Background & Summary

In June 2024, the Asian Network for Free Elections (ANFREL) published its final report on Indonesia’s general election, held on February 14th 20241. With nearly 280 million people, including about 205 million registered voters, Indonesia is the world’s third-most populous democracy, following the United States and India2. ANFREL commended election authorities and voters for successfully managing an exceptionally complex, simultaneous election for the president and vice president, the two-chamber national legislature, and 38 provincial legislatures in a single day.

However, ANFREL also highlighted several concerns. These included voter difficulties with the electoral system’s complexity and the overwhelming number of participants (report, p.37), controversial legal changes to the vice-presidential candidate’s age requirements before the election (p.70), alleged misuse of state funds for campaigning (p.26), unregulated use of artificial intelligence during campaigns (p.85), and threats to press freedom, as reported by the Alliance of Independent Journalists (Aliansi Jurnalis Independen, AJI Indonesia) (p.38).

ANFREL’s report concludes with recommendations for legal reforms, improvements in the electoral process, and enhanced voter access, information, and support. It specifically calls for:

“citizen groups to uphold norms, enhance accountability, and drive reforms by utilizing existing mechanisms, influencing cultural changes, and advocating for reforms... A crucial step for local civil society groups is to transition from merely collecting and reporting isolated incidents of abuse to implementing a more systematic and structured approach to data collection, resulting in more robust findings.” (1, 31)

The National Democratic Institute (NDI) Open Election Data Initiatives (OEDI) defines essential election data for long-term trend analysis in three main categories: a) information on the institutional framework of each election, b) pre-election data on candidates and processes, and c) post-election outcomes (ndi.org, openelectiondata.net). Providing permanent open access to data in all of these categories in the interest of democratic accountability is challenging, especially for early elections after Indonesia’s re-democratization in 1999, when election result collection was not yet digitized for electronic dissemination.

Despite these challenges, Indonesia saw momentum for open access to election information after re-democratization. Civil society groups, in collaboration with the KPU, made significant progress in distributing voter information before elections and publishing results afterward, a collaboration that continued until after the 2019 election. However, Indonesia’s legal framework does not mandate information openness regarding pre-election candidate data. Moreover, public data distribution and maintenance depends on the commitment of election officials, who are reappointed every five years, bringing varying perspectives and commitments.

This article has as its main objective to create a publicly available Indonesian Election Archive (hereafter IEA)3 that provides systematic, longitudinal election data across multiple levels of elected administration in Indonesia. These data include longitudinal data in all three key categories: a) institutional framework, b) pre-election candidate information, and c) post-election results. At the same time this project documents the collaboration between the Association for Elections and Democracy (Perludem), a leading Indonesian civil society group, with the KPU and U.S. scholars to make all of this information publicly available permanently.

Thus, setting the stage, the remainder of this introductory Background section outlines the main features of the Indonesian institutional system for electing representatives at various administrative levels, and recounts the KPU-Perludem collaboration to disseminate voter information and gather data. The three substantive sections, following the introduction, concentrate on introducing the IEA3 data. These include a section on Methods where we detail the collection of the data; the Data Record section where we delineate the types of data included in the IEA3 and explain where these can be accessed; and the Technical Validation section where we explain the steps we took to verify the accuracy of the data provided in the IEA3. The IEA3 data presented in this article is a standalone contribution. Furthermore, for interested readers, a Supplementary Appendix outlines a longer-term goal to compile Indonesian election information beyond the data provided here.

Indonesian elections’ framework

To briefly outline the institutional framework, Indonesia is divided into five administrative units, as illustrated in Fig. 1 (Reproduced4, with permission from Cambridge University Press). Representatives are elected in three and a half administrative levels and appointed in one and a half. For the national level, elections are held for the President and Vice President, and two chambers of the national legislature: the Dewan Perwakilan Rakyat (DPR, People’s Representative Council) and the Dewan Perwakilan Daerah (DPD, Regional Representative Council). Below the national level are 38 provinces, including two special regions (Aceh and Yogyakarta) and the capital, Jakarta. Of those, four were created in Papua before the 2024 election, increasing the total number of provinces from 34 to 38, the number of electoral districts from 80 to 84 and the lower level legislative (DPR) seats from 575 to 580.

