The Transparency of Facebook Disclaimers in the 2022 Lebanon Elections

A sketch on a background image, mostly in turqoise color and with various shapes
In this article, Rayan Shaya examines the effectiveness of the political advertisement disclaimers on Facebook and Instagram during the 2022 elections. Using data collected in the Meta Ad Library, the team developed a framework to see how the type of disclaimers intersected with the type of attributed person or page.

Lebanon is a parliamentary democratic republic that held its most recent parliamentary elections in May 2022. These elections followed the 17 October Revolution in 2019, during which there were widespread and lengthy demonstrations against the political elite. Although the Lebanese population's first-hand experience with economic and monetary crises served as the impetus for these protests, some identified the underlying causes as a structurally defective economic system, scandalous political practices, and corruption that has been tolerated by successive administrations for decades (Arab NGO Network for Development).

The 17 October Revolution empowered civil society, bringing together diverse groups of citizens who expressed their frustrations about a corrupt political class. The movement led to the emergence of independent candidates and civil society organizations campaigning for social justice and anti-corruption measures. The revolution also sparked a culture of civic engagement, reshaping Lebanon's political landscape and fostering a more dynamic environment, which has ultimately bolstered the strength of the opposition.

Political campaigning has become critical in the increasingly heated elections. Previously, traditional media (news stories, TV and out-of-home advertisements) were mostly dominated by the long-standing parties and wealthy candidates who could afford advertisements even in times of profound crisis. This skewing of advertisements towards the wealthy and legacy parties has led to a particular eagerness among candidates to find outlets outside of the traditional media and pursue a large public reach on social media platforms, including Facebook and Instagram (which are both owned by Meta). Furthermore, the presence of opposition candidates and parties on these social media platforms has compelled the traditional parties to multiply and strengthen their campaigning efforts. As the political parties rely more heavily on social media platforms and international attention to misinformation during political campaigns around the world rises, our team aimed to understand if the campaign messages which ran during the Lebanese election through Facebook were being appropriately attributed, or rather, if the measures in place by Facebook were accomplishing transparency.

Our research investigates the transparency of disclaimers in paid political advertisements on Facebook and Instagram: the description of the person, organisation, or ‘page’ (the account for a business, organization, or institution) responsible for paid advertisements (from here on ‘ads’) run on the platforms during the Lebanese parliamentary elections in 2022. The Business Help Center for Facebook.com states that ad transparency tools are available in 243 countries around the world, including Lebanon. Some countries have to include disclaimers for “social issues, elections and politics” though for Lebanon only ads related to “elections and politics” have mandatory disclaimer requirements (Facebook Business Center). Our main research tool was Meta Ad Library. The Ad Library is a platform that displays all ads run through Meta products (Facebook, Instagram and Messenger). The Ad Library aims to increase transparency on every ad that deals with “social issues, elections, or politics” (Facebook Ad Library). While all ads on the platform are stored in the library for seven years and are searchable even if they are inactive, in Lebanon, Meta only rolled out its Ad Library in March 2022 to provide more transparency on Facebook and Instagram political campaign spending during the election year.

Methodology

To first narrow down the ads related to the Lebanese elections, our team decided to use the keyword "مرشح" which translates to “Candidate” in Arabic. We had two other keyword options, “elections” and “campaigns” but these were ultimately too broad. With the broad keywords, we risked extracting posts from different types of elections or campaigns outside of national political elections (universities, sport clubs, and awareness campaigns, for example). Therefore, “Candidate” was selected to represent the content and focus of our study. Using “Candidate,” we extracted a total of 1,155 posts and they were distributed as follows:

Pie chart showing the breakdown of irrelevant advertisements, advertisements without disclaimers and the number of analysed advertisementsFigure 1: Types and number of extracted advertisements

As Figure.1 shows, seven posts were irrelevant to the elections or politics, 117 posts ran ‘Without Disclaimer’ and were subsequently taken down from the platforms, which left us with 1,031 posts to analyse.

The images below show two different ads: Figure.2.1 represents a typical ad from a candidate, with a disclaimer appearing on the ad. Figure.2.2 represents an ad without a disclaimer, which was removed by Meta.

