Information Disorder Analysis CentreMainstreamMedia LiteracyResearch

A typology and distribution analysis of information disorder on the 2023 general elections in Nigeria

By Kemi Busari and Silas Jonathan


The study analysed claims about the 2023 general elections extracted from fact-check articles by three fact-checking organisations. The findings revealed that most of the claims were directly related to the candidates, were propagated across different social media platforms, and most contentious claims originated from groups/pages, political influencers and individuals on various social media platforms. Significantly, the study found that the claims were targeted to either promote or decimate the appeal of the leading presidential candidates. The results also showed that most claims/ misinformation that was fact-checked originated from Facebook, while Twitter was primarily deployed in their propagation. The study anticipates more dispensation of false information before the elections.

1.0 Introduction 

The usage of social media in Nigeria has increased significantly over the last few years. Statista estimates the number of active users in Nigeria in 2022 to be 34 million, from the 18 million recorded in 2017. Although some studies note social media’s importance for the vital role it plays in the democratic process, others worry about their utilisation by political players and their potential to impact democratic outcomes (Mutsvairo & Rønning, 2020). This role of social media becomes critical, especially in the build-up to major elections when platforms are weaponised by political actors to spread false information, thereby depriving electorates and citizens of accurate and factual information when it is most needed. 

Wasserman and Madrid-Morales, in a 2019 study, noted the usage of online disinformation campaigns to influence political agendas in Nigeria, Kenya, and South Africa. They posited that exposure to such information on social media, especially during electoral campaigns, leads to distrust among users. This aligns with a study by Apuke and Omar (2020), which found that false information exacerbated tensions, caused conflict and intensified regional and religious crises during the 2019 Nigerian electioneering campaign season. Igwebuike (2020) finds that political actors deployed different means to legitimise falsehoods in the build-up to the 2019 general elections.

Fact-checking and media literacy are the two main strategies currently used at the global level to address the spread of political disinformation, misinformation, and manipulation (Dittrich, 2019). Understanding the types of disinformation and misinformation being addressed by fact-checkers and where they flourish on social media is essential. Therefore, this study seeks to analyse and discuss the typologies of political misinformation found by fact-checkers ahead of the 2023 Nigerian general elections. Further, the social networks of various actors used to spread misinformation and disinformation aimed at shaping voting behaviour are also explored in this study. 

2.0 Methodology and conceptualisation

2.1 Methodology 

This study aimed to answer two questions: 

  1. What was the nature and prevalence of political misinformation about the 2023 general elections?
  2. What are the sources of political misinformation about the 2023 general elections?
  3. Which social networks were engaged in spreading political misinformation within the study period?

Data was primarily sourced from published fact-check reports from three fact-checking outlets signatories of the International Fact-checking Network (IFCN): DUBAWA, FactCheckHub, and Africa Check. A search was launched on each website with the keywords ‘2023 elections’, ‘2023 Nigerian elections’, ‘general elections’, ‘presidential candidates’, ‘political parties’, and ‘general elections 2023’ to extract fact-check articles about the elections. Out of the 54 articles found, only 50 were published during the study period – between March 1, 2022, and September 1, 2022 – and formed part of the study sample. Sixteen reports came from DUBAWA, 15 from Africa Check, and 19 from FactCheckHub. All the fact-check reports were unique to platforms; if two organisations fact-checked the same claim, only one such was chosen.

The selection of this period is predicated upon the timetable of the Independent National Electoral Commission (INEC), the body saddled with the responsibility of conducting free, fair, and credible elections for sustainable democracy in Nigeria. March 1, 2022, marked the official commencement of political activities as authorised by INEC. Within the six-month study period, political parties engaged in primary elections at federal and state levels and submitted the names of victorious candidates to INEC. Within this same period, INEC published the names of candidates submitted and provided an opportunity for substitutions to be made. Several online and offline discussions accompanied these exercises, and fact-checkers were engaged in determining the veracity of claims made.

