Political expression of academics on social media

Social media platforms allow for immediate and widespread dissemination of scientific discourse. However, social media may distort public perceptions of academia through two channels: the set of topics being discussed and the style of communication. This column uses a global dataset of 100,000 scholars to study the content, tone, and focus of their social media communications. It finds systematic differences in the views expressed by academics and the general public, in both the topics and the tone of discussion. There are also clear differences between academics, depending on their gender, field, country of affiliation, and university ranking.

Social media platforms may be important marketplaces for the exchange and dissemination of ideas, where academics and researchers hold significant roles as knowledge producers and influencers (Cagé et al., 2022, Gentzkow and Shapiro 2011). Policymakers may actively scout these platforms for policy insights or ideas, while journalists increasingly rely on digital sources like Twitter to shape news agendas (Muck Rack, 2019). The backlash against the journal Nature‘s endorsement ¹ of Joe Biden in the 2020 US presidential election illustrates the risks of political expression in scientific discourse, highlighting its potential to polarise public trust (Zhang 2023). This example raises questions about whether academics should maintain political identities when expressing views, to safeguard public trust in their scholarly independence and expertise.

The COVID-19 pandemic has highlighted the importance of trust to ensure compliance with public health measures that are rooted in hard science (Algan et al., 2021). Concerns also arise over the disconnect between academics and the public on issues like Brexit or populism more broadly, influenced by biases associated with political affiliations (Den Haan et al. 2017, Mede and Schäfer 2020) and differing perspectives on economic policies (Fabo et al. 2020).

The upsides of academic engagement on social media, by providing new and more direct ways of science communication, may inadvertently shape public perceptions of academia through selective topic engagement and differences in communication styles. Unlike traditional media, social media allows for immediate and widespread dissemination of scientific discourse. Most academics are not trained in such communication. Further, not all academics engage with social media, meaning that those who do may disproportionately influence public perceptions of academia through both the specific topics they choose to discuss and the styles and tones in which they communicate. Our new paper explores patterns in academics’ political expressions using a global dataset linking the Twitter profiles of 100,000 scholars to their academic records, spanning institutions across 174 countries and 19 disciplines from 2016 to 2022 (Garg and Fetzer 2024). Leveraging scalable large-language model (LLM) classification techniques, we analyse the content, tone, and substantive focus of their communications.

We document large and systematic variation in politically salient academic expression concerning climate action, cultural, and economic concepts. Views expressed by academics often diverge from the general public opinion in both the topics they focus on and the styles in which they are communicated.

Key findings

Finding 1: Academics are much more liberal and less toxic on Twitter than the general population

Academics on social media are markedly more vocal on politicised issues compared to the general population. Specifically, academics are 10.7 times more likely to express opinions in favour of climate action, 6.2 times more expressive about virtues of cultural liberalism, and 23.3 times more vocal about advancing economic collectivism than the average US social media user. However, academics consistently exhibit lower levels of toxicity and emotionality in their discourse compared to broader US Twitter users, with toxicity rates halving to around 4%, while the general population maintains rates between 8-9%. This divergence in political expression and communication style may contribute to public misconceptions about academic consensus, potentially affecting public trust in academia and influencing policy debates.

Figure 1 Differences in academic expression between academics and general population in the US

Note: This figure explores the divergences in expression between academics and the general US Twitter population, leveraging two distinct datasets: one comprising tweets from 100,000 US-based academics and another from a sample of 60,000 users representative of the general US Twitter population. The analysis highlights notable differences in both behavioural expression and political stances from January 2016 to December 2022.

Finding 2: American professors tend to be more egocentric and toxic on social media, but professors from other countries tend to be nicer than average

We find big differences in the tone and style of expression between academics, characterised by their field, institutional ranking, gender, and country of affiliation. Academics based in the US and those from top-ranked institutions tend to exhibit higher levels of egocentrism and toxicity on social media. In contrast, academics from other countries generally demonstrate lower levels of toxicity in their online discourse. We also find that humanities scholars and academics with extensive Twitter reach but lower academic prestige display heightened egocentrism. Conversely, academics with lower Twitter reach, regardless of academic standing, and those associated with top 100 universities, exhibit elevated toxicity levels compared to their counterparts.

Figure 2 How academics express themselves online: Average by author characteristics

Note: This figure presents the average levels of three behavioural metrics—Egocentrism, Toxicity, and Emotionality/Reasoning—quantified from tweets by a balanced panel of academics from 2016 to 2022, using data linking 100,000 Twitter profiles to their academic profiles. Each panel categorises data by specific groups such as gender, fields, Twitter reach, and academic credibility, university rankings, and country. 95% confidence intervals are indicated by error bars.

Finding 3: Climate action: Discipline polarisation and expert selection

On average, academics are 10.7 times more vocal in advocating for climate action. Male academics exhibit a stronger preference for technological solutions over behavioural adjustments to tackle climate change. STEM scholars prioritise technological solutions, reflecting their focus on innovation and engineering approaches, while social sciences and humanities scholars often emphasise behavioural adjustments and societal transformations as primary solutions. This diversity in academic emphasis allows policymakers to select experts whose perspectives align with their policy goals. Moreover, academics affiliated with top-ranked US universities, and those with larger social media followings but lacking expertise in climate issues, show relatively lower support for proactive climate measures.

Figure 3 How academics talk about climate action online: Average by author characteristics

Note: This figure illustrates academics’ average stances on three pivotal climate change policy topics—Climate Action, Techno-Optimism, and Behavioural Adjustment—across different academic groups and personal characteristics, using a sample of 138 million tweets made by 100,000 academics between 2016 and 2022. 95% confidence intervals are indicated by error bars.

Implications

The over-representation of certain views – especially from high-reach, low-expertise academics – and the under-representation of other views could result in distorted or poor communication of science. Given these complexities, it is crucial for the academic community to engage in a more inclusive and balanced manner, ensuring that the marketplace of ideas on social media enriches rather than distorts public discourse and policy formulation. Further research should aim to quantify the impact of these ideological divisions on public trust and explore strategies for mitigating potential biases in academic communication on social media. Additional research is required to explore why academics express themselves politically, considering motivations such as name recognition, ideological drive, and the desire to share or evangelise knowledge.

Source : VOXeu