This repository hosts election results data for the 2018 Afghan parliamentary elections for the Wolesi Jirga (the lower house of parliament), as reported by the Independent Election Commission of Afghanistan (IEC).

The primary output collects in tidy format all candidate results reported across all polling stations for which the IEC released data, and can be found in the files prelim_af_candidate_ps_data_2018.csv and final_af_candidate_ps_data_2018.csv, comprising preliminary results and final certified results, respectively.

These output files, which also include additional metadata for candidates and polling centers, are too large for Github (approximately 975 MB apiece) and are hosted remotely on Google Drive. A "lightweight" version of the preliminary and final polling station results data, which includes only the polling station code, candidate code, and vote variables, is also available. (Both files are approximately 95-100 MB and the final results file is also remotely hosted due to Github file size limits.) Accompanying keyfiles can be joined to expand these lightweight dataframes as needed.

This data is being released in open source for the purposes of contributing to public understanding of the elections and to allow for the analysis of available IEC reporting. Please read all accompanying documentation and link, cite, and credit as appropriate.

Note that as of initial publication, there are substantial as-yet-unexplained gaps between the IEC's polling center-level data for final Kabul results and its calculations of aggregate provincial totals; extreme caution is advised in conducting analysis on the data until these gaps are resolved by the IEC. This repository remains a work in progress and data and analysis herein should not necessarily be considered final or authoritative. See the section on caveats, gaps, and errors below for more details on these and other gaps.

• Summary of files
• Summary of results data
• Background to the election
• Primary sources for this dataset
• Caveats, gaps, and errors
• Contact and acknowledgements

Each row in the primary dataset represents a candidate's vote total for each polling station. This data has also been re-aggregated into polling center (prelim / final)- (both remote hosted due to size, or see lightweight versions - prelim / final), district (prelim / final)-, and provincial (prelim / final)-level summaries.

Additionally, the file prelim_final_ps_candidate_net_vote_change.csv summarizes net changes per candidate per polling station between the preliminary and final results, noting where polling stations did or did not report results; due to file size (approx 300 MB), it is also hosted remotely on Google Drive.

These datasets include zero vote counts where reported, but may not include data (such as polling centers or candidates) in cases where the IEC did not report any results at all. (See section on caveats and gaps below for validity checks and investigation of completeness or possible gaps in the data.)

For some keyfiles, see:

• candidate_key.csv for a list of all candidates for which the IEC reported results, grouped by electorate, with unique ID codes, names in English, Dari, and Pashto, and IEC-designated preliminary and final winner status, as well as formal party affiliation (as registered on the ballot), gender, and incumbency status;
• ps_key.csv for a list of all unique polling station codes, and their corresponding parent polling center, district, and province codes (see note on gaps, below);
• pc_key.csv for a list of all planned polling centers, polling station counts, reporting status in the preliminary and final results, and voter registration data;
• district_key.csv for a list of all unique districts as reported by the IEC (as well as the 21 subdivisions of Kabul); see also district_code_keyfile for a key used to track district codes across election cycles and datasets, drawing on the work of the Afghanistan District Maps project;
• province_constituency_key.csv for a list of provinces and constituencies, their codes, names in English, Dari, and Pashto, and the number of general / female reserved seats per constituency.

For a brief explanation of some of the main variables used, please also see this codebook. If you have questions about a variable definition or source, please feel free to contact the author.

The file all_pc_ps_candidate_vote_counts.csv provides a basic count by province of polling centers, polling stations, candidates, and votes reported in the preliminary and final results.

There is substantial inter- and intra-provincial variation within the results dataset. For a constituency-level summary of major indicators, including calculated variables on turnout against voter registration and population, average candidate vote shares, margin of victory thresholds, votes for winners and losers, votes for female and incumbent candidates, and distribution of votes between more urban and rural polling centers (as calculated through a nearest-neighbor polling center density index), see the constituency_summary_table.csv file.

