An Incumbency Bias: Mainstream Media’s Election Coverage on Twitter

Photo Courtesy of Daily News 

Based on my previous study of looking at how Sri Lankan partisans on Twitter perceived their favoured presidential candidate, and opponents, I wanted to establish how mainstream media (MSM) accounts on the platform presented (and, in a thinly veiled manner, promoted) candidates as well.

Undergirding what’s now called information disorders, Sri Lankan mainstream media is, and has always been, deeply partisan. What’s evident now on social media can be understood through a study of media ownership. As the seminal Media Ownership Monitor website by Verite Research notes,

How can people evaluate the reliability of information if they don’t know who provides it? How can journalists work properly if they don’t know who controls the company they work for? And how can media authorities address excessive media concentration if they don’t know who is behind the media’s steering wheel?

Even with the significant enshittification after Musk’s acquisition, Twitter continues to play an important role in the Sri Lanka’s information ecologies, and political communications landscape. MSM’s Facebook accounts feature a considerable amount of propaganda, adding to, and further complicating the viral spread of propaganda over the platform through pink slime, gossip, meme, and proxy accounts for each candidate, their political party or coalition during electoral moments. But Twitter often serves as an entry point to political controversies, developments, and debates. Tweets featured on Sri Lanka’s MSM accounts often amplify content published on respective websites, and other social media platforms like Facebook, YouTube, or TikTok. Unlike in New Zealand, where I also study information, and media ecologies very closely, Musk’s acquisition of the platform has not resulted in any discernible decline in usage or mass migration to alternatives like Threads, Mastodon and Bluesky. Even though it is wrong to generalise Sri Lanka’s political pulse from Twitter alone, the platform remains integral to the distribution of news, information, political campaigns, and the delivery of propaganda.

This is significant.

Why does it matter?

Very few read what they go on to share on social media. This is obviously a problem, and helps illuminate why what’s noted in tweets, and how (including through embedded media) is so important. What’s in each tweet crafts or influences, at scale, public perceptions, attitudes, responses, and behaviours – including, ultimately, the exercise of franchise during electoral moments.

Twitter, before Musk, tried to address this. In 2020, the platform introduced an interstitial that asked users if they had read the article they were sharing – often reflexively. The test, which started on Android, then extended to iOS, noted that “Sharing an article can spark conversation, so you may want to read it before you Tweet it…”. As an article published in the Guardian by Alex Hern around the same time noted,

The problem of users sharing links without reading them is not new. A 2016 study from computer scientists at Columbia University and Microsoft found that 59% of links posted on Twitter are never clicked. Less academically sound, but more telling, was another article posted that same year with the headline “Study: 70% of Facebook users only read the headline of science stories before commenting” – the fake news website the Science Post has racked up a healthy 127,000 shares for the article which is almost entirely lorem ipsum filler text.

These patterns, and pathologies are present in Sri Lanka too. In what’s a high media consumption but low media literacy context, content on social media without any critical framing or fact-checking essentially serves to establish partisan presentations as news coverage. This dynamic arises from, and feeds into a very low quality political news reporting culture – where holding politicians to account is enduringly hostage to, and governed by media ownership structures, the partisan leanings of Editors, advertising revenue from state enterprises as well as oligarchs connected to those in power, and chilling effects because of pushback against critical coverage – or in other words, the actual practice of journalism.

Political narratives parading as news updates on Twitter frequently embellish, conflict with or fall short of conveying complete, critical captures of politics, politicians, and policies. Engaging in discourse, and shaping one’s views based on what’s tweeted by MSM accounts risks adding to what academics call ‘affective polarisation’ – in which one’s own bias is confirmed, and those with a different perspective, violently rejected. Emotive, partial, partisan narratives in tweets published by MSM possess the capacity to mould perceptions at scale, and influence behaviours in the absence of a more critical, and comprehensive context.

This content also contributes to an incumbency bias through which Ranil Wickremesinghe benefits from more frequent, uncritical presentations than his two chief rivals. We can establish this in the data. As a measure of bias in the presentation of candidates (inadvertantly or by design), I studied the amount of words in the tweets dedicated to the incumbent President (RW), the Leader of the Opposition Sajith Premadasa (SP), and the NPP’s Anura Kumara Dissanayake (AKD). This analysis counted direct mentions, and closely associated content, but doesn’t account for indirect references or broader contextual discussions.