Sample Facebook advertisement with name and identifying features removed showing the term candidate in ArabicFigure 2.1: Facebook ad showing a political disclaimer Sample Facebook advertisement that was removed by Meta due to lack of a disclaimerFigure 2.2: Ad without disclaimer that was removed by Meta

In order to begin categorising the ads, our analysis followed these steps:

  • Extracting the data from Meta Library and transforming them to Excel files (.xlsx)

  • Opening each advertisement to determine and categorise the content and key information about owner and type in relation to:

    • Relevancy of the advertisement to politics
    • Dividing the type of pages by answering the questions “what is this page?” and “who is the user?”
    • Investigating the types of disclaimers by answering “what is the disclaimer?”
  • Analysing the results and statistically applying them to graphics and findings

After opening and categorising each of the 1,031 posts, the next step in our investigation was to code the types of disclaimers. The final list and codes can be seen in Figure 3:

Pie chart showing the types of disclaimers by percentagesFigure 3: Types of disclaimers by percentage

Figure3 shows that ads from the “Candidate” category slightly dominate, with 20.08% of the total ads. Ads with these disclaimers clearly demonstrate that they originated from the political candidates themselves or by their direct team. It is relatively easy to sponsor an ad via Meta with an email, confirmed phone number and a legal ID number. After an ad is submitted to Facebook with the necessary information, it is placed under review by an internal Facebook team, verified, and then published.

However, the categories themselves do not provide enough information or context to know which group is producing and promoting more content, what type of content this is, or how influential that content is. For example, while “Candidate” has 20.08% of the ads, which is slightly more than “Ad Firms” with 18.62%, the content and type of ads can also make a difference to our understanding of how effective their political influence campaigns are. More research into the context, networks and technology is needed, which we examine through the types of pages connected to disclaimers. See the difference in the role of the type of disclaimers in regards to the pages they sponsor in figures 4.1 and 4.2:

Pie chart showing the breakdown of types of pages by  an adfirmFigure 4.1: Types of pages as reported by the Ad Firm disclaimers Bar graph showing the type of pages candidates claimedFigure 4.2: Types of disclaimers as reported by the candidate disclaimers

The above Figures (4.1 and 4.2) reveal that a “Candidate Disclaimer” only sponsors the “Candidate Page” page itself; whereas an “Ad Firm Disclaimer” varies in which types of pages they sponsor candidate ads from, such as political, personal, social and even commercial entities like a restaurant.

Considering that two of Meta’s platforms (Facebook and Instagram) are a prominent space with the biggest reach in the modern world [Facebook had 2.98 billion while Instagram had 1.62 billion monthly active users in April 2023 (Kepios, April 2023)] advertising firms play a big role on these platforms in generating revenue, or votes, through targeted ad placement. Advertising firms study the best strategies for the widest and most accurate reach possible, based on data such as recorded interests, demographics and geographic location – and have the advantage of often working over many years, campaigns, and even countries to gather this information. By reaching audiences that are supposedly the most likely to be swayed by an advertisement, the advertising firms can increase the number of clicks, site visits and, in our instance, potentially votes.

We further categorised the types of groups behind disclaimers to identify more details, which matched up to three categories: Advertisement, Political, and Others (See below in table 1). For more understanding on the relationship between the type of disclaimers and the type of pages on Meta platforms, we grouped the types of pages also into three categories: Political, News, and Others (See below in table 2).

Table demonstrating the breakdown of the Advertisement, Political and Others disclaimer categoriesTable 1. Definition of disclaimer types Table demonstrating the breakdown of the Political, News and Other page categoriesTable 1. Definition of pages sponsoring disclaimers

Our team wanted to visualise both the categories and the relationships between each grouping to better understand from where the ads were originating and if the political ads were clearly marked from political pages. The two pie charts below show the percentage of the grouped type of disclaimers (Figure.5.1) and percentage of the grouped type of pages (Figure.5.2). The distribution of the grouped disclaimers and the pages shows that the “Political” category is dominant in both. From this insight, our team concluded that generally people who are involved in politics are comfortable adding the disclaimer on political or pages related to politics.