A "net change" row in the summary table also notes differences between preliminary and final results for each constituency. Note that seven constituencies (Wardak, Baghlan, Badakhshan, Kunduz, Daykundi, Kuchi, and Sikh) saw no net change in vote totals between the reported preliminary and final results. Final Kabul results saw substantial changes that do not correspond to official IEC summary totals; see note on Kabul discrepancies, below.

Some initial visualizations and analysis of the available data can also be found in the graphics subfolder.

The Wolesi Jirga's five-year term in office last expired in June 2015, but members' terms in office were extended indefinitely amidst ongoing negotiations within the Afghan government over changes to election administration and other broader electoral reform debates. After multiple delays, the elections were ultimately held on October 20 2018 in 33 provinces and for the Kuchi (nomadic peoples) and Hindu / Sikh minority constituencies (for which there was only one candidate). (Kuchi and Sikh voters used the same polling center locations as other general voters but cast separate ballots for their respective nationwide constituencies.)

Following the assassination of Kandahar provincial police chief Abdul Raziq two days prior to election day, voting was delayed in that province and held a week later, on October 27 2018. Elections for Ghazni province - whose electoral results were the subject of a monthslong political standoff following the 2010 elections - remain indefinitely postponed, and as of current writing are planned to be held during the next round of national elections for the presidency, currently scheduled to take place in September 2019. (The Ghazni parliamentarians originally elected in 2010 continue to serve in the current parliament in the meantime.)

Afghan parliamentary elections are held under the Single Non-Transferrable Vote system, in which each province (and the Kuchi and Hindu/Sikh groups) serves as single constituency electing multiple members, for which voters may cast a single vote. (Electoral reform debates over the preceding years have included proposals for the creation of single member districts, or some form of proportional representation, but these changes have not been adopted.) A set quota of female winners per province is also guaranteed under law. (In the 2018 election only one female candidate, Raihana Azad, candidate code 24-1297-16 in Daikundi, won a non-reserved seat.) For more background, see the Electoral Law or summaries by the International Foundation for Electoral Systems or the Afghan Analysts Network, among other resources.

The 2018 parliamentary elections are Afghanistan's first to link voter registration to a specific polling center location. At the demand of some opposition political parties, the IEC agreed to introduce a biometric voter verification process at polling centers prior to voters casting their ballots, although observer reports indicate widespread technical and procedural problems with implementation of this system. The IEC and ECC ultimately appear to have adopted a procedure that would still accept votes cast in the absence of biometric verification in most cases.

Preliminary election results were initially scheduled to be finalized by early November 2018, with final certification due in December 2018, but the tabulation process was repeatedly delayed. After collecting polling station forms and ballot boxes for tabulation centrally in Kabul, the IEC began publicly releasing preliminary results in tranches starting on November 23 2018, and completed the preliminary reporting process on January 15 2019. The publication of results for Kabul were repeatedly delayed and the last batch to be announced. The Electoral Complaints Commission, a separate body tasked with adjudicating complaints, initially invalidated all results for Kabul province on December 6 2018, but later reversed its action and subsequently cooperated with the IEC in reviewing the remainder of the preliminary results.

Beginning January 20 2019, shortly after the completion of the publication of the preliminary results, the IEC began releasing final certified results data for provinces, again in tranches. On February 13 2019, President Ashraf Ghani issued an executive order modifying Afghanistan's electoral law and firing all members of the IEC and ECC commissions, in part in response to allegations of fraud and mismanagement of the parliamentary elections and in anticipation of presidential elections then scheduled to take place in July (which were subsequently postponed until September). The IEC secretariat halted the certification of the final results for the remaining 17 out of 35 constituencies until a new commission was sworn in. Commission members were elected in a vote by the already registered presidential candidates on March 1 2019, chosen by the president, and sworn in on March 4 2019.