Content analysis was conducted to draw insights and patterns from the 50 fact-checks collated. The study focused on the claim that led to the fact-checks and not the fact-checks themselves. Additionally, the claims that led to the fact-check reports were categorised by the kind of sources they originated from to identify the personalities that first shared them. To ensure the validity and reliability of the samples collected, Tweetdeck was used to track the fact-check articles on the Twitter pages of the three organisations within the stipulated period; fact-checking organisations usually post whatever they publish on their Twitter handles for visibility purposes. Ultimately, a correlation between the keywords searched on their websites and the results from Tweetdeck was established. 

A social network distribution of the data (collated fact-checks) was then conducted to track the claims spread across social media platforms. To achieve this, each fact-check article was treated as an independent entity to track where a claim originated and its journey to other associated platforms. Other related characters who further spread and promoted a claim to other social media platforms were also identified, together with the unique platform they used. Claimants identified to have shared misinformation more than thrice were treated as unique entities to understand whether they were inorganic or unauthentically dissipated. 

Sentiment analysis of all the claims using Hugging Face machine learning and text classification models was conducted to understand the claims’ affiliations. Other tools, including Tweetdeck, Crowdtangle, Wayback machine, Google search, and Excel, were utilised for data collection, classification of claims by the verdict, tracking the chain network of claim distribution, and the main actors actively spreading misinformation in the build-up to the 2023 general elections.  

2.2 Conceptualisation

2.2.1 Platforms for distribution

The study categorised the data collected based on the platform from which it first emerged: 

  • Twitter: This platform has been noteworthy in hosting political discussions towards the 2023 general elections. Hashtags such as #2023elections, #obidients #BAT, #Atiku, and #Kwankwaso among others, have trended on the platform. These are hashtags directly related to the 2023 general elections in Nigeria. 
  • Facebook: It is the largest social media platform globally and was one of the most used social media platforms by Nigerians in 2022. Facebook allows for closed and open groups/pages to be formed. 
  • Instagram and TikTok:   These platforms share some correlation in content distribution. Claims shared in video formats can easily reach many users in the shortest time possible.
  • WhatsApp and Telegram: Both social media platforms are closed. However, claims can easily be reshared between users and groups. 
  • Blogs and websites: These platforms are also effective channels for sharing misinformation and rumours. Interactive blogs such as Nairaland have millions of active visitors who are fully engaged on every topic shared on the forum. 

2.2.2 Other related terms 

Other terms used in this study are explained below:

  • Misinformation: when false information is shared without an intention to cause harm.
  • Disinformation: when false information is knowingly shared to cause harm.
  • Fact-check article/ report: The published copy of verified information with a specified verdict.    
  • Claim: Statements of fact that can be subjected to verification or fact-checking.
  • Claimants/ Characters: The source of a claim, misinformation or disinformation that is subjected to a fact-check. In this case, the term refers to the first person who shared the claim. 
  • Associations (associates): Other characters who shared, retweeted, or forwarded the report on other platforms. This includes other characters who are party to the chain network of a claim distribution. 
  • Coordinated inauthentic behaviour: Efforts made by accounts seeking to influence behaviour by relying on fake accounts and a chain network of users across various social media platforms.
  • Platform: This refers to the channel or medium the claimant or associates used to share the claim, that is, Twitter, WhatsApp, Facebook, and others.
  • Verdict: This is the conclusion of the verification carried out on a claim. Although the verdicts system varies for DUBAWA, FactCheckHub, and Africa Check, the categorisation of the verdict in this study is based upon the general verdicts used by the organisations. Therefore, the following verdicts are used across the organisations for the 50 fact-check articles:
    • True: A fact-check is deemed true when all elements of such a claim pertain to factual information. It is also used contextually and verifiably at the time of assertion.
  • False: A fact-check is deemed false when all elements of such a claim do not pertain to factual information at the time of assertion. In essence, imposter, manipulated, and fabricated content are considered false.
  • Misleading: A fact-check is deemed misleading when elements of a claim are too complex to be termed true or false.
  • Insufficient evidence: When the claim(s) is unverifiable, usually about urban myths or quantifiable data. 

