Comparable 2022 General Election Advertising Datasets from Meta and Google.

Area of Science:

• Political Science
• Communication Studies
• Data Science

Background:

• Digital advertising significantly influences U.S. federal elections.
• Understanding online political campaigns requires accessible, comprehensive data.

Purpose of the Study:

• To introduce two novel datasets detailing digital ads from Meta and Google during the 2022 U.S. midterm elections.
• To provide researchers with structured data for analyzing digital political advertising.

Main Methods:

• Data collection via ad transparency libraries and web scraping from Meta (Facebook, Instagram) and Google (YouTube).
• Processing included automatic speech recognition (ASR), face recognition, and optical character recognition (OCR).
• Data enrichment through labeling and classification tasks for enhanced comparability and utility.

Main Results:

• Creation of two comprehensive datasets on U.S. federal election digital ads (2022 midterm).
• Datasets contain metadata, ASR transcripts, OCR text, face recognition data, and classification labels.
• The processed data enables detailed analysis of ad content and campaign strategies.

Conclusions:

• The datasets offer valuable resources for analyzing the digital election advertising landscape.
• Facilitates research in political science, communication, and data science for insights into digital political campaigns.
• Enables deeper understanding of campaign strategies and public engagement in online environments.