The remaining provincial results were subsequently released in further tranches. On April 26 2019, approximately six months after the elections were held, President Ghani inaugurated the opening of the new parliamentary session, with results for Kabul (which elects the largest provincial delegation to parliament) and Paktia provinces still uncertified at the time. Paktia results were announced the following week, but on April 28 2019 the Electoral Complaints Commission released a statement again warning that it was preparing to invalidate the recounting and certification process for Kabul. After further review, the ECC certified the final results on May 8 2019, which were announced by the IEC on May 14-15 2019. As noted, this final tranche of data for Kabul contained significant gaps at the polling center level that do not correspond to the IEC's official provincial-level totals, although this does not appear to have been publicly reported at the time of announcement. Kabul and Paktia parliamentarians took the oath of office on May 15 2019.

All 2018 election results in this repository were drawn from the IEC results website and initially collected over the course of the December 2018 - May 2019 period. Once cleaned up enough to be legible, the underlying R code used to scrape, clean, and reorganize all data will also be published in the near future to allow for replication of results, although no guarantees are made as to its long-term functionality.

The IEC's pre-election candidate lists can be found here. The IEC also published lists of disqualified candidates. (Both sets are mirrored in the repository here.) However, checking these lists against results data and other sources reporting on disqualifications found that most if not all of the candidates appearing on the IEC's pre-election "disqualified" lists were in fact eligible to, and did, contest the elections. For a correct list of candidate disqualifications, see this Afghan Analysts Network report, or the original Electoral Complaints Commission Dari-language disqualification announcement here.

Candidates were assigned ballot symbols by the IEC, but these do not appear to have been shared across parties or provinces. Almost all candidates contested without a formal party affiliation. Party name transliterations and translations from Dari used in this project (see the party key) are unofficial, but where available are derived from the Afghan Ministry of Justice's list of officially registered political parties (Dari, English), and the party's position in that list sequence is noted where available. Candidates were assigned unique ID codes, which the IEC reported in its results data but not in any of the earlier pre-election candidate lists. (See also caveats below about possible unobserved gaps in the available candidate data.)

Incumbent Wolesi Jirga members who sought re-election in 2018 are noted in the candidate key with the past_winner variable, with their corresponding candidate codes in the 2010 and 2005 data noted if they were reported as winners in those periods. (Note that 2018 candidates are classified as past winners if they were designated a winner in either the preliminary or final round of the 2010 elections; only the final winners would be incumbent sitting parliamentarians. No preliminary winner status data was reported for 2005.) Because the IEC does not use the same candidate codes between election cycles, candidates can only be matched by name (which is not standardized across time periods), province/electorate, and in some cases candidate photos used on the published candidate lists. 2010 incumbency coding is derived in part from the Afghan Election Data Project; in one instance, a candidate identified there as an incumbent (Ghazni candidate Abdul Qadir Wahidi, 2010 candidate code 2710795) could not be definitively matched to a 2005 winner, and has not been categorized as a past winner in this project.

As of initial publication, candidates who lost past elections but who ran again are not currently matched to past candidate codes, which would allow for a more robust study of re-running candidate performance. If you have further information on candidate ID matching or additional candidate biographical or other metadata and would be willing to share, please get in touch.

For the IEC's pre-election plans for polling centers, see this list (or in Dari here), which was published September 27 2018. The polling center key notes (in the column pc_change_initial_final_plan) polling centers that were either added to or dropped from the pre-election plan when compared against an IEC polling center security assessment completed in January 2018, which was obtained from an election observer. (The column planned_2018 denotes polling centers that the IEC planned to open prior to the election.)

Some polling centers reported results despite having been dropped from the pre-election planning list. The file all_pcs_reporting_not_planned.csv (a subset of the polling center key) highlights those polling centers (35 centers in both preliminary and final, or 0.68% of 5106 planned). For planned polling centers that did not report results, see the caveats and gaps section, below.