3.0 Limitations of the study

The major limitation of this study is the small sample size arising from the reliance on fact-check reports published by just three fact-checking organisations: DUBAWA, FactCheckHub, and Africa Check. This was necessary to examine claims already debunked or verified by credible outlets. The three fact-checking organisations presented themselves as the best sources for such data due to their membership in the International Fact-Checking Network (IFCN) and subscription to its code of principles.

Another limitation of the study is the predominance of fact-checks rated as false. This is unavoidable since fact-checkers are predisposed to select claims for verification that are likely to be rated as false. 

Finally, this study is non-generalisable due to the earlier limitations stated.

4.0 Key findings 

Fifty (50) reports from three fact-checking organisations were analysed to understand the nature and prevalence of political misinformation about the 2023 general elections. To commence the analysis, the sources of the claims were tracked and categorised by the nature of the claimant. Six main claimants were identified: political parties, candidates, individuals/ citizens, political influencers (clearly affiliated with a political party or promoting a particular candidate), groups (social media pages/ groups) and others (claims from WhatsApp, Telegram or those whose origin could not be traced).

Unsurprisingly, the majority of claims tracked (84%) were rated false by the fact-checkers, 12 per cent as misleading and the remaining 4% as true. Fact-checkers tend to focus on potentially false claims for verification, while true claims are only subjected to a test when they are controversial.

4.1 Typology of claimants based on their nature

The study found that accounts categorised as “groups” shared the most claims. Those categorised as “individuals” and “influencers” also shared many claims. These three categories were the original sources of over 80% of the claims.  Significantly, only three claims were found to have been shared by the major presidential aspirants or their running mates – one claim each by Peter Obi of the Labour Party, Kashim Shettima of the All Progressives Congress (APC), and Atiku Abubakar of the Peoples Democratic Party. All three claims were found to have been originally on YouTube, categorised under websites for this analysis. These claims were also noticed to have been made during interviews that featured the political candidates. 

Figure 1: Categories of claimants

4.2 Dominant topics/subjects

A significant number of the claims revolved around the major candidates and their political parties, that is, Peter Obi, Labour Party (LP); Atiku Abubakar, People’s Democratic Party (PDP); Bola Tinubu, All Progressives Congress (APC); and Rabiu Kwankwaso, New Nigeria People’s Party (NNPP). As indicated in the chart below, close to half of the claims (24) were related to Peter Obi or his party, followed by Atiku Abubakar (12 claims) and Bola Tinubu (six claims). Other emerging topics included claims about the Personal Voter’s Card (PVC), INEC and other political elements.  Notably, aside from Kashim Shettima, Bola Tinubu’s running mate, and the four major candidates, no other candidate was featured in the claims. There seems to be a concentration on these leading presidential candidates who dominate discussions in the media space. 

Figure 2: Topic of claims.

4.3 Dominant sources of claims/ misinformation

An examination of the claimant’s activities revealed that some accounts repeatedly shared false information. For instance, three accounts shared false information more than once on their platforms. Aside from that, there appears to be continuous activity around the propagation of controversial information on these platforms. Celebrity vibes, a Facebook page, repeatedly shared information that was all confirmed to be false. Another Facebook page, Prime guard, and a politician/political appointee, Joe Igbokwe, shared information fact-checked to be false more than once. These pages are all on Facebook, and while they appeared non-political, they promoted some presidential candidates. Primeguard and Celebrity vibes, for instance, shared controversial and unverified pro-Peter Obi and pro-Labour party claims. On the other hand, Joe Igbokwe, described as a public figure on his profile, shared pro-Tinubu content. 

