determine what level of data thegranularity primary or source level: data (e.g., due

to privacy concerns), then the rationale for theThis restriction should be clearly stated, and the - Where was the data originally collected? might give an indication as to the lowest level. least restrictive means should be used to protect the legitimate interest at issue. - What is the lowest level possible where an individual data point might vary from another data

point? field of election forensics uses different statistical methods35 to analyze and The emerging - Is it data related to voting? If so, then polling station is likely the (e.g., lowestpolling level. stations). identify suspicious patterns in data published at the primary level - It is the dataplaces relatedused to qualifying to be(e.g., listed administrative on the ballot or voter registration? If so,markets) then ‘indiAdditionally, for polling buildings, schools, (voters or voters’ is likely the lowestand level. are oftenviduals’ the places where votersignatures) registration is conducted complaints are lodged. Thus, access to data at the most granular level is key for citizens and other stakeholders to assess the transparency of most parts of the election process.

Questions to help determine what level of data is the primary or source level: • Where was the data originally collected? This might give an indication as to the lowest level. • What is the lowest level possible where an individual data point might vary from another data point? • Is it data related to voting? If so, then polling station is likely the lowest level. • It is data related to qualifying to be listed on the ballot or to be registered to vote? If so, then ‘individuals’ (voters or voters’ signatures) is likely the lowest level.

Leaders of the G8 governments explicitly acknowledged the importance of this principle when they signed the G8 Open Data Charter (and Technical Annex)36 in June of 2013. In the Open Data Charter, the governments of Canada, France, Germany, Italy, Japan, Russia, the United States and the United Kingdom commit to releasing high-quality open data and state that the data will be “in their original, unmodified form and at the finest level of granularity available.”

35    For example, various digit tests and outlier detection methods have been used to cast suspicion on election results in places such as Russia, Iran, and Ukraine. 36    gov.uk/government/publications/open-data-charter/g8-open-data-charter-and-technicalannex#principle-2-quality-and-quantity

25


82


SECTION 4 OPEN ELECTION DATA IN PRACTICE: EXAMPLES FROM LATIN AMERICA

83


INTRODUCTION The open election data principles outlined in Section 2 are meant to provide clear, concrete guidance to those responsible for publishing election data, as well as a framework by which election observers, political parties, the media and other oversight actors can evaluate how “open” a given country’s election data truly is. In this section, NDI moves from principles to reality by applying the open election data principles to existing election data in South and Central America. The Institute chose South and Central America because, while anecdotal information has suggested that election management bodies (EMBs) in some countries of the region release election data in a variety of categories in an open manner, these efforts have not been assessed systematically. As a result, potentially valuable lessons and examples may not be captured and drawn from within and outside the region. In addition, South and Central America provide a diverse set of countries in which to assess how well the open election data principles in Section 2 of this report are adhered to in practice. NDI analyzed which categories of election data are available online for the most recent elections in 18 countries of South and Central America.105 For each of the countries, NDI assessed what data was available online in the categories described in Section 3. For each set of data that is available, the Institute then evaluated the data based on each of the open election data principles. Among the 18 countries, NDI has chosen six countries to highlight through in-depth examples of how relevant, granular data is being provided in open, analyzable and bulk formats – in other words, it meets many of the open election data principles. The six countries are Argentina, Brazil, Chile, Costa Rica, Mexico and Peru. There are other countries that could have also been highlighted, but NDI has selected these six countries because the data they provide in several different categories stand out as meeting many of the open election data principles. NDI hopes that EMBs, civil society groups, political parties, media, international organizations and other stakeholders focused on making election data more open can draw lessons from the examples we have highlighted below.