Fig. 1
figure 1

Administrative levels in Indonesia.

Full size image

Below the national level, each province elects a provincial legislature (Dewan Perwakilan Rakyat Daerah Provinsi, DPRD Provinsi or DPRD-I) and a Governor, except Yogyakarta, where the Governor’s office is held by the hereditary sultanate. Provinces are further divided into regencies (Kabupaten) and cities (Kota), which elect municipal legislatures (DPRD Kabupaten/Kota or DPRD-II), Regents (Bupati), and Mayors (Walikota). Below these are sub-districts, whose heads are appointed civil servants, and rural villages, which elect their own heads. Urban villages, however, do not hold elections, as their heads are appointed civil servants. The IEA3 data presented here focuses on providing institutional data for all elected offices above the village level and election data for national level representatives, the President and Vice president and the two houses of the national Legislature the DPD and the DPR.

While partial election results and some candidate information from recent Indonesian elections are available online, they remain scattered across various sources including KPU documents (kpu.go.id), compilations by Indonesia’s Statistics Office (bps.go.id), the Global Elections Database (globalelectionsdatabase.com), the Election Passport (electionpassport.com), and the Constituency-Level Elections Archive (electiondataarchive.org). In addition, various scholars who study Indonesia often provide the data they have worked with publicly on their websites. Similarly, numerous organizations provide information on the legal framework surrounding the elections. These include, for example, The International Foundation for Electoral Systems (ifes.org), The Electoral Knowledge network (aceproject.org), and the Carter Center (cartercenter.org). To our knowledge, no sites provide comprehensive election data that emphasize multiple elected bodies, align with the NDI’s framework, maintain consistent granularity, or are systematized for examining trends and outcomes across time and diverse elected bodies.

Government and civil society collaboration to collect and distribute information in 2014 and 2019

Prior to the 2014 election, the KPU sought to enhance information access for voters, recognizing that over half of Indonesia’s electorate is under 40 and technologically adept5. To achieve this, the KPU collaborated with Perludem to digitize pre-election information and make it widely available. Perludem data enumerators organized and digitized KPU data, including candidate information, electoral district maps, and election regulations. This work spanned multiple regions, including Jakarta, Aceh, Bali, East Java, and Papua.

To further improve access, Perludem developed an Election Application Programming Interface (API)—a user-friendly, open data platform providing verified election information. Perludem also hosted a hackathon, inviting Indonesia’s tech community to create applications based on the election API, accessible via smartphones and websites. This initiative inspired the KPU to develop its own API for disseminating election data, expanding public access further. After the 2014 election, the KPU and Perludem formalized their collaboration through a Memorandum of Understanding to increase voter participation, enhance government accountability, support marginalized groups, and promote a healthy political ecosystem.

Before the 2019 election, Perludem, in collaboration with the KPU and Google Indonesia, launched a platform designed to centralize voter information, focusing on young, tech-savvy voters who actively use social media. The platform, pintarmemilih.org ("Smart Voting"), provided comprehensive election data but became inactive after the election, with information migrating to the main KPU website.

Methods

In response to calls from international election authorities for a tripartite collection of election-related information, the IEA3 systematizes and makes available data on: a) the institutional structure governing elections, b) pre-election candidate data, and c) post-election results. Specifically, we provide user-friendly access to:

  • Laws defining the election framework, including nomination rules, from 1999 to 2024.

  • Pre-election candidate lists from 2014 to 2024 and

  • Post-election results from 2014 to 2024 at various levels of detail.

The data collection employed a multipronged approach over time. First, Perludem staff collected documents during the collaboration with KPU described above, next Perludem and UMD researchers accessed public KPU websites pertaining to earlier elections (2014 and 2019) and downloaded PDF documents with official published candidate lists and results. For the latest election (2024) the team also scraped information in other formats from official KPU websites. Additional information was obtained from the website of the Indonesian Audit Board (Badan Pemeriksa Keuangan (BPK)). Finally, the researchers on this project continue to work closely with KPU staff to source additional data where needed. Thus, some of the official documents used here were supplied directly by KPU staff. To facilitate tracing to the original sources of data provided in the IEA3, the archive includes a file that, for every variable provided for every office, specifies the official file and/or the website from where data for individual variables were sourced.

