Most recent version 🔗 of Party Facts codebook (Github).
Project leaders
Contact
References
PF-Web page partyfacts.herokuapp.com // PF-Data Github repository // PF-Dataverse archive
see full credits at Party Facts “about section”
Initial version of the codebook was created by Phillip Hocks and Jan Schwalbach (University of Bremen) in 2015. This work was funded by a small research grant of the University of Bremen (M8 Plus).
The Party Facts project is a gateway to empirical data about political parties and a modern online platform about parties and their history as recorded in social science datasets. It makes use of social media technologies to create a collaborative data infrastructure following an approach to collect data successfully applied by the Encyclopedia of Life (EOL).
Political scientists have accumulated a large amount of data on political parties. This information is included in mass surveys, data handbooks and various datasets on election results, voting records, party characteristics and party positions. With this information we can trace the dynamics of party competition across countries and time. However, the many existing datasets with crucial information about political parties are difficult to link and there is the need for a platform that helps to combine existing sources.
With Party Facts we want to establish an infrastructure that supports political scientists in linking parties across datasets. Our work is based on the experience we gained in recording and linking party information in the Manifesto Project and the ParlGov project with initial data for Party Facts derived from these two projects. In the Party Facts project we link main datasets of political science and provide a platform for other scientists to add party information and additional datasets.
The main output and workflow of the project takes place on the Party Facts Website (PF-Web).
Dataset imports and exports are provided in a public Party Facts Data Github repository (PF-Data). Github is an online service for software development and project management which allows us to coordinate, administrate and adapt the project and datasets as a team.
Long term archiving is provided at the Party Facts Archive Dataverse (PF-Dataverse).
The Party Facts project aims to gather information on all parties in the world which won at least 5.0% of the votes in a national election.
The project may add parties
The respective threshold for including and linking parties from external datasets depends on the quality of the party level information (names, first/last year, election results) in the external source.
We aim to avoid including small and short lived parties. If feasible, we remove these smaller parties during the import.
Examples for additional, optional inclusion criteria of parties:
Core parties are observations that have been gathered by the Party Facts project. They can be added and edited on the Party Facts webpage. Currently we include around 5900 core parties.
External parties are observations with party information extracted from other datasets – see Party Facts Data. Currently we include around 45000 external parties.
Party Facts covers every country in the world that is included in an external dataset.
Currently, it includes 217 countries. The main focus of the project is on the national level but parties from relevant sub-national units may be included as well, if they are covered by several datasets.
Country definitions follow V-Dem Country Coding Units. V-Dem does not cover several smaller countries where we use ISO 3166-1 country code definitions.
For some datasets we harmonize the sub-national or autonomous regions data. This may require updating the country data in the import script. See for example the ParlGov import for Greenland and Faroe Islands.
The aim of the project is to combine and harmonize party information from different social science datasets. Parties from external datasets have to be linked to the corresponding core parties within the Party Facts project. We distinguish between core parties and external parties. Core parties are the party units created in Party Facts. Information about external parties is extracted from the respective datasets. A direct linking between parties of different external datasets is not implemented and can be achieved through core party linking.
While linking a dataset to the project, please keep the following issues in mind:
There are other technical options for linking parties:
Most external parties are linked to exactly one core party. A few external parties may be linked to multiple core parties.
The German Greens provide an example, in Party Facts a new party is recorded after the Greens from the West and East merged in 1993. The ParlGov project does only record one party. Hence, this observation (external party) needs to be linked to the two Green parties in Party Facts (core parties).
If an external party is linked to multiple core parties we distinguish between one primary link and secondary links. The primary link serves as the main link whereas all other links are secondary links. It should be linked to the core party with the longest time overlap.
We distinguish between the two types to make merging datasets easier and to avoid merge ambiguities (many-to-many relationships, m:n). Initially, the first link of an external party is set as the primary link. The primary link for an external party with multiple links to core parties can also be reset on the page of the respective external party. A user comment for this party is created for any change in the primary link.
For public datasets we complete the linking of all parties that meet the Party Facts population criteria (>5%, see above). Public datasets are visible on PF-Web without login and are archived regularly in the PF-Dataverse.
For smaller parties, we may apply a staggered approach. All parties >5%, are linked during the initial import. Parties \<5% and >1% (or 2%) are imported but may not be linked completely in Party Facts. A section “Linking status” in the “readme” of the dataset documents the status.
Note – Staggered inclusion for parties \<5% introduced in March 2021. Previously, all public and archived datasets were completely linked.