SP, and AKD – along with their respective political parties, and coalitions – are consitently reported on, and represented less than the incumbent President, and narratives that feature him. The data reveals a media landscape on Twitter that heavily favours Wickremesinghe through uncritical presentations of what he did, said or promised – influencing public perceptions in the lead-up to the election.

This bias is further compounded by paid campaigns on some of these Twitter accounts. For example, the Daily Mirror Twitter account is repeatedly running a plethora of campaign ads clearly partial to what’s a bias present in its presentation of candidates. The two may be connected. It is unclear how much is being spent on these ads (given that each candidate can spend as much as 2 billion rupees), but featuring paid propaganda obviously presents significant challenges around impartial, accurate, and fair journalism coverage of candidates since critical perspectives, including those which put into context what Wickremesinghe promises or says – will place at risk a revenue stream that’s likely to be in the millions of rupees, if not much more.

Though outside the scope of this study, the coverage of other presidential candidates – including, this time around, Namal Rajapaksa from the SLPP – is even poorer. This speaks to an enduring, and intentionally constructed information asymmetry, in which the propaganda connected to those in power, including through MSM, is much greater than rivals during electoral moments. This is a familiar problem, and pathology that significantly adds to what I’ve penned earlier as factors that undermine electoral integrity in Sri Lanka.

Data collection, and analysis

All the tweets for this study were captured on Friday the 13^th (what better day to study Sri Lanka’s presidential campaign horrors!). Tweets from 1 August to 13 September were studied – covering the period when the date for the presidential election was announced, nominations were handed in, and the campaigns officially started.

Accurate determinations of reach are impossible on Twitter, especially with Musk’s revision of algorithms to prioritise his own ego, but the subscription figures for the accounts studied – featured below – provide a proxy indicator of how many these accounts reach. Allowing for obvious overlap between the accounts, the biggest difference in demographics would be with Hiru, and Ada Derana Sinhala, which both tweet near exclusively in Sinhala, in comparison to the other accounts which tweet exclusively in English.

Given the sheer size of the Sinhala text corpus linked to these two accounts (nearly 41,000 words), I used Anthropic’s Claude 3.5 Sonnet to translate all the tweets to English prior to analysis. The same model from Anthropic was used for preliminary text analysis, in addition to OpenAI’s just released OpenAI o1 preview model.

Nearly all the tweets studied featured photos of the candidates, various image collages, still frames from videos or some other form of visual or memetic presentation. These were not studied, but are important to consider as markers of meaning, adding to, and strengthening what’s noted in this article. It goes without saying that all the tweets featured links to news stories on the respective websites or other social media accounts (e.g., YouTube). These were also not studied.

The accounts studied generated nearly 12,500 replies, and nearly 12,000 retweets, which weren’t studied. Each retweet carries the potential of even more replies, and quote tweets in what’s a sea of associated commentary, and content that Twitter as a platform is just not open to any more for the deep study of. These numbers, however, are clear proxy indicators of a conversation on Twitter that’s carried out on the basis of what’s presented over the accounts studied – whether to reflexively share, and agree with or to vehemently decry, and deny the validity of through hyper-partisan frames. In other words, what MSM accounts tweet, and how, matters.

Reach of accounts studied

Engagement with tweets studied

The construction of political personalities

In my earlier study of propaganda on Twitter I noted,

What’s “real” and “authentic” in politics is often manufactured. Today’s election campaigns are highly sophisticated and feature dark arts that instrumentalise our fears, anxieties and aspirations. To know this is to also realise that all election campaigns manipulate their audiences. This is, after all, propaganda’s raison d’être. The RW, SP and AKD one consumes in the media are fictional constructs intended to serve partisan ends.

This article, which looks at a much larger dataset, confirms this observation. MSM’s presentations of three leading candidates show some similarities, and key differences – which hint at the partisan bias of respective media networks, and platforms. What’s particularly insidious about these constructed personalities is they are based on tweets which appear to be journalistic coverage, and not official propaganda – thus appealing to more diverse constituencies. But when studied at scale, and over a longer period, each account’s stress as well as silences reveal partisan bias.

