Search
Close this search box.
Search
Close this search box.

Justice in the COVID-19 Crisis:
What People Are Saying
on Social Media

Trends beyond words | Words behind trends

Photo by Gordon Johnson from Pixabay

Justice problems on
social media

What people are saying on Facebook and Twitter during the pandemic

Throughout the COVID-19 crisis, people have been sharing their views on the justice problems happening around them on social media. In April 2020, HiiL began collecting data from Facebook and Twitter to identify trends in social media activity and find out what people around the world have been saying. Based on an analysis of hundreds of thousands of social media messages, this report aims to answer the following questions:

Most of all, this experimental project aims to explore the potential and limits of a new source of data and knowledge. Below, we explore what social media can tell us about justice.

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Our methodology

How we listened for justice problems on social media

To collect rapid, almost real-time people-centred data on justice during the COVID-19 crisis, we asked HiiL national experts in the nine countries where we have presence to identify hashtags, keywords, and phrases which people use to discuss justice problems on social media. This led to a list of country-specific keywords. The keywords were aggregated and coded into eleven justice problems categories: crime; domestic violence; medical bill disputes; financial disputes; business problems, housing disputes; disputes between neighbours; problems with accessing government benefits; and land disputes.

With the help of an external knowledge partner – GemSeek – we used the locally-validated keywords to collect data from Twitter and Facebook in South Africa, Nigeria, Lebanon, Burkina Faso, Uganda, Mali, Kenya, Ethiopia, and Tunisia. Messages were collected retrospectively from 1 January 2020 to create pre-COVID-19 data. We also gathered messages from Twitter and Facebook that contained the keywords but were not linked to the nine focus countries. This dataset is referred to as the “global dataset”.

A natural language processing algorithm was developed, trained, and applied to classify each social media message into categories. The algorithm was set to distinguish between general COVID-19 messages and messages related to justice problems. For example, somebody might say, “I lost a good job.” This is clearly a justice problem. Another person might say, “Good job with your efforts on the final exam.” This is not a justice problem.

The algorithm retained the relevant messages and removed others, which were marked as noise. The relevant messages were then classified into the aforementioned eleven justice problem categories.

After all of these steps, the dataset for the nine focus countries consisted of 587,286 messages (from January-July 2020). Of these messages, 73 percent are specific to a justice category and 27 percent are COVID-19-related. Nigeria and South Africa account for the majority of the messages, with 36 percent and 33 percent of all messages, respectively. The next countries in terms of frequency are Kenya, Uganda, and Lebanon with 14 percent, 10 percent, and 4 percent, respectively. The other countries have relatively low message counts.

A machine learning algorithm estimated the sentiment of each social media message in the country dataset. Sentiment analysis is a classification method that aims to identify the degree of feeling or emotion embedded into a text. In practice, the algorithm checks the texts for occurrences of positive, negative, and neutral words and phrases. Based on the results of this analysis, a sentiment score is generated and attached to each social media message.

The global dataset consists of over 79 million messages.

Analysing these messages produced the following visualisations, which provide global, country-level, and comparative insights into social media activity around justice problems over the past six months.

Instead of explaining the trends, we invite the readers to interact with the data, formulate their own questions, and seek answers. Along this journey, we will share some observations and interpretations.

Read more about our methodology

The global view

Crime and domestic violence concerns rise steadily. Talk of employment, financial disputes, and issues with medical bills grew at the peak of the pandemic but has since decreased.

Crime and domestic violence

Globally, talk of crime and domestic violence on social media has increased steadily since the start of the global pandemic in March. This trend is consistent with HiiL’s COVID-19 expert survey findings, which anticipated spikes in crime and domestic violence in low- and middle-income countries. In July, mentions of these problem types declined slightly. This may be due to summer re-openings that made it easier for people to access social services and receive assistance for their crime and domestic violence-related needs.

Financial, medical and
employment disputes

Mentions of medical bill disputes, financial disputes, and employment disputes also began to rise in February but peaked between March and April. Mentions of employment disputes increased noticeably between June and July. Mentions of other problem types have remained relatively stable during the COVID-19 crisis.

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Country deep dives

The peak of mentions of justice problems followed by 2 to 3 months the peak of the COVID-19 messages.

In the charts that follow, we zoom in on social media trends in nine countries where HiiL works and observe how justice problems have been talked about in those specific places.

Justice problem mentions peak in May, June and July

An interesting trend appears when we compare messages about the COVID-19 crisis and messages about justice problems. The peak of COVID-19-related messages is in March, when the pandemic became global. A lag appears when we look at messages about justice problem – they peak in May, June, and July. This trend is particularly visible in countries with large volumes of messages – South Africa, Nigeria, and Kenya.

One interpretation of this lag is that the pandemic was the root cause of many justice problems around the world. It does not take long for people to feel the justice implications of COVID-19. The big question is: how will the trends in justice problem mentions change over time? Will they subside, maintain the current levels, or increase?

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Mentions of justice problems increased in South Africa, Nigeria and Uganda

Mentions of justice problems increased dramatically in South Africa and Nigeria between the months of April and June. Justice problems were also mentioned more frequently than usual in Kenya during the same period. Between June and July, talk of justice problems in these three countries has started to decline.

In Uganda, mentions of justice problems on social media increased slightly after March but have since returned to pre-COVID-19 levels. In other countries, justice problems were not discussed more than usual on social media during the COVID-19 crisis.

Crime is the most commonly mentioned justice problem on social media in Lebanon, Mali, Tunisia, Burkina Faso, and Ethiopia. In Uganda, Kenya, and Nigeria, housing disputes are most often talked about. In South Africa, social media mentions of justice problems mostly relate to employment disputes.

