Raul Magni-BertonProfessor of Political Science at Sciences Po Grenoble and at the Pacte laboratory

How would you define objectivity in the social sciences?

In the social sciences, as in all the sciences, objectivity depends on the quality of the evidence provided. Although the quality of evidence in the social sciences is generally lower than in the natural sciences, there are regularly "truths" that are very difficult to dispute, because there is so much heterogeneous and converging evidence. 

Thus, for example, it is well documented that the success of violent revolution, rather than injustice, is based on high levels of poverty, which drives individuals to seek income through paramilitary activities. This is consistent with quantitative studies: poverty and unemployment are major factors in the outbreak of civil wars, associated with income concentrated in a few hands. In sample surveys, too, a sense of injustice has less influence on involvement in violent activities than poverty and the desire to earn an income. Finally, ethnographic surveys show that, in civil wars, combatants are often not hostile towards members of the opposing camp, and sometimes even know friends and family there. 

This set of elements constitutes "objective" knowledge, in the sense that all the studies carried out confirm this idea. In general, the level of objectivity of the social sciences can be compared to that of judicial truths, which also deal with human affairs and are often based on several pieces of converging evidence. Like the social sciences, forensic truths are generally reliable, but you can never be sure of a miscarriage of justice, especially when information is lacking.

Is researcher neutrality possible and desirable?

Neutrality on the part of the researcher is not at all desirable. Research requires a great deal of time to reflect on possible events and their causes, to collect and process data. It requires strong motivation and hard work. For most researchers, these motivations come from the desire to promote their ideas and convince others. This attitude is contradictory to neutrality: you need convictions strong enough to devote time and energy to it, knowing, moreover, that these convictions can lead to error. 

Neutrality, provided it is possible, does not give rise to such motivations. Nor is it the best guarantee of objectivity. Indeed, as noted above, it is the ability to prove one's case to others that creates objectivity, not the personal convictions of those providing the evidence. If it weren't for Marxists wishing to demonstrate that revolutions are guided by the presence of class consciousness, we wouldn't know that this thesis is not congruent with what we observe empirically. If the lack of neutrality on the part of researchers is a resource, rather than a problem, it must also be stressed that scientific progress is only possible if researchers' convictions diverge. Indeed, the quality of evidence is never as high as when there is controversy. So, for example, if opponents of the extreme right ask "why are there so many votes for the extreme right?", supporters of the extreme right ask "why aren't there enough of them? But the more ideas there are in circulation, the more likely it is that, after rigorous examination, one of them will prove to be true. 

How important are methods to you as a researcher?

Methods play a central role, as they are what distinguish a researcher from a non-specialist. Everyone has good ideas, but researchers can go out and test their ideas, turning them into true or false ideas. Without methods, scientific controversies are no more interesting than any other kind of controversy. Methods are used to decide between divergent ideas. 

So what is a method? A method is a consensual procedure for getting everyone to agree on a piece of evidence. For example, the proposition that "confidence in politicians is falling" can be tested against reality, provided we agree on the method. We need to know between which dates to observe this phenomenon, and within which spatial limits. We need to know how to measure trust in politicians and, once measured, how to go about finding the information. It's very important that those who believe and those who don't believe in the proposition "trust in politicians is falling" agree on these various parameters that allow us to verify whether it's true or false. For this reason, it is very important that there is as broad a consensus as possible on the methods. 

Just as diversity of ideas is a source of richness, so consensus on methods is essential if they are to fulfill their role. This consensus does not mean that there is no diversity of methods. In the social sciences, there are many methods - qualitative, quantitative, experimental, simulations, etc. - all of which have their validity. However, they must all be used rigorously. Rigor means the ability to create a broad consensus. It is precisely on this requirement for rigor that we need a consensus, which is lacking today in the social sciences, it seems to me. 

Could you present an example of research, ideally from your own work, to illustrate the issues and tensions surrounding objectivity and neutrality in the social sciences?

In recent years, one of my research topics has been terrorism. It's a delicate subject, as it's responsible for some very recent tragedies, and for political debates that are always very heated. At the time, it was a question of a political Islamism with a totalitarian vocation, aiming to punish blasphemy and behavior that didn't conform to this ideology, such as French secularism. On the other hand, until recently, public debates gave little space to the idea that these attacks were aimed at punishing France for its military commitment against several Islamist groups, including the Islamic State, responsible for several attacks around the world. 

Yet this hypothesis has been widely corroborated by studies of transnational terrorism in general, and reaffirmed by several terrorists themselves. Until now, I thought that this absence from the public debate was probably due to the propaganda phenomena that exist in war-torn nations, fed by an abnormally large amount of information that remains confidential. However, it's clear that, as soon as I agree to transform my personal opinion into a scientific thesis, I have to be prepared for reality to prove me wrong. 

With two colleagues, we collected all the attacks claimed by the Islamic State around the world, as well as each country's intervention in Iraq, its bombings, its laws on blasphemy and secularism, and other factors such as the presence of a civil war, the distance from Iraq, the number of Muslims, discrimination against Muslims, and so on. The results of this study had the effect of "radicalizing" me politically. Indeed, the results went far beyond my suspicions: bombing in Iraq is by far the main predictor of terrorist attacks on its soil. Other factors play a lesser role, but secularism or blasphemy have no effect.

Of course, it's always possible that we've measured them wrongly, and we're currently looking for other measures. But as it stands, we can conclude that not only is the main reason for terrorist attacks not mentioned by our political representatives, but that the one that is mentioned is, in the current state of our tests, non-existent. Here's an example where I, as a researcher, wasn't "politically neutral" to begin with, and where reality turned out to be even "less neutral" than I was.