1. Wickremesinghe is, across the half-dozen accounts studied – portrayed as an active and strategic leader deeply engaged in the upcoming presidential election, with a strong focus on economic reforms and policy proposals. All accounts highlight his campaigning activities, such as signing nomination papers, releasing manifestos, and building political alliances with various parties and figures. They also note his efforts to address economic issues, emphasising his commitment to economic recovery and reform, and his warnings about potential economic consequences if current programmes are altered. However, differences emerge in the emphasis and tone of these portrayals. Some accounts, like Newswire and News1st, underscore his substantial support base and active policy engagement but also acknowledge significant opposition and critique from rivals. Colombo Gazette and Daily Mirror introduce elements of controversy, mentioning legal challenges, criticisms over his language choices, and concerns about selling state assets or campaign conduct. Ada Derana and Ada Derana Sinhala focus on his vigorous campaigning and policy promises, reflecting a leader defending his policies amidst mixed support, while also showcasing his strategic vision and human touch through economic reforms and religious dialogue. Hiru emphasises his participation in religious and official events and the significant support he receives from numerous political parties, yet also highlights criticism from opposition figures questioning his policies and leadership.
2. Premadasa is presented as an active and prominent opposition leader, vigorously campaigning for the upcoming presidential election with a strong focus on social welfare, economic development, and anti-corruption measures. All accounts highlight his efforts to build broad political alliances and garner support from various politicians, parties, and notable personalities, including former cricketers and minority group leaders, reflecting his strategy to unite diverse groups under his leadership. His ambitious policy proposals, such as eradicating poverty within a specific timeframe, providing direct financial support to low-income families, reforming the education sector, and empowering farmers and fishermen, are common themes across the accounts, emphasising his commitment to comprehensive national development. However, differences emerge in the emphasis on internal party dynamics and challenges; while some accounts like Newswire and News1st mention internal conflicts within the SJB, including the expulsion of members and resignations of key figures, suggesting potential obstacles to his leadership, others like Colombo Gazette and Ada Derana Sinhala focus more on his growing support base and strategic alliances, portraying him as a unifying figure despite these challenges. Criticisms are also noted in certain accounts, with references to allegations of impractical promises, accusations of copying other parties’ manifestos, and questions about his commitment to reducing executive power.
3. Dissanayake is framed as an active and transformative presidential candidate leading the National People’s Power (NPP), with ambitious plans for economic reform and national rebuilding. All accounts highlight his vigorous campaigning and policy-focused approach, emphasising proposals aimed at economic relief, reducing the cost of living, and implementing governance reforms. Common themes include his pledges to reduce fuel and electricity prices, remove taxes on essential items, and establish a clean political culture free from corruption. He is uniformly depicted as a strong opposition voice, criticising the current government and other candidates, and positioning the NPP as a significant alternative to traditional politics. However, differences emerge in the emphasis and tone of these portrayals. Some accounts, such as Newswire and News1st, focus on specific policy initiatives like offering subsidies to low-income families, updating price formulas, and introducing educational reforms, portraying him as a serious contender with a comprehensive policy platform. Colombo Gazette and Ada Derana highlight his manifesto and commitments to strengthening law and order and engaging in international relations, reflecting a leader with both domestic and international focus. Ada Derana Sinhala emphasises his confidence in electoral success and his stance against political privileges, showcasing a leader determined to enact significant systemic changes. In contrast, Hiru presents a more critical tone, with less frequent mentions and references to historical critiques of the JVP, indicating a varied reception. Additionally, while some accounts note criticisms over the substance of his proposals and accusations of opportunism, others focus on his growing support and strategic campaigning.
4. The portrayal of internal party dynamics is another significant aspect that varies across the accounts. While Premadasa’s SJB is shown dealing with internal challenges, such as the resignation of Field Marshal Sarath Fonseka as SJB Chairman and the expulsion of certain members, there is also mention of efforts to replace MPs who have left, indicating proactive steps to strengthen the party. Similarly, Dissanayake’s NPP is depicted managing internal disputes, with references to pursuing a defamation suit against an SLPP MP and refuting allegations regarding the ‘People’s Struggle’.
5. Premadasa’s ambitious pledges, such as eliminating poverty within two years and providing direct financial support to low-income families, are sometimes met with scepticism regarding their practicality in the tweets studied. Dissanayake’s comprehensive policy platform, including significant economic reforms and governance changes, is occasionally scrutinised for feasibility. In contrast, Wickremesinghe’s policy initiatives are presented in the tweets studied without such scepticism, focusing on his commitments to economic recovery and warnings about potential consequences if current programmes are altered.
6. Tweets framing Dissanayake’s emphasis on abolishing political privileges, such as pensions and special allowances for retired presidents, and adjusting public servants’ salaries to match the cost of living, signals a focus on systemic change. Premadasa’s promise to abolish the Executive Presidency, and modernise the education sector presents a commitment to significant governance reforms. These aspects contrast with President Wickremesinghe’s portrayal, which is more centred on maintaining current economic programmes and international cooperation, suggesting a more conservative approach to structural changes.