Importantly, the data that were collected are all in the public domain. Specifically, in Indonesia, Law 14, 2008, on public information disclosure establishes the importance of transparency of government provided information. To further cement the public provision of government information, in 2012 the Indonesian government launched the Open Government Indonesia (OGI) initiative to address issues at the intersection of civil society and government, particularly electoral administration and voter information (ogi.bappenas.go.id). The OGI secretariat, including the KPU, collaborates with civil society organizations - including Perludem - to promote transparency, participation, innovation, accountability, and inclusiveness in governance and public services, including election governance.

Below, we describe in greater detail how each part of the IEA3 data was compiled.

Institutional Data Collection

First, we compiled a list of elections held in Indonesia at the National, Provincial and Regency level. These are enumerated in Table 1. Following re-democratization, Indonesia held its first direct election for the national legislature’s lower chamber in 1999, and the first direct presidential election in 2004. In 2009 and 2014, the President, both chambers of the national legislature, and the provincial legislatures were directly elected in the same year but at different times. The first simultaneous election for the President, both chambers of the national legislature, and provincial legislatures occurred in 2019, followed by the second in 2024.

Table 1 Election schedule for the three most aggregate levels of administration from 1999-2024.
Full size table

Regional elections followed a different timeline. The first direct regional elections were held in 2005, with each province, regency, and city adhering to its own schedule. This temporal scattering created complexities in election management and dispute resolution. To address this, the government introduced simultaneous regional elections in three stages. In 2015 elections were held in 269 regions (thereof in 9 provinces, 224 regencies, and 36 cities). In 2017 elections were held in 101 regions (thereof in 7 provinces, 76 regencies, and 18 cities). In 2018 elections were held in 171 regions (thereof in 17 provinces, 115 regencies, and 39 cities). By 2020, the local governing bodies elected in 2015 reached their term limits, with those from 2017 and 2018 reaching theirs in 2022 and 2023, respectively. During this transitional phase, no regional elections were held in 2020, 2022, or 2023. Instead, interim caretakers governed these regions. The first simultaneous regional elections for all regions were held in November 2024.

Next we identified key documents governing nominations, election procedures, and the translation of votes into representation at the National, Provincial and Regional level for all elections held for these bodies in Indonesia to date. This collection does not include rulings on administrative borders or changes of these. The law on political parties clarifies nominations and is, therefore, also included. For all the listed election laws from re-democratization in 1999 to the 2024 elections, we collected original legal texts. These are all available via the Legal Documentation and Information Network of the Audit Board (Badan Pemeriksa Keuangan or BPK) of the Republic of Indonesia (bpk.go.id). Specifically, using the legal database of the Audit Board (peraturan.bpk.go.id/), we searched for each original legal text and downloaded the document once it had been located. Next, we compiled the main body of all the individual files into a single PDF file, collating all the pertinent law in chronological order. The 2017 electoral law includes numerous appendices that were not included in the collated file but can be accessed on the website of the BPK.

Pre-Election and Post-Election Data

As the primary election organizer, the KPU is the main source for pre-election candidate information and post-election results. Because the data for the 2024 election were collected in real time, these data were obtained from the KPU website (kpu.go.id).

In contrast, the collection of longitudinal election data in Indonesia is exceedingly complex because these data were originally collated and transmitted between various administrative levels in a multitude of documents.

To give the reader an idea of the complexity of the existing source documents, including the multiple agencies and different levels at which the data have been collated and made publicly available over time - we briefly describe the data reporting procedures here. In the 2004 and 2009 elections for national level offices, results were collected manually at the polling stations via a form called C1. Polling station officials then sent the form along with aggregate counts to the sub-district. Sub-district officials verified poll counts and sent so-called form D, or sub-district level aggregates, to the regency/city level. Regency/city level officials sent on, to the provincial level, a new form D, based on regency/city level aggregates. Province level officials aggregated regency/city level results for their province and sent to the national level. Finally, KPU national level officials aggregated provincial level results and made these public.

