UCL Mathematical & Physical Sciences


How does news spread amongst a population?

30 April 2019

How long does an event or a topic remain in the spotlight? Are big cities more newsworthy than small? Following the 2017 Puebla earthquake, a research team including UCL academics investigated the behaviour of the main media channels in the country, and obtained striking results.


On the 19th September, 2017, Mexico was devastated by the Puebla earthquake. The event caused widespread damage in the Mexican states of Puebla and Morelos as well as the Greater Mexico City area, including the collapse of more than 40 buildings. Around 370 people were killed by the earthquake and related building collapses, and more than 6,000 were injured.

In order to understand the spread of this type of news amongst a population, the research team focused on analysing the coverage that different media channels gave to different cities both in ordinary circumstances and just after the occurrence of the 2017 earthquake. By taking into account the online newspapers with the largest audiences, a significant proportion of the national written news output is captured.

As well as analysis of print and online newspapers, the research team also included data from the newspapers’ activity on their associated Twitter accounts. Each time an update to the corresponding newspaper's website is issued, this is usually announced, at least once, via a related post on the media outlet’s associated Twitter account. Therefore, with little effort or cost (compared to that of producing the news update), mass media outlets can reach a broader audience by promoting their updates on their social media channels.

To quantify the coverage given by a set of media outlets to a city and a topic, the researchers defined a media coverage index for each city. Only cities where the impact of the earthquake was “moderate” or higher according to the Modified Mercalli intensity scale were included. The resulting index is a weighted average of the proportion of tweets published by each newspaper that mention the earthquake, weighted by the number of followers on their Twitter accounts. 
In ordinary circumstances, the media coverage index accounts only for the proportion of news devoted to each city. When the period just after the earthquake is considered, the media coverage index takes into account the proportion of news related to both the earthquake and each of the affected cities.

Media activity was high close to the occurrence of this event and on the following day, slightly more than 80% of the news items published by the media channels included in the study were related to the earthquake. The proportion of news related to the earthquake stayed high for the next few days, however, one week after, it decreased to 25%. One month later, only 10% of the news items were devoted to this topic.

By reviewing the media lifecycle related to the earthquake, the interest in the topic decays exponentially: every day, the amount of news items about the earthquake reduces by 10% with respect to the previous day. This means that if on the day of the earthquake 100% of the news were earthquake-related, it is possible to estimate that the day after, 90% would keep talking about the earthquake; two days after, 81%; three days after, 72.9%; the fourth day, 65.6% etc.

During ordinary circumstances, it was found that the media coverage index scales as a power-law of city population size, with a scaling exponent of 1.37. This means that the media coverage index per 100,000 inhabitants for Mexico City is 2.5 times larger than for cities like Puebla or Toluca, which have a population that is roughly ten times smaller. In summary, the media coverage per 100,000 inhabitants scales superlinearly as the city size is increased. The media seldom talk about events that occur in small towns and therefore, Mexico City receives a disproportionate amount of coverage compared to the others.

In the immediate aftermath of the Puebla earthquake, distribution of media coverage changed. Under normal circumstances, larger cities receive more attention from the media per 100,000 inhabitants. During the first few weeks following the occurrence of the earthquake, approximately the same amount of news items per 100,000 inhabitants were devoted to cities of different sizes such as Puebla, Cuernavaca and Mexico City. The distribution of the coverage normalises after approximately four weeks.

“Usually, the more populous a city is, the more coverage per capita it gets from the media. However, our research has found that this behaviour is no longer observed during the weeks following the occurrence of an earthquake that affects a whole region involving cities of different population sizes,” said Carmen Cabrera Arnau, PhD Student, UCL Mathematics.

Given the influence that the media has on society, the observed behaviour can have important consequences: the media can determine the amount of attention and financial aid that the governments and other institutions give to the affected regions. Therefore, the amount of coverage that they give to a city after an emergency situation such as an earthquake, is crucial for its recovery process. 



  • Social media concept. Technological abstract background © metamorworks