105    NDI did not assess data in Belize, French Guiana and Suriname.


ARGENTINA ELECTION CYCLE ASSESSED: October 27, 2013 Legislative Elections (national and local levels)

85


Election management body: Argentina provides an important example of open data on election officials at the polling station level (pollworkers). The data is provided in a granular way, down to the individual names and positions of polling officials at the polling station (mesa) level. The complete data is provided in bulk and is available for download in a non-proprietary (PDF), non-discriminatory and license-free format. The main drawback, however, is that it is not analyzable, since it is provided in PDF. Political party campaign finance: Argentina’s political party campaign finance data is provided in a user-friendly and interactive format. It also provides the user with a variety of ways to download (in an analyzable format) different sets of data using filters, as well as visualization features (Figure 1 below). However, it is not granular. While it is admirable that data is provided down to the provincial level by each political party, the expenses and income are only provided at the level of major types of income and expenses. It does not provide any detailed line item data or individual expense/income data. Furthermore, the data is not provided in bulk and is through a proprietary software (Excel Online via Microsoft’s OneDrive).­

Figure 1 Argentina – Campaign Expenditure Data, Mendoza Province Data can be downloaded in Excel or accessed in more granular detail (by each individual party among coalitions or “fronts”).

86


Election results: Argentina’s preliminary election results106 are a particularly good example of data that meets nearly all open election data principles. The data is provided through Argentina’s Portal of Public Data, which is a user-friendly platform. Notably, the platform is non-proprietary (see open election data principle six), utilizing an Open Data Commons Open Database License ODbL. Argentina provides preliminary election results data for download in an analyzable, non-proprietary format (CSV). The results are granular (provided down to the polling station level) and are available to download for each election type (national or subnational levels). In addition, unlike most countries, the complete dataset for the results for all types of elections held on October 27, 2013 if available in a bulk download in CSV format. Final election results107, however, are provided in a much less open manner through a different website. They are only available for download in PDF (not analyzable), and the data is provided only down to the provincial level.

106    datospublicos.gob.ar/data/dataset/a4e05b80-6331-4a6f-a110-35ebc29e8ffc 107    elecciones.gov.ar/resultados-definitivos.php#.U_IRpPmSxDB

87


BRAZIL ELECTION CYCLE ASSESSED: October 5, 2014 general elections: President, National Congress, state governors and state legislatures.

88


Data standards: Brazil’s election data has the best use of timestamps compared to other countries in the region. Most data is time-stamped, as well as every website page. Data in many cases is also documented well. Election management body: Brazil provides user-friendly and detailed information on the officials and staff108 of its electoral management body, the Tribunal Superior Electoral (TSE). The information includes an interactive organizational/staffing chart with hyperlinks to (and a full list of) the names, positions, individual email addresses and phone numbers of staff. Voter education - Candidate information and status: Brazil’s “DivulgaCand” platform109 provides a significant amount of data on each electoral candidate. It provides a user-friendly, interactive way to gain information about each candidate’s platform, status and official documents submitted to the TSE. While the provision of official candidate documents is admirable, they are not analyzable, as they are only provided in scanned image files (PDF). In addition to individual candidate data, complete lists of candidates for all elections are available for download in a non-proprietary, analyzable format (CSV). There is no bulk download for all candidates in all elections, however. Political party registration: Brazil’s data on political party membership110 stands out when compared to most countries in Latin America. The TSE provides granular data on all individual current and former party members for each party at the state level. The data can be downloaded in a non-proprietary, analyzable format (CSV). While very few countries provide individual member data at all, Brazil goes a step further by providing a rich set of data on each member. This includes date of registration, voter ID number and geographic location by state, city, electoral zone and electoral section. For former party members, it includes date of and reason for cancellation, as well as the date of renewal/reinstatement for those who have rejoined parties. Each file downloaded is accompanied by documentation that explains the data coding scheme. Brazil’s party membership data meets open election data principles. Polling stations: Brazil’s polling station data meets most open election data principles. The TSE provides a complete list of all polling stations111 for bulk download in a non-proprietary format (PDF), but unfortunately it is not analyzable. The list includes several geographic levels, as well as the unique identifier and address of each polling station. Election results: Brazil provides three different options for obtaining access to election results with different levels of detail and different interactive features. Each of them has different positives and drawbacks when considering open election data principles, as described below. • Summary-level results: Brazil’s DivWeb platform112 provides an interactive way to access election results, but it is not granular data, as it only provides vote totals by race (e.g., national-level elections only have national-level results). The data is available in a non108    tse.jus.br/institucional/o-tse/organograma-tse#PRES 109    tse.jus.br/eleicoes/eleicoes-2014/sistema-de-divulgacao-de-candidaturas 110    tre-sp.jus.br/partidos/filiacao-partidaria/filiacao-partidaria 111  justicaeleitoral.jus.br/arquivos/tre-pa-locais-de-votacao-do-estado-do-para-para-as-eleicoes-2014 112    divulga.tse.jus.br/oficial/index.html

