German Political Facebook Ad Analysis During 2021 Election

4,770 political advertisements from the 2021 German federal election were manually coded by the Digital Pandemic Campaigning project. The dataset, created by Simon Kruschinski, covers ads from seven major parties and five lead candidates run on Facebook and Instagram. Ads were collected and analyzed during the hot campaign phase from August 30 to September 27, 2021.

Use Cases

• Analyze correlations between ad format and coded mobilization strategies.
• Train classifiers to predict policy topics from coded ad content variables.
• Investigate patterns of negative campaigning by examining coded attack targets and strategies.
• Study audience targeting by modeling relationships between coded demographic targets and ideological positioning.
• Assess the prevalence of populist messaging by analyzing coded dimensions for anti-elitism and people-centrism.

Strengths

• Contains 4,770 manually coded political advertisements.
• Covers ads from seven political parties and five lead candidates.
• Includes coded dimensions across 10+ categories like policy topics and populism.
• Ads were collected during a defined 29-day hot campaign phase.

Limitations

• Dataset does not contain the original ad text or media content, only derived variables.
• The standardized codebook is not publicly shareable, limiting reproducibility.
• Sample is limited to one election cycle and one country.

Provenance

Source

Digital Pandemic Campaigning (DiPaCa) project, Simon Kruschinski.

Collection Method

Manual content analysis of ads collected from the Meta Ad Library.

Time Range

August 30, 2021 to September 27, 2021.

Freshness

Data covers a fixed period in 2021; dataset was last updated in March 2026.

Geography

Germany.