Pie chart showing the percentage of grouped disclaimers as per the above table 1Figure 5.1: Percentage of grouped disclaimers Pie chart showing the percentage of grouped pages as per the above table 2Figure 5.2: Percentage of grouped pages

However, we can see there are more diverse connections among the advertising firms or single individual page users and owners (in ‘others) and what type of page they sponsor. Figure 8 shows how the “Political” disclaimers only appear on political pages whereas the “Others” disclaimers can be seen across all types of page categories.

Flowchart showing the connections between types of pages and types of disclaimersFigure 8: Flowchart showing relationship between page and disclaimer types

Even when the disclaimer was not visibly affiliated with politics but was nevertheless promoting a “Political Page,” it was listed as a “News & Media Page,” which was in fact promoting a candidate’s page. Our team felt this shows less transparency of intent from the disclaimers made by News outlets as it lends the news institutions credibility to the ad and obscures the fact that the ad promoting a candidate.

For example, out of the five posts we found that were stated to be paid for by “Others” in connection to “Political,” four were connected with the “News and Media” category and only one had a disclaimer from the “Page Itself,” though this page was, to our team, already obviously affiliated with one party. In the other four cases, the pages had a disclaimer with the name of the page and not the party, person or company sponsoring the ad., clouding the original transparency. This is particularly true in situations in which the parties or candidates have numerous pages that are created or aimed at discrediting others or campaigning for candidate. The more pages there are, the less viewers are able to quickly discern if the disclaimer is spreading correct information about what political group they are connected with or who is paying for the pages. That the originator of the disclaimer can be the name of the page, not of the person or company sponsoring the disclaimer, highlights a limitation of the Ad Library. In the course of our research, we found out that some “News & Media” are obviously affiliated with some candidates and parties, but do not come with clear statements or declarations.

We could also see from the “Advertisement” disclaimer group that the advertising firms, freelancers and private ad firms in which these types of companies created advertisements for all types of pages – political, news, social groups, and corporate organisations, showing the importance of these agencies and freelancers in ad sponsorship across Meta’s social media platforms. These firms are also therefore involved in the spread of election messaging, without necessarily being directly connected to a political group or campaign.

Advertising or marketing firms are more aware and knowledgeable in reach strategies including targeting political, news, social, and personal pages beyond just the candidate’s page, which make them a popular alternative for the candidates or parties that can afford to pay for these targeted services.

Sponsoring advertisements via Meta platforms is easy to do and available for any accounts, which leads to a dominance – and transparency - of direct advertisements from candidates. Parties are able to run pages under other names to discredit opponent parties and candidates, and it’s difficult for readers to link those accounts to the original creators, which obscures Meta’s original goal of transparency. Based on this research, we recommend a new restriction or update from Meta that compels users in Lebanon who create the disclaimers to show the exact name of the individual or page admin as shown on an official ID or bank username transactions, instead of being able to show any fake name or only a page name as a disclaimer.


Suggested Readings

Digital Listening: Insights from social media: an explainer report by the Data and Politics team at Tactical Tech

Tools of the Influence Industry: Insights into the Influence Industry research by Tactical Tech

References


About the Author:

Rayan Shaya is a mechanical engineer turned management consultant with experience in investigating electoral processes, including the 2022 Lebanese elections with the European Union, which has enhanced his skills in detecting fake news and hate speech. This amplifies his ability to promote transparency and combat misinformation. He is also socially and politically active, passionately advocating for positive change in his community. Therefore, Rayan has actively participated in multiple programs focused on youth participation in change-making. To connect with Rayan, find him on LinkedIn at Rayan.Shaya or Instagram at @rayanshaya.


If you want to find out more about the firms that support political parties engage with Google advertising, head over to The Influence Industry Explorer.

The influence industry is led since 2016 by Tactical Tech’s Data and Politics team addressing the pervasive data-driven technologies used by political groups within elections and political campaigns.

First published: June 21, 2023

Read another case study

  • by Glyn Thomas

    Google Ads in Election Campaigns

    In this article, we will look in depth at the different advertising possibilities offered by Google and YouTube, how they are used by political campaigns and what exactly consultants and agencies might be asked to do by a candidate or party.

    Read more +

Learn more about developing research