In 2014, election results were manually counted as before, with the addition of polling officials sending form C1 to KPU officials at the regency/city level for scanning. These scans were forwarded to the national KPU data center, which published form C1 on the public counting system (situng) without calculating aggregates. Public access to situng later became unstable. In 2017, provincial elections maintained manual counting, but the KPU also published every iteration of form D on situng alongside form C1.

In the 2019 simultaneous election, the manual counting remained, as did KPU publication of all forms C1 and every iteration of form D on situng. Subsequently, the 2020, simultaneous local elections retained manual counting, but the publication system shifted from situng to sirecap, a mobile app. Polling officials photographed form C1 using sirecap, which aggregated the data. After verification, results were sent to the national data center, and the KPU published the polling station count with C1 photos at infopemilu.kpu.id. In 2024, manual counting and C1 publication on the KPU website continued unchanged.

To further complicate matters, the mandate of the KPU does not extend to maintaining a public archive of election information for prior elections. Therefore, some KPU documents made publicly available around past elections can no longer be accessed on the web or were only ever made publicly available offline. Consequently, while the IEA3 provides consistent information for each office over time, the level of aggregation and the origin of the source documents varies.

As described above, for the 2014 and 2019 elections, the IEA3 data collection benefited from a collaboration between Perludem and the KPU. Perludem affiliates systematized KPU data and worked with the election tribunal to make the data publicly accessible. Thus, much of the IEA3 collection of data for elections prior to 2024 originates from the collaboration between KPU and Perludem during the 2014 and 2019 elections, where the KPU made available a great variety of original documents that Perludem staff helped publish and archived at the time.

IEA3 researchers found other official documents on the web pages described above that are not actively maintained, but had not yet been retired. We scraped websites, and downloaded and scraped PDF documents to extract pertinent information, which was then cleaned, verified and organized to align with data from previous elections, ensuring comparability at the appropriate level of granularity. Finally, and importantly, when requested, KPU staff endeavor to make original documents from their offline archive of past elections available to fill any holes that we find in our data collection.

Data Records

The IEA3 data is available for download at the Open Science Framework (OSF.io)(DOI 10.17605/OSF.IO/HJQZD). Table 2 lists the archive files. The first file, “Election_Law”, contains qualitative information that is the law governing the electoral framework of Indonesia. Next, are three quantitative election information files, pertaining to each of the offices covered in the archive, respectively the Presidency and the Vice Presidency (Pres_VP), the Upper house of the National legislature (DPD), and the Lower House of the National Legislature (DPR). Each of these files is described in greater detail below. Two additional files are included in the archive. First is a file entitled “Codebook”, the second auxiliary file included, is titled “Sourcebook”.

Table 2 The files in the IEA.
Full size table

The codebook includes a project description, details the bibliographic citation and explains where the data can be accessed, and lists the variables in each of the quantitative files and describes them. Altogether, in addition to a version number, the quantitative files contain 16 variables listed in Table 3. Identifier variables including election, year, month and province are common to all three quantitative election files. As detailed in the table, other variables are provided for some and not other elections as appropriate. For example, DPD candidates all run as independents, therefore, no party information is provided for DPD elections. To make this data collection as compatible with earlier efforts as possible, the codebook format and the provision of the quantitative data on the lower house of the national legislature closely follows the format of the earlier CLEA data collection on the lower house of the Indonesian legislature. This is with permission from CLEA directors who, in turn, have expressed interest in linking to the IEA3 so that users can obtain the greatest coverage of Indonesian elections (A. Hicken, personal email communication, 5-30-25).

Table 3 Codebook variables.
Full size table

Because of the variety of the source files described above, to facilitate tracing of the information contained in each variable in each of the quantitative documents to their source, the source-book details, by year and office, the source of the variable information as shown in Table 4. The Table illustrates that the source documents vary between office, and sometimes between variables for a given office. Furthermore, the source documents vary between election years for each office. If the source document for a given variable for a given office contains information at a level more disaggregate than the level at which the information is being provided in the IEA3, we aggregate the information to the appropriate level (province for the presidency and the DPD and district for the DPR).