Unlike in 2010, the IEC did not publish a count of planned polling stations per polling center in advance of the elections, which would allow for identification of any reporting gaps within polling centers; see caveats and gaps section, below. A list of polling stations that did report results in either the preliminary or final results can be found in the polling station key. (The polling center key also includes a count of reporting polling stations per polling center, based on the preliminary and final results.)

The IEC previously reported some public, provisional pre-election voter registration data by polling center, in Dari only, on its website here. The detailed final polling center-level voter registration data used in this repository (which includes changes between preliminary voter registration and finalized registration after an IEC voter list review process conducted in October) was obtained from an election observer.

Voter registration data was missing for 2347 polling centers (31.6% of 7413 total reviewed), most of which were to have been dropped in the final plan. 78 planned polling centers (1.5% of 5106 planned) were missing voter registration data in this source, and 9 polling centers (0.19% of 4674 polling centers reporting preliminary results) reported votes despite the absence of voter registration data.

The author received a dataset of latitude / longitude coordinates for 7413 polling centers from an election observer (ommitting 38 centers from the known universe of centers, all of which were dropped in the pre-election plan, and did not report any results). Although polling centers are public sites, this location data was shared with the request that it not be published in raw format so as to avoid any potential security risk to the physical polling center locations. Accordingly, absent reporting of this information by another source, the raw coordinate data will not be included in this dataset.

However, the available coordinate data was used to calculate nearest neighbor distances for the five closest polling centers for each respective center (in kilometers using the Vincenty ellipsoid formula, and in lat/lon decimal degrees), which can be found in the file pc_plan_nearest_neighbor_coordinates.csv and may be used for further geospatial analysis of results.

For administrative boundaries and shapefiles, see the other data section below.

The IEC initially released voter disqualification data in a machine readable format (viewable in Wayback reference here), but midway through the preliminary results reporting process removed this (very limited) data from its website and replaced it with a collection of zipped files containing scans of polling stations where results had been invalidated. (As of initial publication, only zip files for 28 of the 36 provinces or nationwide constituencies have been released, despite the announcement of all final results.)

These files are also mirrored here but at this stage have not been processed into a machine-readable format for further analysis. As of this writing the ECC has not published its own detailed disqualification reporting by polling station or polling center. Net changes between preliminary and final results can be calculated using the primary data outputs, but may also be the product of IEC data gaps.

As part of its preliminary results reporting, the IEC also released scanned images of most polling station results forms, which were downloaded as part of the data scrape process and may potentially be used as a visual check on the machine-readable results. Those images are too large to host in Github, but are mirrored here. For a full list of files, see ps_scan_file_list.csv; scan filenames use serial numbers which do not match directly to polling station numbers, and these have not yet been directly matched to their corresponding results data in the dataset. Full analysis of the scanned images has yet to be undertaken. As a note of caution, some preliminary comparisons suggest that in at least some cases the polling station scans are mislabeled on the IEC's website, and filenames or link titles may not correspond to the actual scan contained in the file.

Links to some scanned forms did not work in some cases; these observed file breaks are logged in the file ps_scans_broken_links.csv. Some polling centers also failed to include links to any scanned forms. The file ps_scans_reporting_check.csv offers a province-level summary of the difference between polling stations reporting in the preliminary results and accompanying scanned forms missing from the IEC website. (Note that this is not a completeness check against the number of polling stations originally planned by the IEC, only against the reported polling station results; see caveats and gaps section below for more.) Notably, no polling station results scans were made available for Paktia province, either due to broken file links or otherwise missing links.

Historical data for the results of the 2010 and 2005 parliamentary elections, drawing in part on previous work by Democracy International and the Afghanistan Election Data project (a project of the National Democratic Institute and Development Seed), is also included for reference in the past elections subfolder of this repository.

Full polling station (preliminary / final) and polling center (preliminary / final)-level datasets for the 2010 preliminary and final results are too large to be hosted on Github, and are mirrored here; lightweight versions of the files are also included in their respective data subfolders.