In six out of the nine countries, crime and domestic violence are among the top most frequently talked about justice problems. This is consistent with the global dataset, which shows a post-COVID-19 spike in these two problem types. Financial and employment disputes, which were also talked about more after the onset of the pandemic in the global data, were also among the top five most mentioned problems in seven out of the nine focus countries.

The focus on financial and employment disputes on social media globally and in the countries where HiiL works can likely be attributed to the global economic recession that followed mass business closures and put many people out of work. Crime and domestic violence mentions are likely related to increased financial distress and strict lockdown policies in many low- and medium-income countries.

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Country view

In the following charts, we invite you to explore the social media data specific to the nine focus countries. By selecting one country or comparing two of them, you can discover which kind of justice problems people talk about most, how that has changed over the course of the COVID-19 crisis, the language people use to describe the problems they are facing, and the way they feel about them (on average).

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Comparison view

In the charts below, you can explore pairs of countries and compare how social media users talk about their justice problems during the COVID-19 pandemic. 

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Sentiment analysis

Justice problems in COVID-19 have negative connotations. Messages about crime and domestic violence are the most pessimistic.

How people talk about justice on social media

On average, the sentiment score of the messages classified as referring to a justice problem is 38. This is below the middle of the scale and indicates negative feeling. The average sentiment score of the messages classified as COVID-19-related is 45. A 10-point difference is considerable. The implication is that COVID-19 is concerning particularly when people see it in relation to (or as the cause of) a justice problem.

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Click here to load the graph.
It is being generated in real time. So, please bear with us.

Sentiment over time

Messages classified in the crime and domestic violence categories have the lowest sentiment score – 27 and 26 respectively – on the scale ranging from 0 to 100. Texts about public benefits, housing disputes, and medical bills have the highest sentiment score. On average, texts about public benefits have a sentiment score of 47. However, this score decreases steadily after March and April 2020. 

01

Findings: Global level

Mentions of crime and domestic violence rise during the pandemic. People are concerned and anxious.

The increase in global mentions of domestic violence between March and July 2020 is consistent with what we know was happening on the ground: between March and July 2020, lockdown orders caused rates of domestic violence to rise in countries around the world.

The same cannot be said for the most dramatic post-COVID-19 social media trend: the spike in crime problem mentions. Between March and July 2020, talk of crime nearly doubled while actual crime rates fell in several countries.

American criminologists attribute the COVID-19 crime decline to the reality that the majority of crimes in the US are petty offenses committed by young people in peer groups – a type of crime made more difficult by lockdown orders. Meanwhile, rates of more serious offenses are often committed privately (or remotely) by individuals – including domestic violence, serious battery, and homicide – increased between March and July 2020 or remained the same. Similar trends were observed in the UK and Australia.

The discrepancy in social media and crime data suggests that the driver of social media mentions may not be direct experience with the justice problems under study. Rather, rising social media mentions may reflect increased anxiety around crime and domestic violence during the pandemic. The low average sentiment score of 27 and 26 for messages about crime and domestic violence respectively supports the conclusion that they were motivated by negative emotions rather than neutral and objective analysis.

Increased anxiety around crime and domestic violence might be the result of authoritarian COVID-19 measures imposed by some governments, anti-police riots in countries around the world, or a more general climate of uncertainty and disorder created by the pandemic.

It might also be explained by increases in in-home media consumption, including increased engagement with news, streaming services, TV, radio, and messaging services in countries around the world. Social media use in particular has spiked globally during the pandemic. A deluge of information – as well as misinformation – about the virus’ spread and related economic and political insecurity may have triggered increased concerns around crime and domestic violence. 

The same reasoning can be applied to medical bills, financial disputes, and employment problems, which were also discussed more frequently after March 2020 but later leveled out.

02

Findings: In the nine countries where we zoomed in

Justice problems are felt after the first wave of the pandemic, especially those related to housing and employment.

At the country level, the onset of the COVID-19 pandemic coincided with a significant increase in justice problem mentions on social media in Nigeria, South Africa, Kenya, and Uganda. Beginning in April 2020, these numbers spiked dramatically in Nigeria, South Africa, and Kenya and began to decline to normal levels in Uganda. 

By contrast, the COVID-19 pandemic has not seemed to have a significant impact on the number of justice problem mentions in Ethiopia, Tunisia, Lebanon, Mali, or Burkina Faso.

Three of the countries that experienced notable increases in justice problem mentions after the onset of the pandemic in March 2020  – Nigeria, Kenya, and Uganda – also shared housing as their most frequently discussed justice problem. Meanwhile, South Africans on social media primarily talked about employment problems.

In the four countries where justice problem mentions on social media did not increase noticeably – Ethiopia, Tunisia, Lebanon, Mali, and Burkina Faso – crime problems were the most frequently discussed.

03

Findings: What social media tell us

More and more people use social media to share views, ideas, and concerns. Justice problems are an important part of the conversation.

With the growing usage of social media, there is a need for more reliable and refined methods to extract knowledge from the vast amount of data. This report is a step towards a better understanding of how people encounter justice problems in their daily life. The results are intriguing. HiiL and its partners will continue working towards evidence-based people-centered justice.

About the authors

Dr. Martin Gramatikov, Director Measuring Justice

Isabella Banks, Justice Sector Advisor

Share

Your ideas and opinions

We want to hear from you

Your opinion, thoughts and ideas are important to us! 

Feel free to share them through the form.

Table of Contents

The graphs on this page are being generated in real-time using our big data storage solutions. Loading time might vary based on the quality of your internet connection.