No. Most of the analyses of the democratic peace have made used of data from the Correlates of War Project, initiated by J. David Singer. It’s a highly regarded project that for over 30 years now has served as the main database on international war and internal war. But it has a relatively high threshold of 1,000 battle deaths for something to be called a war.

Many of the armed conflicts today are low-level conflicts, which sometimes go on for a very long time. I’m also the editor of the Journal of Peace Research, and we started looking around for suitable database of comprehensive conflict data in order to publish annual updates in our journal. After considering several alternatives, we decided to team up with the Uppsala Conflict Program at Uppsala University. Since 1993 we have published an annual update on armed conflict with a threshold of 25 battle deaths in a given year. This gives us wider coverage than the Correlates of War Project.

That is a joint article between scholars at Uppsala and PRIO that we published after we had spent considerable time, effort, and money backdating the dataset to 1946, giving us a time series for the entire period after World War II. Our aim was to create a new standard dataset for the study of armed conflict, the first choice for scholars analyzing armed conflict statistically. I think we’ve had some success. The subsequent updates are also frequently cited, but this was the first article with a long time series.

No, their data are still being used by many people, including ourselves. We’re not aiming to make them obsolete, but to supply a data series that includes the smaller conflicts. We are also trying to add new features to the database, such as more accurate start and end dates for the conflicts, data on the geographical center of the conflict and extent of the battle zone, and revised data on battle deaths.

That’s quite arbitrary. We just had to set a threshold and we had to be consistent. We picked 25 because we figured it was high enough to be a significant conflict and low enough that it would include most of the conflicts we were interested in. Some conflict datasets have thresholds as low as a single death, or even none. The main reason we wanted a higher threshold was to avoid bias by over-reporting conflicts in open societies with accurate news reporting. Armed violence causing more than 25 deaths in a year is likely to appear on the media’s radar screen, almost regardless of where it occurs. Our colleagues in Uppsala who record the information are dependent on news sources for a great deal of their coding, particularly for the more recent conflicts.

That means military or civilians killed in battle-related violence. It does not include indirect deaths—people dying of starvation or disease in the aftermath of a conflict. The Uppsala/PRIO dataset as presented in the 2002 article and subsequent updates does not include the absolute number of battle deaths. But it does have two thresholds: 25 deaths for being recorded as an active conflict and then 1,000 battle deaths for being coded as a war. In two articles principally authored by Bethany Lacina and published in 2005 and 2006, we describe a new dataset on battle deaths, matched to the basic conflict data. Obviously, it’s early days yet, but we hope that the battle-deaths dataset will also become a standard tool for conflict researchers.

We find, contrary to what you might believe from the media, that conflicts have actually become less bloody over time since World War II. If you look at a longer time period—the entire twentieth century—you see it’s really an inverted U-shaped curve. Wars became more deadly through World Wars I and II. Of course, if World War III had broken out as a nuclear war between the two superpowers, that would have been even deadlier. After World War II, the risk of getting killed in battle worldwide has decreased substantially. But battle deaths are very dependent on single wars. The curve for battle-deaths has a large peak for the Korean War, another one for the Vietnam War, and a third for the Iran-Iraq War and the Soviet-Afghanistan war. But those peaks are progressively lower, so the overall impression is a curve that declines. The current war in Iraq could reverse that trend, although I believe that war-weariness will affect the US long before the casualties reach the levels of the Vietnam War. A nuclear strike in the Middle East would also ensure a dramatic reversal. We are dealing with low-probability events, but potentially extremely hazardous ones.

That is the question we would most like answer, but it will take a lot more work than we have done to date. Going back, though, to my first point about the democratic peace, I think that one factor is that a higher share of countries are democratic today then ever before and a higher proportion of the world population live under democratic governments.

Another factor is the end of the Cold War and the fact that superpowers are no longer fuelling wars around the world. Both the Korean War and the Vietnam War started out as local conflicts. Then the superpowers got involved on the two sides, and the wars ended up being a lot larger and bloodier than they would otherwise. There were numerous conflicts like that in the Third World during the Cold War, although not quite as deadly as those two. With the Cold War now gone, I think the international community is more concerned with settling conflicts than with fighting by proxy. The UN has been able to play a much more active and constructive role, although it failed to prevent the 2003 Iraq War.

Apart from the problem of trying to understand a complex world, I am also struggling to understand what my younger colleagues are doing, in terms of advanced statistical methods, mathematical modeling, and so on. I realize that I’m very fortunate to work in an active research environment with lots of talented young people, so this is obviously a luxury problem.

It’s one of them. Most of my highly cited articles are co-authored. This is quite a normal pattern in other fields. Indeed, in the hard sciences, it would be pretty unusual to have a single-author scientific paper that isn’t an editorial. This is becoming much more common in social sciences, particularly in those areas where people make use of methods inspired by, and sometimes borrowed from, the hard sciences.

My research now falls into three major areas. We’ve been talking about democracy and peace, and about developing new conflict data. The third area is the question of environmental factors in conflict, and this is my main research area at the moment. The working group that I head within the Centre for the Study of Civil War is on environmental factors. In other words, how do factors like resource scarcity or abundance influence the probability of conflict? There’s been a lot of interest in that area, particularly after the end of the Cold War. Many scholars as well as policy-makers thought that the end of that ideological conflict would lead to an increase in other forms of conflict, such as ethnic wars or struggles over scarce resources. In the work within my working group at the center, we find that resources do matter for conflict. But my view is that the apocalyptic scenarios for global scarcities are considerably exaggerated.