School attack: please do not pay attention to the tragedy – 03/28/2023 – Bernardo Guimarães

The murder of the teacher at the school dominated the news and social networks. Have we paid too much attention to the tragedy? Could all this attention spur new attacks?

This question illustrates well the difficulties of establishing cause and effect relationships in social sciences such as economics. But despite the obstacles, there are methods that, combined with the creativity of researchers, manage to find reliable answers.

I start with the difficulties.

Data from school killings and press reports can point to patterns. For example, we can be shown that after school murders, there is on average more news; and following the news, there are more incidents on average.

This, however, does not allow us to unravel what is causing what. It may be that news encourages attacks. But it could also be that other factors are driving more violence and more news. In both cases, we would see more bombs stamped on newsstands and exploding in children’s schools.

To establish cause and effect relationships, researchers in the natural sciences conduct experiments.

At the age of 7, we planted a bean in a pot and left another bean without seed. Then we compare the two.

To test coronavirus vaccines, we randomly select a treatment group that gets the vaccine and a control group that gets a placebo. Then we compare the results.

But for most questions in the social sciences, this approach doesn’t work.

We were unable to randomly assign different monetary or fiscal policies to groups of countries. The deficits and interest rates of each country are not drawn, they are chosen, and thus reflect characteristics of each country that also affect the economy. Therefore, we cannot learn just by studying the patterns in the data.

In the same way, what appears in the newspapers and goes viral on social networks is not chosen by lot.

Empirical researchers need to find ways to find random variations in press coverage. These random variations create something like treatment and control groups.

For example, natural disasters (earthquakes, tsunamis) happen randomly to us (we cannot predict their occurrence) and reduce the space in the media for other news.

If a major earthquake had occurred in the United States on Monday (27), the newspapers would have given a little less space to the murder of the teacher at the school in São Paulo.

On the other hand, it is unlikely that this earthquake, by itself, would affect the occurrence of violent events in the coming days here in Brazil.

So, events like this earthquake create a group of events with less repercussion in the media for a random reason – something that happened in another country, unrelated to violent acts in Brazil. This is sort of a control group.

The treatment group includes the events with the greatest media coverage due to the lack of other hot topics on the day.

Then, using statistical techniques, we compare what happens next in the treatment and control groups.

This is a very simplified explanation of the instrumental variable estimation technique, widely used in social sciences such as economics.

Work comparing these treatment and control groups shows that greater media coverage actually causes more terrorist actions and mass killings. The effect is big. [referências aqui e aqui]

According to empirical evidence, our attention to murders like Monday’s encourages similar incidents.

PRESENT LINK: Did you like this text? Subscriber can release five free hits of any link per day. Just click the blue F below.