Table 4 Sourcebook: Sample list of variable source documents.
Full size table

Notably, all of our sources are official, either websites or PDF documents, and sometimes several different documents are available per election for cross-validation. The source-book is not a comprehensive list of all the documents we consulted, but does provide at least one definitive source for each variable, per office and year.

Archive data on Institutions

The first data file made available in this archive includes a compilation of the laws defining the institutional framework of Indonesia’s elections. This framework is highly complex and has undergone significant changes since re-democratization at the turn of the century. Earlier studies provide valuable insights into administrative structures and changes before, during6,7, and after re-democratization4,8,9,10,11.

Table 5 lists the pertinent laws included in this archive, organized by relevant office and indicating whether they apply to current or past elections. In total, 16 distinct laws have shaped the principal features of elections at the three administrative levels (eight offices), for which data are collated and made public at the national level. These levels and offices are the Presidency, the lower level legislature (DPR), the upper level legislature (DPD), the provincial level legislature (DPRD-I), Governor, the district level legislature (DPRD-II), and Regent/Mayor. Laws governing village-level elections, held on varying schedules with results collected locally only, fall outside the scope of our work.

Table 5 Law: Number and year.
Full size table

Many of the laws address multiple offices. Table 5 organization clarifies which laws contain provisions relevant to specific offices of interest. Additionally, the last row highlights four key laws on the organization and structure of political parties, as these are critical for understanding electoral candidate nominations and the regulation of the parties.

For effective longitudinal analysis of election-related data in Indonesia, it is crucial to understand these administrative structures and their evolution over time. The collected laws detail several significant changes to the electoral framework between 1999 and 2024. These include foundational shifts, such as the introduction of direct elections for previously unelected or indirectly elected bodies and an increase in the number of administrative districts. Additionally, the rules governing elections have evolved, including changes to the legal threshold for entering the national legislature, a shift from a closed-list to an open-list system, and the gradual implementation of gender quota legislation. Adding complexity, rules regarding candidate nominations have also evolved over time. These changes are documented in the election and party legislation included in the archive.

The IEA3 collection provided here includes all of the law listed in Table 5, organized chronologically, in a single PDF file. The file name is: “Election_law”.

Archive data on Elections: Candidates and Results

The next three data files made available in the archive are in CSV format. The files contain information on three elected offices, respectively the President and Vice President, the Upper House of the National Legislature (DPD), and the Lower House of the National Legislature (DPR). Each file is labeled for the office and each contains information for the 2014, 2019, and 2024 elections. The granularity of the data varies by office. Presidential data and data for the DPD, is organized by province, and for the DPR by the district.

The files include information on names and gender of all candidates running for office in any of the years covered. Between 2014 and 2024, over 24.6K candidates competed in DPR elections. Tables 6, 7, and 8 detail the number of candidates by party and gender for the 2014, 2019, and 2024 elections, respectively. On average, most parties had slightly more male than female candidates, with a gender ratio of approximately 1.6 males to every female candidate. The current DPR candidate data also includes party affiliation for each candidate, as this information was systematically collected across elections. For DPD elections during the same period, the data records over 2.5K individual candidates.

Table 6 Count of DPR Candidates in 2014 elections by party and gender.
Full size table
Table 7 Count of DPR Candidates in 2019 elections by party and gender.
Full size table
Table 8 Count of DPR Candidates in 2024 elections by party and gender.
Full size table

The files also include post-election data for the offices of President and Vice President, the DPR, and the DPD. In line with the presentation of the candidate data, results for the Presidency and the DPD are presented at the provincial level. One exception here is that official results have yet to be released for DPD Valid and Invalid Votes for the 2024 election, and are therefore not made available here. Consistent with earlier data collections (e.g., CLEA), DPR election results are presented at the district level.