The bias in presentations: RW vs SP, and AKD

Though there’s a common bias towards the promotion of the incumbent President, the RW, SP, and AKD presented in the six MSM Twitter accounts studied differ. Generally, across the half-dozen accounts nearly 6,000 tweets studied, Wickremesinghe is depicted as an active, and strategic leader focused on economic reform, political manoeuvring, and international cooperation, with less emphasis on controversies or criticisms. In contrast, SP, and AKD are frequently associated with internal party challenges, criticisms of their proposals or scepticism about their methods, which portray them in a less favourable light.

1. Daily Mirror: Wickremesinghe’s portrayal tends to be more favourable, emphasising his experience and economic management skills, while downplaying criticisms of his political manoeuvres. Sajith Premadasa’s coverage is more mixed, highlighting his popular appeal but also questioning the feasibility of his promises. Anura Kumara Dissanayake’s presentation, while generally positive in terms of his reform agenda, seems to carry an underlying scepticism about his party’s ability to govern effectively. The bias is evident in the framing of issues, with Wickremesinghe’s actions often contextualised within broader economic recovery efforts, Premadasa’s portrayed more in terms of political opposition, and Dissanayake’s frequently presented as idealistic but potentially impractical.
2. Ada Derana Sinhala: Wickremesinghe receives the most extensive and diverse coverage, focusing on his current actions and policies, which presents him as an active leader dealing with immediate issues. This contrasts with Sajith Premadasa’s portrayal, which primarily highlights his future promises and campaign rhetoric, potentially undermining his current relevance. Tweets on AKD present him as a radical change agent with bold claims, but with less emphasis on concrete policy details, which could be interpreted as lacking substance. The disproportionate coverage and the framing suggest a bias that favours Wickremesinghe’s established position, while potentially diminishing the credibility or immediate impact of his two chief rivals.
3. Ada Derana: RW receives the most coverage in Ada Derana’s tweets, often portrayed in the context of his incumbent role, with a mix of official duties and campaign activities. This gives RW a more authoritative presence or aura in the tweets. SP is primarily presented as the main opposition figure, with a focus on his campaign promises, and criticisms of the current administration. AKD, with the least coverage among the three, is often portrayed as an alternative to the traditional political establishment, with an emphasis on systemic/structural change. This disparity in coverage, and narrative framing suggests a bias towards maintaining the current political narrative, with RW at the centre, SP as the main challenger, and AKD as a less prominent, albeit radically transformative, option.
4. Hiru: RW receives the most coverage, often in the context of official duties, but with a mix of performatively neutral reporting, and criticism. SP is presented more favourably in the tweets, with emphasis on his campaign activities, policy promises, and public support. AKD receives significantly less attention, mostly in the context of policy critiques and campaign activities, but with less detail compared to the other two. This disparity in coverage, subtle framing, and tone suggests a bias towards established political figures, particularly Premadasa as the main opposition leader, while potentially marginalising AKD’s campaign in the overall narrative.
5. Newswire: RW is portrayed more in terms of his official actions and decisions, often in a neutral or slightly favourable light, emphasising his role in economic recovery, and governance. SP is presented as the main opposition figure, with coverage focusing on his campaign promises and critiques of the government. AKD/NPP, while receiving less overall coverage, are often framed as outsiders offering an alternative to the established political system, with their policy proposals, and criticisms given particular emphasis. This framing subtly positions/presents Wickremesinghe as the established (or stable) leader, Premadasa as the conventional opposition, and Dissanayake as the radical change candidate – potentially influencing readers’ perceptions of each figure’s role, and potential.
6. Colombo Gazette: Mindful that the number of tweets by the Colombo Gazette is smaller in comparison to the other accounts, Wickremesinghe receives significantly more coverage, and is presented in a broader context, encompassing both his governance activities, and campaign efforts. This extensive coverage, while not always positive, makes him the dominant character in the electoral coverage. SP primarily portrayed in the context of his campaign activities, and growing political alliances. AKD receives notably less attention, with his limited mentions focusing mainly on specific policy proposals, and challenges to the status quo.
7. News1st: RW is often presented in the context of his role as incumbent President, lending his mentions a tone of authority. SP receives slightly less coverage, with a mix of positive campaign reporting, and mentions of party conflicts. AKD receives the least coverage, but is generally presented more positively as an alternative to the established political order, with more detailed policy proposals mentioned when he is discussed.

The full report is available here : CLICK