89


proprietary format (HTML on webpages), but it is not analyzable. Also, as of December 19, 2014, the platform does not function properly and does not provide results for the October 2014 elections. • Primary-level results: Brazil’s Divulga platform provides access to data at a more granular level and can be queried in a variety of ways. The platform also requires proprietary software (Java). Data is provided in HTML tables but can also be downloaded in a nonproprietary, analyzable format (CSV). Polling station-level data can be downloaded, but not in bulk. Instead, it can be downloaded by each candidate in each race (see Figure 2 below) • Historical results: Brazil’s Repositório platform provides open election results data back to 1945 available for download in a non-proprietary, analyzable format (CSV). In most cases preliminary and final results are available. The data is more granular in more recent elections. Because file sizes are so large, they are split into groups, but even then it can take a significant time to download for slower internet speeds.

Figure 2 Brazil - Downloadable Polling Station-level Results for Each Candidate Users can select any candidate from any 2014 election and download a CSV file with the results from all polling stations for that particular candidate (apps.tre-ms.jus.br/SDR2014).

90


CHILE ELECTION CYCLE ASSESSED: November 17, 2013 Presidential, Parliamentary and Regional Councillors.

91


Party/candidate registration: Chile provides a good example of data on candidate registration status.113 In addition to providing a complete list of all candidates’ status, for regional councillor candidates it provides data on the justification if candidates were rejected. This provision of data is extremely rare within the region and across the world. The justification data is provided in a non-proprietary format (PDF); however, it is not analyzable (image file). Election results: Chile provides an exemplary model for making its final election results open. Final results114 are easy to locate (discoverable) and are available at the primary (polling station) level. They can be accessed by complete, bulk download in analyzable but proprietary format (Excel).

113    To access data, click on “Resoluciones de aceptación y rechazo de candidaturas” on left sidebar. presidenciales.servel.cl/ss/Satellite?c=Page&cid=1349388530591&pagename=SERVEL%2FPage%2FS_ ListadoElectoral 114  presidenciales.servel.cl/Resultados/index.html

92


COSTA RICA ELECTION CYCLE ASSESSED: February 2, 2014 Legislative, Presidential and Gubernatorial Elections and Presidential Second Round on April 6, 2014.

93


Electoral management body processes - Budget and expenditures: Costa Rica’s Superior Tribunal of Elections (TSE) provides detailed, granular budget115 and expenditure116 data by year. Data is provided in various levels of granularity (summary level and detailed line item level) in a non-proprietary but not analyzable format (PDF). Particularly notable is that the TSE provides data that tracks budgeted amounts versus expenditures, which is rare among countries in the region. Electoral management body processes - election observation organizations: Costa Rica is one of, if not the best example in Latin America of providing open data on election observation organizations.117 Regarding international observers, Costa Rica provides the total number and name of each accredited international observation organization, as well as the number of observers for which each organization requested (which is rarely provided by other countries) and received accreditation. One drawback is that the data is only provided in HTML tables, so it is not analyzable. Polling stations: Costa Rica provides a complete list of all polling stations118 for bulk download in a non-proprietary – but not analyzable – format (PDF). The list includes several geographic levels, as well as the unique code and address of each polling station. Electoral complaints, disputes and resolutions: Costa Rica has developed three ways of obtaining data on electoral complaints and resolutions, which, taken as a whole, help make this data open. • Users can search for and obtain the official resolution document in PDF for individual cases (primary level data) via a search platform.119 • Another web page allows users to download in a non-proprietary format (ZIP) the complete set of resolution documents120 (one document per resolution) for each year. The data is provided at the primary (case) level, but it is not analyzable, since each case document is a PDF file. • A third web page provides a PDF document (non-proprietary but not analyzable) that contain overall statistics on resolutions121 for each year and type of case.