These datasets have been restructured to allow for joining with the current available 2018 data (and cleaned to correct some errors related to candidate codes, the Kuchi electorate, and zero counts / incomplete ballot reporting in some polling stations). However, caution is advised that polling center and polling station codes do not correspond across election periods (nor do provincial or district codes - but see other data below and the district_code_keyfile for a join key at those levels).

The code used in this data cleaning process is included for reference; however, some additional manual cleaning of old data was also undertaken that may not be fully reflected in this code, and caution in advised in running it without careful checks on the process. (The raw final results file for the 2010 elections is too large to host on Github, but can be accessed directly via the Afghanistan Election Data project here; or see the IEC's 2010 Final Certified Results page.)

Polling center- or station-level results for the 2005 elections are currently unavailable. (If you have access to such information and would be willing to share, please get in touch.) Although the electoral system used for presidential elections is not directly comparable to the parliamentary SNTV system, additional data for the 2004, 2009, and 2014 Afghan presidential elections may be added to this dataset in the future to allow for further cross-election cycle analysis.

For a parallel election results data collection effort, see the Development Seed Afghan Elections Data project and the Afghanistan Election Data project.

For a comprehensive resource on Afghanistan's district administrative boundaries (which are a subject of periodic political disputes, and have not been definitively standardized across datasets or official government documents), their relationship to one another, and changes over time, see this resource by Map Sync.

The Map Sync project also collects district population data estimates drawn from Afghanistan Central Statistics Organization estimates for the 2018-19 Solar Year, published in June 2018; a 2016 'Landscan' estimate used by Resolute Support / US Forces-Afghanistan and the Special Inspector for Afghanistan Reconstruction; a 2017 population estimate by iMAPP; and a 2016-17 CSO / UN Office for the Coordination Humanitarian Affairs population estimate. This population data is mirrored here in the file district_population_data.csv.

District and provincial boundaries and codes do not correspond across the 2005, 2010, and 2018 election periods. There is also no consistent English transliteration standard for district, provincial, or individual candidate names across (and in some cases within) election periods. IEC district codes for the 2018 election appear to match the most recent Afghan Geodesy and Cartography Head Office-published figures (both encompassing 421 districts), which should allow for data joins with election results and voter registration data. (Note that the one exception to this is the Gizab and Chenarto districts, which the IEC codes in 2018 as districts 2507 and 2506, reversing the AGCHO's codes for those district.) The district_code_keyfile can be used to make comparisons across election cycles, but may require further manipulation of results data to account for district splits.

For mapping purposes, UN OCHA has published a set of shapefiles for district boundaries, which can be found on the Humanitarian Data Exchange and is also included in the district data subfolder of this project for reference. Further transformations are required to re-base 2018 districts or election results data to match these district boundaries, however. For an unofficial, ArcGIS-only version of the 421 districts used in the most recent CSO and IEC data, see this Map Sync map.

This repository remains a work in progress and caution is advised when using it for analysis. The author cannot verify the accuracy of, or account for any discrepancies in, the underlying data, and makes no guarantees as to its completeness. The results collected at this point may be further altered by legal complaints, IEC amendments to the results, or other disputes and adjudication processes.

Due to the very narrow margins of victory separating winners and runners-up under the SNTV system, disqualification decisions (or revisions during the results adjudication and final certification process) can have a significant determinative impact on electoral outcomes. Although more details may be released, there is currently very little transparent reporting on these decisions, despite the months-long process of counting, review, and reporting. As noted earlier, candidates and some observer groups have highlighted concerns over the fairness or accuracy of the results announced by the IEC, alleging manipulation to favor certain candidates, which contributed to the removal of the IEC and ECC in mid-February before the final certification of all results.