Aside from the multiple fact-checked claims sourced from them, the three pages were all riddled with unverified allegations that appeared to promote their preferred candidates and taunt their opposition. For example, Joe Igbokwe, an apparent pro-Tinubu supporter, shared claims that targeted mainly Atiku Abubakar and Peter Obi, the latter mildly. Prime guard, a pro-Peter Obi page, also shared claims targeted at Atiku Abubakar, while Celebrity Vibes, also a pro-Peter Obi page, shared allegations against Bola Tinubu.  

Figure 3: Claim topic per most common claimant.


4.4 Accounts associated with claimants found to have repeatedly shared claims

The study found that some accounts were disseminating misinformation across several pages. Seven accounts played a significant role in propagating and reposting false claims from Celebrities vibes and Primeguard: Peter Obi support group, Peter Obi fans, Peter Obi for president, Peter Obi Labour support, Peter Obi for youths, Peter Obi Labour Supporters and Obidientgists. They were all on Facebook.

With the name “Peter Obi” and multiple pro-Peter Obi posts on their timeline, it appears these accounts shared a coordination network and acted as support groups for Mr Peter Obi. 

4.5 Frequency of platforms in the distribution of claims by claimants

The analysis shows that over half of the claims examined in this study originated from Facebook; all but one was false, and another was misleading. Twitter came next as a distant second. This finding is similar to the Portuguese experience. Baptista and Gradim (2022) found that Facebook was the top platform for spreading false information during the elections and was used to influence people and promote political discourses in the country. 

Several reasons could have accounted for the high number of claims sourced from Facebook; two of the organisations, DUBAWA and Africa Check, are members of Meta’s Third Party Fact-checking Programme and may have been able to find claims more easily on Facebook than any other. Secondly, as mentioned earlier, Facebook was one of the most used platforms by Nigerians in 2022, thereby increasing the possibility of false claims on the platform. According to Statista, there were nearly 37 million Facebook users in Nigeria as of July 2022, accounting for 86% of all social media users and 16.8% of the Nigerian population.  Finally, it is also possible that the fact-checkers paid significant attention to Facebook for the foregoing reasons.

Figure 4: Most used platforms to share claims

4.6 Spread of claims from original source to other platforms

Aside from the original media they were first posted, a number of claims were found to have re-appeared on or travelled to different platforms to varying degrees. Some were reshared to as high as four different platforms. Eight claims were traced to four other platforms. Half (four) were tracked to have originated from Twitter, and three from Facebook. On average, claims from any platform were likely to spread to at least two other platforms. Figure 5 shows that 14 of the 50 claims were reshared to two other platforms aside from the original source, 13 claims resurfaced on three other media, and eight on four other outlets.

Figure 5: Number of other platforms a claim was further reshared to

4.7 Major platforms where associates most reshared claims 

Twitter was the most used platform for propagating claims from other platforms, mainly Facebook. Regardless of where claims originated, 80% of claims were reshared to Twitter. Additionally, many of the claims were discovered to have moved from Facebook to websites. A correlation was found between Facebook groups’ claims and others on websites. It can be hypothesised that the websites own these groups on Facebook. Most of the Facebook groups tended to act as blogs, with reports transferred to websites and vice versa. Only 23% of the claims resurfaced on WhatsApp/Telegram. This could be due to the difficulty fact-checkers face in gathering claims from WhatsApp and Telegram, as they are more private platforms than the others.  

4.8 Interactions with claims across platforms 

In analysing the interactions with claims from original posts and their replications on all platforms, the findings showed that those originating from Facebook had the most interactions across all platforms. Claims identified to originate from websites were noted to have been replicated and reposted to other platforms. While claims from Twitter also had a good number of interactions, it had fewer numbers compared with Facebook. In fact, some claims initially shared on Twitter garnered more interactions when the replicated versions were found on Facebook. This may be because of the high number of Facebook users in Nigeria.

