We examined more than 1,300 journalist killings between 2002 and 2013. Here’s what we learned.

A journalist from the Balochistan Union Of Journalists holds a photograph of his colleague Imran Sheikh, who was killed in an explosion the day before, during a silent sit-in to protest bomb blasts and condemn the killings of members of the media outside the Quetta Press Club in Quetta, Pakistan, on Jan. 11, 2013. (Naseer Ahmed/Reuters)

Reporting the news from a war zone is a risky business. But just how many members of the media are killed — and what does that tell us about the trajectory for how countries view human rights?

To look at these questions more closely, we collected data on killings of journalists and media personnel that occurred from 2002 to 2013. We built on the work of the Committee to Protect Journalists, Reporters without Borders, and the International Press Institute.

Here’s what we found: There were over 1,300 press corps deaths between 2002 and 2013. In our just-published study we find some surprising results that underline why we should be paying more attention to these deaths:

1) Journalists are killed all across the world — well beyond the battlefield.

Journalists die reporting on stories across the globe, and not just when caught in crossfire. Roughly one-third of all killings between 2002 and 2013 occurred in countries that were not involved in an armed conflict. The figure below shows where journalist killings tended to occur.

Journalists are a big target in Mexico, one of the most dangerous countries for members of the press worldwide, but also in Brazil, the Philippines, Indonesia, Nepal and Egypt.

A number of countries are hotspots in the global map of where journalists are killed.

2) Outside of war, more journalists are killed in countries where repression is limited.

The deaths of journalists also occur in countries that aren’t considered repressive regimes. Outside of armed conflict zones, most journalist killings actually occur in countries with some, but not extensive, repression. Figure 2 divides the extent of state-sponsored repression into five categories, based on the Political Terror Scale. “Medium repression” refers to countries where political imprisonment, murder and execution are extensive, but have not expanded to the whole population. This is the context in which most journalists are killed outside of war.

Most journalists killed on duty died in countries with a medium level of political repression.

3) Most journalists are not murdered by criminal gangs.

Outside of war, many journalists are killed by state agents — not by criminal gangs, as is often assumed. Or the perpetrators remain unconfirmed. While the large proportion of deaths at the hand of unconfirmed perpetrators hints at how difficult it is to bring those responsible to justice, it does not adequately reflect the extent of the problem. The Committee to Protect Journalists suggests that the killers go free in nine out of 10 cases.

Many of the countries with the highest number of journalists killed almost never bring the perpetrators to justice. These countries include Russia, Brazil, Pakistan, Bangladesh and Mexico.

It is hard to say who is behind these killings by unconfirmed perpetrators. It is relatively easy to identify criminal gangs or nonstate groups, such as ISIS, as perpetrators. Governments are likely to be keen to uncover the involvement of opposition groups or criminal gangs, as this could help to justify harsher state responses. But states have few incentives to follow up and confirm the killing where their own forces were involved.

In many cases, it is never clear exactly who was responsible for the death of a journalist.

4) Journalist killings are often harbinger of worse to come.

In our recent study we show that the killing of even a single journalist can act as a warning of worsening human rights conditions. We find that torture, killings, political imprisonment and disappearances of people generally become more likely in the two years following the killing of a journalist — regardless of who committed the crime.

Murdering a journalist generally signals instability and increasing tension, which are followed by increasingly invasive and harsher government behavior. When we account for the killing of journalists in predicting human rights protection more broadly, we can identify countries that are unlikely to improve their human rights records — even when other characteristics, such as democratization or economic development, would suggest that things are getting better. A stronger economy is unlikely to make up for the risk that is highlighted when a journalist is murdered.

5) In countries that generally respect human rights, journalist killings are particularly pertinent warnings.

Few analysts would expect a substantial improvement of human rights in Syria or Sudan — or predict a sudden turn for the worse in Norway or Canada. It is much harder to predict what will happen in countries that are somewhere in the middle.

This is where it really helps to pay close attention to the fate of journalists: The death of journalists can help identify those countries with “average” levels of repression that are most at risk to further deteriorate. These are also the countries where predicting future human rights is the hardest — but also the most important from a policy perspective. Policy initiatives are likely to have the biggest impact in these moderately repressive countries, because achieving improvements in highly repressive countries is extremely challenging.

The killing of journalists can therefore be a pertinent early warning signal for deteriorating human rights in moderately repressive countries. Examples of such countries would be Peru, Sierra Leone, Malaysia and Tanzania.

Anita Gohdes is assistant professor of international relations at the University of Zurich. Find her on Twitter @argohdes.

Sabine Carey is professor of political science at the University of Mannheim. Find her on Twitter @Sabine_Carey.

The project received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC Grant Agreement no. 336019.