Inclusive Crisis Management by Governments: Using Digital Ethnography and Sentiment Analysis as a Sensing Function and Policy Tool

This report demonstrates how real-time social analytics can provide a tool for inclusive policymaking and crisis management. Using a dataset of over 873 million online interactions drawn from more than one hundred social and mainstream media channels, Equiception used bespoke AI tools to access, analyze and track trends in public sentiment and emotion in response to the pandemic management of eight governments between January and July 2020.

Our methodology provides policymakers with a gauge that offers a more representative and inclusive insight into public sentiment, allowing them to monitor levels of, and shifts in, trust and confidence across diverse communities, including those that may be under-represented in traditional voice channels.

The use of social media, sentiment and opinion mining in government is not new, but the application of such digitally enabled analytical techniques to gauge the impact of policy interventions and calibrating policy is novel. Our methodology enables policy makers to gain inputs and insights at unprecedented scale and speed that can be used at every stage of the policy making and execution cycle, from agenda setting to formulation, implementation and evaluation, and, most importantly, in engineering real-time interventions and course corrections.

The innovation in our approach is the use of digital ethnography to listen to all the voices and conversations around a pertinent theme, detect the sentiment and emotion, contextualize them, and discern the narratives. This approach of dynamic monitoring of the public discourse around COVID-19 can be critical in managing top-down actions by governments especially when they involve restrictions on individual liberties and freedoms during a crisis.

Our analyses of data from eight countries demonstrate how government leaders, policy makers, and administrators failed to appreciate the socially patterned impacts of the virus as well as the control measures taken to contain it. Even nations that were widely praised as being among the best responders to the first wave of the pandemic, such as Singapore and South Korea, overlooked the impacts on major high-risk groups (migrants and gig economy workers). Using a methodology similar to ours could have helped them identify the at-risk groups and their lived experiences and adjust the planning or implementation processes to include their voices and needs.

One notable, common feature in the data is the significant upswing in joy or satisfaction when the government announced the initial lockdown or equivalent. What is even more significant is that this phenomenon occurred in every country studied, i.e., the sentiments were uniform across different societies. There can be several explanations for this seemingly paradoxical behavior, a universal spike in expressions of joy during times of extreme uncertainty and potential anxiety about the future. A possible explanation is that there was a collective “rally round the flag” effect that has been documented in times of crises such as wars or terrorist attacks, when the perception of a threat leads citizens to seek the protection of, and certainty from their governments.

While the spike in expressions of joy were more pronounced than the other emotions in most countries, in the USA sadness spiked to almost the same level and was reinforced by smaller spikes of disgust and anger. The combined effect of such “trust-negative” emotions outweighed the “trust-positive” impact of the spike in joy.

The rally round the flag effect was relatively short lived in the USA. Had President Trump emphasized the magnitude of the COVID-19 threat, as the leaders of New Zealand and South Korea did, he could have generated a greater rally effect, possibly even across party lines. Instead, he chose to downplay and dispute the threat, thus robbing himself of the full potential of the rally effect and sparking controversies that provoked competing emotional responses.

One of the most common and effective strategies—adopted by President Donald Trump and Prime Ministers Boris Johnson of the UK, Narendra Modi of India, and Anders Tegnell of Sweden —was to gain and maintain in-group trust while fanning distrust among other groups. This worked despite their failures to adequately mitigate the impact of COVID-19. This decoupling of trust from performance and accountability is a troubling and increasingly widespread phenomenon that could undermine the global effort to contain COVID-19 and inclusive governance in times when it is needed the most. New Zealand, Sweden, and the USA saw greater contestation in the emotional reactions to government intervention, with multiple emotion trend lines rising and falling together, indicating high levels of extant disagreement and polarization within those countries.

There were only two countries in the study where sadness was the dominant emotion beyond the initial spike in March 2020: New Zealand and the USA. Closer examination of the emotional responses and their context showed that the sadness displayed by New Zealanders was trust-enhancing that in the USA was trust-eroding. This was a reflection of the fact that Prime Minister Ardern believed the public could be trusted with the truth whereas President Trump concealed the truth and offered no sympathy for the exponential increases in infection and fatality.¹

The data collected in this study reveal a second notable feature across almost all the countries surveyed – systemic shortcomings in the representation and recognition of diverse voices, what we call the “voice deficit” – a marker of the absence of inclusive policymaking that resulted in blind spots, exclusion, and breakdowns in pandemic planning, management, and protection.

This voice deficit led to blind spots in the management of the unfolding crisis. These were evident in government responses in both hemispheres and at all levels of socio-economic development. Examples of groups affected by these blind spots include evangelical churches in South Africa, South Korea and the USA, contingent and gig economy workers in South Korea, the UK and the USA, residents of elderly care homes in Sweden, the UK and the USA, the informal sector and migrants in India and South Africa, and Black and Latino communities in the USA. The blind spots inevitably resulted in the exclusion of these groups from pandemic planning and protections, leading to their increased risk of exposure to the virus itself and its economic impact.

We think this voice deficit could have been avoided had our methodology been used to tap into the lived experiences of the aforementioned social groups and the probable or actual impacts of pandemic control measures on them. Social media data could also have served as a feedback loop and a public sentiment monitoring mechanism to assess and evaluate outcomes.