115    tse.go.cr/presupuesto.htm 116    tse.go.cr/info_financiera.htm 117  tse.go.cr/observador_internacional.htm 118    tse.go.cr/pdf/varios/centros_votacion.pdf 119    tse.go.cr/juris/inicio_juris.html 120  tse.go.cr/juris_anual.htm 121  tse.go.cr/pdf/varios/estadisticas_resoluciones.pdf

94


MEXICO ELECTION CYCLE ASSESSED: July 1, 2012 general elections (presidential and legislative)

95


Electoral management body processes – Decision making: Mexico’s National Electoral Institute (INE) has an exceptional level of transparency in its decision making processes122, providing a wide variety of data on its proceedings, meetings, resolutions and agreements. It provides meeting session transcripts, agendas, and recorded audio and video. Live streams of meeting sessions are also available. Party registration information: Mexico’s data on party registration meets many of the open election data principles. It provides granular data on each party’s status and the relevant resolutions regarding party registration decisions.123 Mexico also provides a list of all deregistered parties124 and, unlike most countries, a justification of why each was deregistered. Deregistration data is provided in bulk, although it is only made available in an HTML table and, therefore, is less analyzable than if it were provided in a machine-readable format. Election turnout and results: Mexico’s turnout and election results data is some of the most open in the world. Both preliminary and final election results and turnout data can be downloaded in non-proprietary, analyzable formats at the primary (polling station, or casilla) level. In addition, Mexico publishes image files of handwritten preliminary and final tally sheets from each polling station, which provides users the ability to compare the image files to the digitally-recorded information available for download. Details on each are below. • Preliminary turnout and results data is available at all geographic levels via the interactive, web-based PREP system125, as well as through a downloadable set of compressed, nonproprietary files (gzip) that provides analyzable data at the primary level. The user can also access image files of handwritten polling station-level preliminary results tally sheets. • Final turnout and results data is made available on the web-based ATLAS database126. Data is available at all geographic levels, and polling station-level data can be downloaded in bulk for each of the 32 states in either nonproprietary (plain text) or proprietary (Excel), analyzable format (see Figure 3 below). There is not a bulk download of all nationwide polling-station level data, however. The site also publishes image files of handwritten final tally sheets at the polling station level. • Historical election results data is also available on the ATLAS website, dating back to 1991. Some of the earlier elections do not have polling station-level data, but more recent elections have the same type of final results and turnout data available as the 2012 election data mentioned above.

122    ine.mx/archivos3/portal/historico/contenido/Sesiones_del_CG/ 123    ine.mx/archivos3/portal/historico/contenido/Partidos_politicos_en_formacion/ 124  ine.mx/archivos3/portal/historico/contenido/Historico_sobre_perdida_de_registro/ 125    prep2012.ife.org.mx/prep/introduccion.html 126    siceef.ife.org.mx/pef2012/SICEEF2012.html

96


Figure 3: Mexico – Final Polling Station-level Results and Turnout Data for Mexico’s 2012 Elections Results and turnout figures can be accessed at any geographic level. Tally sheet image files can be accessed for each polling station (“casilla”) by clicking on the green checkmark button next to each polling station data. Data can be downloaded in plain text or Excel on the right side of the page.

Voter education and Election results - External voting: Mexico provides some of the most comprehensive and open data on external voting127 of any country in Latin America. It developed a “microsite”128 that provides useful, user-friendly information on how, when and where to register to vote, as well as how, when and why to vote. The site also provides disaggregated, granular data on external voting results and turnout129, as well as extensive visualization of the data through charts, tables and graphs. This data, however, is only available in non-analyzable format (PDF). 127  votoextranjero.mx 128    votoextranjero.mx 129    votoextranjero.mx/documents/10157/4fe7cd4d-8e67-487a-8b60-26ce0894e874

97


PERU ELECTION CYCLE ASSESSED: 10 April Presidential and Parliamentary Elections, and 5 June 2011 Presidential Election Run-off.


