Overview

The Cross-Platform Election Advertising Transparency Initiative (hereafter CREATIVE) pilot infrastructure provides cross-platform integration and standardization of electoral digital advertising data for community use. Election advertising is a key link between candidates and the public. Candidates for federal office in 2022 spent an estimated $150 million on ads on two key platforms:  Meta and Google.  The rapid shift in election spending to digital, however, has caused headaches for those who track and study advertising. Transparency in digital ad spending and content is difficult to obtain and often illusory in part because campaign finance laws require minimal disclosure for many ad sponsors -- and no disclosure for some -- meaning that content tracking across media is the only viable way of uncovering some advertising activity.  The Meta and Google ad libraries are important steps in providing more transparency; however, their data come in different formats with different universes, are not user-friendly for research purposes, lack basic metadata that would enable one to find all ads related to a specific campaign for office, and often require the use of novel computational techniques to extract large-scale data. 

CREATIVE enables easy-to-access, shared baseline data for the 2022 midterm general election period, code and information critical to answering vital research questions about elections and democracy, including the role of `dark money' spending and the scope and scale of political misinformation. In addition, CREATIVE infrastructure reduces costs to individual researchers, provides parallel procedures already in place for analyzing TV advertising to maximize comparability to existing data, provides linkage information to other data sources, and has produced cumulative data to benefit the community. The result is a shared knowledge base with the potential for transformative science. Our specific aims included 1) acquiring digital advertising across platforms (including content, spending, impression and sponsor names) along with candidate and sponsor data for the 2022 federal, general election period and 2) processing and integrating data through human and computational methods to produce validated labels for analysis and distribution. 

Intellectual Merit

Scholars seeking to analyze election advertising data frequently perform similar tasks to gather and analyze digital advertising activity, and because of the demands of grappling with incomparable universes across platforms that also lack uniform identification for ease of integration, no one has attempted to systematically understand the universe of federal election activity across platforms. Integrated data on digital ads is essential for a complete picture of how candidates, parties, and outside groups appeal to voters and what the effects of these appeals are. Information on how the volume and content of online activity compares within and across platforms -- along with how it compares with television -- is also required to answer long-standing questions about advertising effects, and to understand how online and social media advertising affect democracy, political polarization and inequality. Current piecemeal efforts by individual scholars to analyze digital advertising end up repeating expensive and time-consuming tasks that exacerbate inequalities in who among the research community is able to produce scientific knowledge given that data acquisition, processing and integration is tedious, computing costs are high, and most social scientists lack the necessary computational skills and expertise to address the problem in a comprehensive fashion.

Broader Impacts

As spending on U.S. election advertising continues to increase year after year -- and the number of ads on digital and social media platforms along with the number of sponsors of ad activity explodes -- the ability to analyze ad spending and content in a systematic fashion is crucial. CREATIVE infrastructure provides community data to help accelerate knowledge and add to our understanding of how candidates, parties and groups tailor and distribute their advertising online. Our datasets can be easily connected via linking identifiers with other government and social science data sources that enable scholars to answer questions about the role of `dark money,' the spread and reach of information in campaigns, including disinformation, among other questions. Furthermore, our validation, labeling and standardization of ad data across platforms provides high-quality, labeled textual and audiovisual data along with human validation that is accessible to researchers and may facilitate work in other fields, including in AI.

Last Modified: 01/28/2025
Modified by: Erika Franklin Fowler