Research / Campaign Casebook
The Campaign Casebook.
One page for each election: what public data show about how candidates and campaigns use social media, and what the same data cannot show. Eighteen elections in fifteen countries, each read on its own terms.
The Casebook describes the data. It does not rank countries, parties or campaigns, and it does not claim that any account or message changed a vote.
How to read it
Three rules for every case.
Public social media data are patchy, collected for different purposes and rarely comparable. The Casebook keeps cases apart so that a gap in one is never filled from another.
Rule 1
One election, one page
Each case is built only from that election's own files. Nothing is added, averaged or ranked across cases.
Rule 2
Source and limits shown
Every chart names its source file. Every case lists what its data cannot show, such as posts, reach, dates or votes.
Rule 3
Patterns need three cases
A pattern needs at least three cases that agree. Fewer is only "emerging", and conflicting results are shown as "varies".
The cases
Eighteen elections.
Choose an election to open its page below. The tag on each card shows how deep the data go: some hold posts, some only counts, some only a list of candidate accounts.
Africa 3 cases
Nigeria 2023
Presidential election 2023
Hashtag-defined camps for the three leading candidates, followed from August 2022 into the last week of the campaign.
Posts + timelineGhana 2020
Presidential election 2020
A large pool of tweets about the two leading candidates, a small part labelled by stance, and what the pool looks like. The files have no dates, so there is no timeline.
Posts + stance labelsTunisia 2022
Parliamentary election 2022-23
Candidates for the 2022-23 Assembly election and whether they list an X account.
Account list + context
Asia 5 cases
India 2022
State Assembly elections 2022 (Goa, Manipur, Punjab, Uttar Pradesh, Uttarakhand)
Which candidates in five state elections list an X handle, and what country experts say about the wider information environment.
Account list + contextTurkey 2023
Presidential and parliamentary elections 2023
Which political accounts one research team followed for the 2023 Turkish elections. The data hold the list of accounts, not the tweets.
Account listJapan 2021
General election 2021
Candidates for the 2021 House of Representatives election and whether they list an X account.
Account list + contextNepal 2022
General election 2022
Candidates for the 2022 House of Representatives and whether they list an X account.
Account list + contextPhilippines 2022
National and local elections 2022
Candidates at national, provincial and local level in 2022 and whether they list an X account.
Account list + context
Europe 5 cases
France 2022 (presidential)
Presidential election 2022 (two rounds)
How the volume of election talk on X rose and fell around the two rounds, shown here as three headline counts only.
Activity countsItaly 2022
General election 2022
A four-month, three-platform view of how election conversation built up and peaked around the vote.
Activity counts + accountsUnited Kingdom 2024
General election 2024 (and 2019)
A method case: the accounts, keywords and 32-narrative codebook one team used to study election narratives on X. No tweets are in the data.
Method onlyGermany 2021
Federal election 2021
Candidates for the 2021 Bundestag election and whether they list an X account.
Account list + contextFrance 2022 (legislative)
Legislative elections 2022
Candidates for the 2022 National Assembly elections and whether they list an X account.
Account list + context
North America 2 cases
United States 2024
Federal elections 2024 (House and Senate candidates)
Which 2024 federal candidates run campaign and office accounts on which platforms, including the alternative ones, handles only; the 3.1 million posts are held elsewhere.
Account listUnited States 2022
Midterm elections 2022
Candidates in the 2022 federal and state races and whether they list an X account.
Account list + context
South America 2 cases
Selected case
Across cases
One comparison that can be tested.
Few questions can be asked the same way in several elections. This one can: whether winning candidates are more likely than others to have an X account. The result shows who has an account, not that having one helps win.
Related
From cases to a working picture.
The Casebook shows what public data can and cannot tell us about campaigns. Polity Lab's election data and strategy work start from the same discipline: say what the evidence supports, and where it stops.