Figure 6: Interactions across platforms

4.9 Disaggregating the data on platforms by the topic or person of interest for the claim 

As shown in Figure 7, Facebook was predominantly used in disseminating claims about the four major presidential aspirants. However, claims related to Peter Obi and his party were found on Twitter to be equal to, or more than, the number of claims other aspirants had on any single platform. This implies the popularity Peter Obi has on the two major platforms. Although claims related to APC and its candidate, Bola Tinubu, were common on Facebook, claims about PDP’s Atiku Abubakar cut across all the platforms outlined for the study. 

Figure 7: Target of claims per platform

5.0 Sentiments Analysis

A sentiment analysis on all 50 claims extracted for the research was performed using Hugging Face machine learning text classification models. The model predicted these sentiments with confidence scores of over 90% in 46 of the 50 claims and a minimum confidence score of 59.4%. Our analysis of the model’s predictions showed that most claims were of negative sentiments. With a large majority of the claims fact-checked being false, this implies that most false claims are put out with less than favourable intentions. 

Figure 8: Sentiment analysis of claims

Figure 9 shows the percentage of the claims by presidential candidates or political parties rated as negative or positive sentiments. While the analysis showed Peter Obi and his party had the highest negative and positive toned claims in quantity, the ratio/percentages tell a different story. The data shows that claims regarding Atiku Abubakar and Bola Ahmed Tinubu were more likely to be negative sentiments than claims regarding Peter Obi. This means that posts that sought to portray Atiku Abubakar and Bola Tinubu falsely were more than claims that falsely portray Peter Obi or Rabiu Kwankwaso.

Figure 9: Target of claims by sentiments

6.0 Discussion/Summary of findings

This study sought to identify the typologies of political misinformation and the social networks of various actors used to spread false information aimed at shaping voting behaviour ahead of the 2023 Nigerian general elections. A content analysis of the 50 fact-check articles sampled revealed that claims that were verified revolved around the four major candidates and their political parties. We found 24 claims directly related to Peter Obi and his political party, Labour Party, while 12 were associated with Atiku and his party, PDP. Tinubu and his party, APC, featured in six claims, while Kwankwaso pointed up only twice, and other topics, such as PVC and INEC, also made six topics. 

A network of coordinated activities was traced to particular claimants. Celebrity vibes, a Facebook page which described itself as an entertainment page, repeatedly shared false claims which appeared to be pro-Obi. Seven other groups, all with the name ‘Peter Obi’ their title, were found to have either reshared or replicated the claims from Celebrity vibes.   

At the same time, the majority of the claims (42) were traced to other platforms. The finding indicates a deliberate effort by some claimants to propagate some claims to other platforms. Burkhardt (2017) asserts that spreading false information on social media is possible because “the manipulation of computer code for social media sites allows fake news to proliferate and affects what people believe, often without ever having been read beyond the headline.”

Moreover, Twitter was found to be the most used platform in propagating claims from other platforms, precisely claims from Facebook. A good number of claims that originated from other platforms to were replicated on Twitter.  The study also revealed that accounts identified as ‘groups or pages’ and by ordinary citizens categorised as ‘individuals’ shared many of the claims. These two categories of claimants account for 28 of the 50 claims. 

Most of the claims were shared on Facebook; it  featured 28 claims (56%), all of which were false. Savino (2017) asserts that false information is likely to spread more on Facebook because ‘fake news articles’ look almost identical to those from reputable media organisations. This finding is perhaps also due to the prevalence of Facebook usage amongst Nigerians. In contrast, Twitter came next as a distant second. The study concludes that while most of the claims about the presidential candidates were found on Facebook, claims related to Peter Obi or his party were more often found on Twitter than any of the other aspirants. This reflects the prominence the Labour party candidate enjoys on the Twitter platform. 

The sentiment analysis in the study showed that a large number of claims were negative, and a majority of claims also false. This implies that the claimants put out most false claims with a negative intention. A false claim leads someone to reason that other negative information about the candidate is accurate, which can alter choices (Thorson, 2016). Additional findings showed that Peter Obi and his party had the highest number of claims that were positively toned and a good number of claims that were negatively toned. At the same time, Atiku and Tinubu had a more significant percentage of claims being negatively toned.