The author cannot adjudicate or verify these charges and is presenting this dataset solely for the purposes of contributing to public understanding of the elections and to allow for the analysis of available IEC reporting, not to validate or invalidate the available data or to make claims as to how that data may correspond to individual or collective voter actions in practice.

The final polling center-level results data released by the IEC for Kabul, collected for this dataset shortly after its publication on May 15, had significant gaps that resulted in substantial differences between the provincial-level aggregate total votes reported by the IEC (on its final results page here, or in raw form in the repository here) and the calculated sum of polling center-level votes. All Kabul candidates' vote totals were affected. These vote differences are recorded in the final_iec_tabulation_check.csv file. In total, there was a difference of 172716 votes between the IEC's provincial sum total and the sum of available polling-center level results, the equivalent of approximately 30% of the IEC's final provincial total for Kabul.

If these differences were applied to the results, they would flip the win / loss status for four sets of candidates — for three general seats, Zabiullah Almas (candidate code 1-1306-43), Ajmal Gullab (1-1718-135), and Ezatullah Aatef (1-1430-80) would replace announced winners Tawfiq Wahdat (1-1066-102), Haji Hafizullah Jalili (1-1209-19), and Urfanullah Urfan (1-1176-45), and for one female seat, Salima Nekbin (1-1601-28) would replace Bibi Haji Parwin Durani (1-588-3).

Missing data on the IEC website is one possible explanation for these differences. 271 Kabul polling stations reported preliminary but not final results (including 29 polling centers, collectively encompassing 112 polling stations, that were entirely missing in the final). Some of these may have been removed from the final results due to their disqualification in the certification process. (See sections below on missing polling centers and polling stations nationwide.) Some of these stations or centers may have contained valid votes that are being counted in the IEC’s provincial totals but are not otherwise appearing at the polling center level on the IEC website. As of initial publication, there was no clarity from the IEC on whether these differences in vote totals are due to disqualifications, data errors, or other reasons entirely. However, these missing polling stations collectively totaled 70267 votes in the preliminary results, which would not fully account for the observed tabulation differences.

Statements by the ECC prior to the release of final Kabul results suggesting that approximately 20% of polling station results were 'missing' appear to be based on comparisons against the larger unpublished count of planned polling stations (see notes below), not missing in comparison to preliminary results. An observer source reported that 3341 polling stations in total were originally planned for Kabul, although no further breakdown of this figure, or counts for other provinces, was available; 2613 polling stations were counted in preliminary results, and 2981 in final.

While this repository is being published with the currently available Kabul data, and follows the IEC's official designation of final Kabul winners, extreme caution is advised in conducting any analysis on the data until these gaps are resolved or otherwise clarified by the IEC or ECC. If and when further Kabul polling center or polling station data is released, the dataset will be updated to reflect these changes.

Three candidates — in Parwan, Mohammadzadah Parsa (candidate code 3-1465-31); in Bamiyan, Mir Hussain Ibrahimi (10-1249-91), and Mohammad Hassan Ranjbar (10-1706-67) — were ommitted entirely from the initial preliminary results reporting, but were included in final results reporting data. (Their candidate metadata has been retroactively added to the master candidate key, but they do not appear in any of the preliminary results files, even as zero counts.)

An investigation of available ballot number sequences and reported results found some further gaps in the IEC's reporting in both the preliminary and final results, which might potentially be indicative of additional missing candidate data not reported in either set of results. These ballot number sequence gaps are logged in the file missing_ballot_numbers_in_data.csv. Because the IEC did not include ballot position or candidate code data on its pre-election list of candidates, these potentially missing candidates in the sequence cannot currently be definitively ID'ed and matched to other metadata from the other IEC lists.

As of initial publication, the IEC's reporting on results for the Kuchi electorate produced some apparent data entry errors. In both preliminary and final results, reporting for Kuchi constituency votes at polling center 0903077 (Hazrat Ali Doshi High School in Dushi, Baghlan) included duplicate data from the votes received by general electorate candidates in the same polling center. (This also appears to have resulted in the erroneous inclusion of non-Kuchi candidates in the IEC's "Results by Ballot Order" constituency-wide summary for the Kuchi constituency.) This duplicate non-Kuchi data has been removed from the dataset published here and the Kuchi results at that polling center retained.