If a party of an external dataset has no corresponding match among the core parties, please add a new core party in PF-Web (login required). This can be either achieved through the core parties list of the respective country or through the linking section.
New core parties should include a vote share or seats share with its year to identify the party. A short description should be included if no share information is available (e.g. leader information: “PM Indra Gandhi (1969-1977)” ).
Observations (parties, links, etc.) may be removed by any user due to coding mistakes or duplicates. It is important to give a short reason for deleting an observation in the comment to document the coding decision.
We record only electoral alliances that take part in at least two elections. All other (one term) electoral alliances are linked to the “alliance” category – see section “Technical links”.
An electoral alliance included in at least three other datasets may be added as well.
Core parties that form an electoral alliance included in Party Facts can be recorded in the following format
<PartyFactsID1> <PartyFactsID2><PartyFactsID>Note — Coding of party changes and party names incomplete.
Relations between parties are shown on the core party’s page, containing successor and predecessor parties.
Renamings of parties are recorded in the description:
<year> <name_english> (<name [optional]>, <name_short>)Additionally the new (or respective old) name of the party may be added to the party name by using a slash to divide the names. Please see chapter “Adding parties” for further instructions.
As a measure to review existing party links, users are able and encouraged to validate links of other users. An active validation of information is an important part of the project. Based on the party names and (roughly) matching dates, a quick validation is often possible. Otherwise the information given about election results may provide additional help.
While reviewing established party linkings there are several options applicable:
You can remove your validation by clicking the highlighted validate-button again. Negative validations are summarized and reviewed on the recent linking page.
In Party Facts we collect party level information from external datasets. We do not include these external datasets but extract party information only. Therefore different datasets from the social sciences are linked but are not combined. Currently around 60 external datasets are imported and around 45000 external parties are linked.
Datasets are grouped into different categories depending on their connection to the Party Facts project, the number of parties they cover and their relevance for social science research.
Core members are datasets, which formed the foundation of the project. Partner projects include those datasets with whom the project has cooperated. Main datasets contain datasets with a wide scientific reach. Datasets which were recently included or with remaining technical problems are grouped under experimental. Any user can import a (scientific) dataset as a user dataset.
Currently we include the following categories for datasets:
Please note that party information from the respective datasets are only included into Party Facts and may not yet be completely linked to core parties.
For each dataset in Party Facts we document the following information and variable names.
Variable names used to import party information from dataset into Party Facts.
For further information about the content of different datasets (e.g. number of parties, countries, time span etc.) and the respective projects, see datasets overview.
Party Facts offers two different datasets, which are directly available in the PF-Web download section and in the PF-Dataverse.
The dataset Datasets parties contains all external parties currently included and linked in Party Facts. The information covers:
The dataset Core Parties contains all Party Facts’ core parties with the following information:
The download section provides merge examples in R and Stata.
In general everyone can add a dataset to the project. But please make sure that the dataset fits the following criteria:
Ideally, a dataset is prepared by a user for an import into Party Facts. A step-by-step guide on how to import a dataset into the Party Facts project can be found in the PF-Data Github repository of Party Facts. It allows you to prepare your dataset for an import into the Party Facts project and to link your dataset to other well-known scientific datasets.
An import of a prepared dataset into the database at partyfacts.org is done by the project maintainers.
External parties with a unique match of a party name (name short, original name or English name) or a “partyfacts_id” are automatically linked during the import into the database.
If you have further questions, please contact the project coordinator.
We use comments for party specific coding information, user interaction, issue tracking, to highlight potential mistakes and to discuss coding issues (PF-Web login required).
To address open issues and questions with respect to the linking of
parties or information about parties, please use the discussion section
below each party (core and external). To address single users
(e.g. concerns regarding a particular linking), @
In the activity section an overview of the comments is displayed. Please mark a comment as solved if it does not need a review by another user.
Use comments if you can’t solve an issue yourself or are not sure about the application of the coding rules (e.g. re-linking a party, updating core parties information).
We review all comments and mark them as solved once the issue has been addressed.
2023-12-15
2022-12-16
2021-12-07
2021-07-29
2021-03-09
2020-07-17
2020-07-12
2020-06-08
2019-11-28
2019-06-07
2019-06-07
2019-01-11
2018-10-09
2018-09-24
2018-08-22
2018-06-18
2017-04-04
2017-06-18
2017-03-16
2016-09-19
2016-05-01
2016-02-02
2016-01-04
2015-11-18
2021-07-15
2021-03-08
Documentation of automatic data changes since November 2018 (user robot).
2021-03-08 — 2021-03-13
2018-11-17
2018-11-11
2018-11-03