7.0 Conclusion

The above findings clearly show a massive prospect of information disorder ahead of the 2023 general elections in Nigeria. Another significant finding is identifying some key characters who have shared false information more than once. These characters may have a wider network of online activities than what appears to be. The study anticipates even more dispensation of information disorder ahead of the 2023 general elections. 


Apuke, O. D., & Omar, B. (2020). Fake news proliferation in Nigeria: Consequences, motivations, and prevention through awareness strategies. Humanities and Social Sciences Reviews, 8(2), 318-327. 

Balli, S. N., et al. “Spread of Fake news on Twitter: A Study.”

Baptista, J. P. & A. Gradim (2022). “Online disinformation on Facebook: the spread of fake news during the Portuguese 2019 election.” Journal of contemporary European studies 30(2): 297-312.

Bartlett, J., et al. (2015). “Social media for election communication and monitoring in Nigeria.” A report prepared for the Department for International Development (DFID), March.

Bevendorff, J., et al. (2020). Overview of PAN 2020: Authorship verification, celebrity profiling, profiling fake news spreaders on Twitter, and style change detection. Paper presented at the International Conference of the Cross-Language Evaluation Forum for European Languages.

Bovet, A., & H. A. Makse (2019). “Influence of fake news in Twitter during the 2016 US presidential election.” Nature communications 10(1): 1-14.

Burkhardt, J. M. (2017). “How fake news spreads.” Library technology reports 53(8): 10-13.

Dittrich, P.-J. (2019). Tackling the spread of disinformation Why a co-regulatory approach is the right way forward for the EU. Bertelsmann Stiftung Policy Paper 12 December 2019. 

Gibson, R. (2015). “Party change, social media and the rise of ‘citizen-initiated’ campaigning.” Party politics 21(2): 183-197.

Horowitz, M. A. (2019). Disinformation as warfare in the digital age: dimensions, dilemmas, and solutions. Journal of Vincentian Social Action, 4(2), 5. 

Igwebuike, E. E., & L. Chimuanya (2020). “Legitimating falsehood in social media: A discourse analysis of political fake news.” Discourse & Communication 15(1): 42-58.

Jensen, M. (2018). “Russian trolls and fake news: Information or identity logics?” Journal of International Affairs 71(1.5): 115-124.

Mare, A., Mabweazara, H. M., & Moyo, D. (2019). “Fake news” and cyber-propaganda in Sub-Saharan Africa: Recentering the research agenda (Vol. 40, pp. 1-12): Taylor & Francis.

Mutsvairo, B., & Rønning, H. (2020). The Janus face of social media and democracy? Reflections on Africa. Media, Culture & Society, 42(3), 317-328. 

Narwal, B. (2018). Fake news in digital media. 2018 international conference on advances in computing, communication control and networking (icacccn), IEEE.

Sastrosubroto, A. S., & S. Pratama (2019). “Fake news and information in Social Media: Problems and challenges, the case in Indonesia.”

Savino, E. M. (2017). “Fake News: No One Is Liable, and That Is a Problem.” Buff. L. Rev. 65: 1101.

Singh, B., & D. K. Sharma (2021). “Predicting image credibility in fake news over social media using multi-modal approach.” Neural Computing and Applications: pp. 1-15.

Thorson, E. (2016). “Belief echoes: The persistent effects of corrected misinformation.” Political Communication 33(3): 460–480.

Vosoughi, S. et al. (2018). “The spread of true and false news online.” science 359(6380): 1146–1151.

Wardle, C. & H. Derakhshan (2017). Information disorder: Toward an interdisciplinary framework for research and policymaking, Council of Europe Strasbourg.

Wasserman, H., & Madrid-Morales, D. (2019). An exploratory study of “fake news” and media trust in Kenya, Nigeria and South Africa. African Journalism Studies, 40(1), 107-123. 

Show More

Related Articles

Leave a Reply

Back to top button