Additionally, in both preliminary and final results, the polling center 3301007 (Boys High School in Farah city, Farah) reported data for general electorate candidates in the Kuchi results, but did not otherwise report Kuchi candidate results for that polling center. The duplicate general electorate candidate data has been removed from the dataset published here.

In addition to the Kabul vote discrepancies noted earlier, in four cases in both the preliminary and final results for four candidates in Farah province, the IEC's reported constituency-wide sums of votes did not match a calculation of the available polling station-level data. This involved undercounts by the IEC of at maximum two votes, and was not determinative in altering candidate rank or winner status.

Caution is again advised that other less evident data entry errors may also be present in the dataset. While all efforts were taken to ensure a complete capture of the published IEC data, doing so required processing through approximately 16,500 sub-pages of the results website, and some omissions may still be present in the data captured here.

The file all_pcs_planned_not_reporting.csv (a subset of the polling center keyfile) lists polling centers (467 in preliminary, or 9.14% of 5106 planned; 520 in final, or 10.18% of planned) that were to have opened according to the IEC's September pre-election polling center plan but which did not report any results in the preliminary data, whether due to closure on election day, data entry gaps, removal through disqualification, or other reasons.

In districts where some polling centers reported results and others did not, the IEC initially (as of mid-December) generated blank results subpages for the polling centers that did not report results. As of the final data scrape (February-May 2019), however, those blank polling center pages appear to have been hidden on the IEC website and do not appear in the list of available polling centers under each parent district, requiring a check against the polling center plan to identify gaps.

Polling stations (within each polling center) should in theory be designated for either male, female, or Kuchi or Sikh / Hindu voters (or in some cases, mixed stations). (In 2010, polling station classifications do not appear to have been fully enforced, with some votes for Kuchi candidates reported from general electorate-designated polling stations and vice versa).

The IEC did not publicly release a list categorizing the 2018 polling stations, and unlike in 2010 did not release a comprehensive list of planned polling stations per polling center, which would allow for a reporting completeness check. In the absence of this benchmark, the file ps_reporting_status_summary.csv highlights differences in reporting status for polling stations between the preliminary and final results, but does not represent a full completeness check against pre-election plans.

Where available, scanned polling station forms do indicate their respective voter categories. If you have access to a comprehensive list of polling station categorization or planned stations in a machine-readable format and would be willing to share, please get in touch.

Feedback, corrections, or suggestions for further expansion or collaboration are greatly appreciated. For questions, suggestions, or to contribute further, please leave an issue request here on Github or contact Colin Cookman by email or Twitter.

Although I am solely responsible for any errors in this project, I am grateful to Asma Ebadi and Lucy Stevenson-Yang for their research and coding assistance; to the Independent Election Commission for releasing elections data in a public form that allows for open analysis; to the anonymous election observer sources who generously provided additional data sources; to the Afghanistan Elections Data, Map Sync, and OCHA Afghanistan / Humanitarian Data Exchange projects for their work on collecting other supplementary data; to the researchers and authors at the Afghan Analysts Network for their detailed reporting and analytical resources which have informed my understanding of Afghanistan's elections; to John Ray, Luke Sonnet, and the Stackoverflow and RStudio communities for lessons in the use of R for data collection, cleaning, and analysis; to Democracy International for providing the opportunity for me to gain firsthand experience as an observer to the 2010 and 2014 Afghan elections; and to Belquis Ahmadi, Staffan Darnolf, Scott Smith, Andrew Wilder, Scott Worden and many other colleagues at the U.S. Institute of Peace for their guidance, support, mentorship, and patience as I worked on developing this project.