Technical Validation

The law collection contains original documents obtained from official sources as detailed above. Consequently, technical validation of this collection consisted of curation rather than checking of the content. Curation of the legal collection was led by the Perludem expert team, with input from other team members. At the margins, which legal documents are essential is based on some judgment calls - for example - the present collection does not include separate regulations on administrative boundary changes, but does include all major legislation affecting the principal architecture of each election. The IEA3 provides the legal texts in Indonesian only. Consequently, no verification of translation was necessary. Users who do not speak Indonesia are encouraged to look to the sources noted earlier, such as IFES and the Carter center, for English translations of the law.

Technical validation of the information in the three CSV files was extensive. The data were sourced from the official election tribunal in Indonesia (KPU) as described earlier. Some of the data was scraped from the tribunal’s websites, other was scraped from official documents publishing results using python code written for this purpose. In some cases, Perludem staff recorded the information from the PDF files by hand.

The information in the three CSV files includes information on candidates running for each office. Each election in Indonesia sees thousands of candidates competing for seats in the DPR, and the DPD. For example, in the 2024 election, 9,917 candidates vied for seats in the DPR alone across 84 electoral districts. Technical validation of the candidate data consisted of ensuring consistency of candidate names and gender as recorded in the original sources. This process is handled differently for each election. Since candidate information for the 2024 election remains available on the KPU website, we directly scraped candidate data from the site, which already includes gender information. For the 2019 and 2014 elections, gender information was assigned based on official source documents and cross-validated by at least three people to ensure accuracy. We also summed the total number of candidates by gender and party as illustrated in Tables 6, 7 and 8, and verified overall counts. We retained the formatting of candidate names as recorded in the original documents, which typically includes their academic degrees or religious titles.

The numerical data in the CSV files were validated in multiple ways and cross-checked by multiple people. While one member of the team might assemble a given file, another member would cross-check the data with the original sources and internally across the document. The quantitative information is of two kinds. First are static variables that denote year and month, geographic unit and elected office. These were entered and checked by different members of the team. Second, are variables that contain variable quantitative date. Generally speaking, the cross-checking of the variable quantitative data consisted of cross-checking whether sums of numbers added up correctly and correcting any instances where this was not the case by referring back to the original documents. Table 9 lists these variables and explains how they were validated. The first column lists the variables themselves. These are Registered Voters (RV), Valid Votes (VV), Invalid Voted (IVV), Candidate Votes (CV), Total Party Votes (TPV), and SEATS. To check the first three (RV, VV, IV) we calculated the difference between RV and the sum of VV and IVV in the geographic unit at which the votes are recorded. For the Presidential and Vice Presidential Election and the Upper House of the National Legislature (DPD) this geographic unit is the province, for the Lower House of the Legislature (DPR) this geographic unit is the District. If the difference was anything other than 0 all three numbers were cross-referenced with an original KPU document. Candidate votes were validated by summing all candidate votes in a geographic unit and subtracting the number of valid votes in the geographic unit from the sum of candidate votes. Total Party Votes were validated by summing the total party votes in a geographic unit and subtracting from that sum the number of valid votes in that geographic unit. Similarly, if the sum was anything other than 0 the source document was referenced for a correction.

Table 9 Validation of quantitative information.
Full size table

SEATS were checked by summing the total number of seats allocated and subtracting the legally mandated number of total of seats overall in a given year. Seats were also checked against the known allocation of seats to a party. Tables 10, 11, and 12 summarize DPR results by party for elections from 2014 to 2024. Of the 24,601 candidates who competed across these three cycles, 1,715 were elected.

Table 10 Number of DPR Seats by Party, 2014 election.
Full size table
Table 11 Number of DPR Seats by Party, 2019 election.
Full size table
Table 12 Number of DPR Seats by Party, 2024 election.
Full size table

Seat allocation was also checked by summing the number of seats and subtracting it from the known district magnitude in a given district (DPR). Table 13 provides an example of district-level data, showing that in Jakarta District 1, six candidates from five different parties were elected to the DPR in 2024. For DPD elections, the total number of upper house members elected in each province was compared to the legally mandated number for each province.

Table 13 Number of DPR Seats by Party in Jakarta District 1, 2024 election.
Full size table

Occasionally a discrepancy could not be resolved with a reference to an original document. In such cases, we searched for and collected additional official documents (sometimes at different levels of aggregation) to resolve the differences uncovered during the technical validation.