Mood assessment - How COVID-19 influenced media tone in the EU

Image by Obi Onyeador from Unsplash

Not surprisingly, perhaps, news can have an impact on society, but news are also a reflection on society, as illustrated in this study. It’s more of a two-way effect, if you will. It is also fairly well established that news events tend to be reported in a negative tone, as illustrated in this study. However, this negativity bias differs between countries. This brief assesses how the COVID-19 pandemic has influenced countries across the EU, and whether this influence differs between countries with a low versus high negativity bias.

Data

GDELT is a great (open) source of news coverage data, and I’d encourage you to read more about the GDELT Project. GDELT data was collected from March 2015 until August 2020 (66 months) from the 27 EU member countries. All the data was aggregated at a monthly level as follows:

  • month_year: represents the month during which the news events were observed.
  • count_of_news_events: its the total number of news events identified by GDELT in a given month, in a given country.
  • average_tone: the average tone of a news event is computed by making use of Google’s in-house algorithms to measure the tone with which the news event was reported across all news articles recognized to depict such an event. The tone is considered neutral at 0 and, although it ranges from 100 to -100, most observations range between 10 and -10. This variable is obtained by computing the average of all events observed in a month, in a given country. If you’d like to learn more about the tone and its significance, have a look at the GDELT documentation.

Missing data: Romania only had 42 months observed while for Slovenia there was no news data available over the period. In addition, the number of news events varied considerably between countries:

Notably, the number of news events were very low in Romania and including it in the analysis could lead to misleading results, so it will be dropped from the analysis.

Analysis

The plot below illustrates that the tone of news has become more negative since the COVID-19 pandemic has started, though it appears to have recovered somewhat since.

Note that on the 30th of January 2020, the World Health Organisation (WHO), declared the COVID-19 pandemic a “global health emergency” - so I assume that from this date onwards the COVID-19 had had a meaningful impact on the news across the EU, even though news of the COVID-19 broke out earlier.

Before assessing the impact the differences in impact COVID-19 may have had between positive and negative news countries, how the negativity bias varies between countries needs to be established.

The plot above illustrates the monthly news’ average tone and 95% confidence interval, and it indicates that news’ tone varies substantially across countries.

While there is some variation in the way news are reported within a country from one month to another, differences in tone across countries seem to be persistently different. In Portugal, for example, the news tend to be reported in a considerably more positive tone than in France.

So, how has the COVID-19 pandemic impacted the the news’ tone across countries? To answer this, countries with a low negativity bias are compared with countries with a high negativity bias. But first two groups are formed - one with countries with an average tone below the average, and another made up of countries with an average tone above the average. The groups are composed as follows:

  • Low negativity bias group: Estonia, Latvia, Portugal, Luxembourg, Bulgaria, Cyprus, Finland, Latvia, Ireland
  • High negativity bias group: Germany, France, Belgium, Sweden, Greece, Netherlands, Hungary, Austria, Poland

Select countries visualisations and social-media posts

To illustrate how some of the high and low negativity bias countries were impacted, three graphs are produced. The first illustrates the difference in negativity bias between the countries. The latter two illustrate how the select low and high negativity bias countries, respectively, were impacted by the COVID-19 pandemic. Note that the plots were produced for social media sharing, so I modified the size.

As the plot above illustrates, the differences between low and high negativity bias countries are considerable. Regardless, both groups seem to have been equally impacted by the COVID-19 pandemic, as illustrated in the two plots below.

Conclusion

I started this brief because I wanted to see how the COVID-19 pandemic impacted the tone of news. Armed with data from GDELT, this brief analysis illustrated that the COVID-19 pandemic has dampened the mood across the EU. It also showed that, although the tone of news tends to be negative across the EU, there are vast differences between countries. In spite of these differences, the effect COVID-19 had on the news seems to be just as bad.

Dragos Tomescu
Dragos Tomescu
Data Trainer (previously, Data Analyst)

A data analytics professional with a passion for understanding society. I write about data-driven applications and their impact on business and society.

Related