Political party registration: Peru’s Registry of Political Organizations130 provides an extensive amount of party registration-related data and information, including user-friendly information on the laws and procedures for application, as well as a list of all parties, their status and relevant application documents. It also provides summary statistics on party membership, as well as a complete list of the names of each party’s members for download in PDF. The only drawback from an open data perspective is that the data is not in an analyzable format. Campaign finance data for candidates: Peru’s campaign finance data131 stands out as a good example in the region, as it meets many open election data principles. Summary-level data on income and expenses for each candidate is available for download in analyzable but proprietary format (Excel). In addition, granular, individual expense and income data is available for download in Excel for each candidate. One drawback is that individual expense and income data is not available for download in bulk (expenditures are in a separate file from income for each candidate), which makes data analysis more difficult. An example of candidate expense and income data can be found here132.

130    200.37.211.165/srop_publico/6Consulta/con_PadronAfil.aspx 131    web.onpe.gob.pe/modFondosPartidarios/gsfp2011/candidato_menu.php 132    web.onpe.gob.pe/modFondosPartidarios/gsfp2011/candidato_detalle.php?t=p&a=ALIANZA%20POR%20 EL%20GRAN%20CAMBIO&d=AREQUIPA

99


ACKNOWLEDGMENTS This report was written by Michelle Brown, Elana Kreiger-Benson, Michael McNulty, and Samantha Moog with input from Meghan Fenzel and Sunila Chilukuri, and with overall guidance from Patrick Merloe. Jessica White performed extensive research on the data available in Latin American countries. Iulia Fedorovych and Oleksandr Sivchenko designed the style of the report. Individual graphics were created by Iulia Fedorovych. NDI wishes to thank the many electoral management bodies whose perspective and comments have been reflected in this document. NDI also wishes to thank the many citizen election observation groups, including the organizational and regional network members of the Global Network of Domestic Election Monitoring (GNDEM). GNDEM currently includes more than 200 members from 82 countries and regional networks from Africa, Asia, Europe and Eurasia, the Middle East and North Africa, and Latin America and the Caribbean. The first and third sections of this report, on electoral integrity and key election processes, respectively, are based on the extensive experience and work of Patrick Merloe and draw heavily from his handbook, Promoting Legal Frameworks for Democratic Elections. The second section, on open election data principles, owes a great debt to, and draws upon, the solid intellectual foundation set by Open Knowledge Foundation’s Open Definition, the “8 Principles of Open Government Data”133 issued in the 2007 workshop of open government advocates, Joshua Tabeuer’s Open Government Data: The Book134, the Sunlight Foundation’s Open Data Policy Guidelines135 and the examples set by the Open Government Partnership (OGP), Google.org and the Transparency & Accountability Initiative (TAI),136 among others. The development of this report was supported through a grant from Google Inc.137 This report complements other resources available on electoral integrity and election observation, such as NDI’s handbooks: The Quick Count and Election Observation (concerning parallel vote tabulations – PVTs); Building Confidence in the Voter Registration Process; Media Monitoring to Promote Democratic Elections; Monitoring and Mitigating Electoral Violence through Nonpartisan Citizen Election Observation; and How Domestic Organizations Monitor Elections. Additionally, NDI, with support from Google Inc., has developed a series of tutorials to help election monitoring organizations make better use of available election data. These and other resources are available at www.ndi.org/elections.

Photo Credits Cover: UN Photo by Martine Perre Section 1, Electoral Integrity: UN Photo by Tim McKulka Section 2, Open Election Data Principles: UN Photo by Martine Perre Section 3, Key Election Process Categories: UN Photo by Kevin Jordan Section 4, Open Election Data in Practice: Examples from Latin America: UN Photo by Martine Perre 133    public.resource.org/8_principles.html 134  opengovdata.io 135    sunlightfoundation.com/opendataguidelines 136  transparency-initiative.org 137    This publication is made possible by the generous support of Google Inc. The opinions expressed herein are those of the authors and do not necessarily reflect the views of Google Inc.

100


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.