Guillaume RouxResearcher at the Pacte laboratory and lecturer at Sciences Po Grenoble

How would you define objectivity in the social sciences?

Objectivity is a way of guarding against the uncontrolled effects of each researcher's own subjectivity: the unconscious tendency, in particular, to favor observations that validate our beliefs or political preferences. Against this "subjectivist bias", as it may arise at the moment of fact-finding, academic institutions offer a set of guarantees: the consolidation of tried-and-tested methods; the obligation for researchers to show their "kitchen" - their research protocol, the way they make findings or draw conclusions - and the critical discussion and validation of results by peers.

This makes the profession demanding, and offers serious guarantees. But can research claim a form of pure objectivity or "universal truth"? That's what Pierre Bourdieu, who took this question very seriously for sociology, argued all his life. Suffice to say, I'm not convinced. This implies that we should always be able to reject the argument of scientific authority: "specialists agree that...". (see the "reproducibility crisis" in medical science and psychology). There's only one condition for this: you must have good reason to discuss the findings you're contesting, based on the way in which they were established. The rest is denial - refusing to know what's wrong.

This position does not imply relativism in the strong sense. Like Max Weber - or, in a different way, Michel Foucault - I would defend a certain perspectivism: we can establish the "truth" of a phenomenon or event from a certain point of view (from the point of view of a discipline, e.g. economics, sociology...). Nevertheless, this vision is partial, and in a sense biased (in the sense that it implies a choice of perspective - if only in the choice of concepts, which amounts to opting for a certain a priori division of reality). Rather than seeking to unify perspectives (the dream of unified knowledge), I would propose to multiply them (a certain pluralism). 

In any case, I think it's important to reject the alternative that either "science produces universal truths", or all research findings are equally and completely "relative" and questionable. It's part of the job to estimate the degree of confidence we place in certain research findings or conclusions. This may differ according to whether or not these results imply the idea of causality, for example, or mobilize concepts whose meaning is not fixed (violence, intelligence, racism, science...).

Is researcher neutrality possible and desirable?

In France, the term "Neutralité axiologique" (axiological neutrality) has become a poor translation of a Max Weber concept. What is a "neutral" researcher? This notion sometimes seems to refer to the ideal of a researcher who is indifferent to his or her object, "without preference", or whose research has no aim of social or political transformation - contrary to the very project of sociology (as Aron reminded us, for example). Equally "neutral" are researchers whose implicit axiological preferences are seen as "reasonable" rather than "radical": constituting a form of shared evidence, they easily go unnoticed, which does not make them "neutral" for all that.

Different is the concept of "freedom with regard to one's own values"(Wertfreiheit), which corresponds more closely to Max Weber's idea. It invites us to distinguish between a normative level on the one hand, and a factual one on the other. In the context of research, from the choice of an object or a question, to the conclusions we propose to draw for the "common good", everything can be a function of normative preferences, except one thing: the establishment of facts. We need to be able to agree on the objectivity of a factual observation, valid for everyone. Distinguishing between facts, on the one hand, and their normative interpretation, on the other, seems essential: if we confuse the two, we can say just about anything. Which brings us back to our first point: the important thing is not the normative or political preferences of a researcher - or his hypothetical "neutrality" - but his ability to objectify, first of all, certain facts; or to produce, to put it better, objectivations.

In general, I think it's a good idea to maintain this distinction between facts and values as a "reading principle". That said, it's not as obvious as it might seem: the "truth criteria" set by an era, a science or a body of knowledge are not completely independent of established values or power relationships. Desrosières has shown how state action can contribute to the legitimacy, even in the scholarly sphere, of certain indicators, measures or procedures - think, for example, of the prestige acquired by statistics. Failing to elaborate, we can refer to authors such as Feyerabend, Foucault, Isabelle Stengers or, more recently, sociologist Andrew Abbott.

There are discussions in the social sciences about the normative and even political implications of methods - discussions that a young researcher must first master. This reminds us that instituted knowledge is never "outside society", alien to its norms and power relations. The academic world can see itself as a field of struggle - between disciplines, approaches or currents. One example is the way in which neoclassical economics came to dominate the academic field, while at the same time attempting to steer public policy. 

What's more, the "discourses" produced by researchers circulate in the social and political world, and produce effects. Should researchers ignore them? The strict separation of political and scholarly registers shows its limits here: even if they are intended to be "cumulative", the research we carry out has consequences here and now, in a given social and political context. I think it's legitimate for researchers to take this into account when considering the type of research they produce.

In general, what seems to me to be lacking is a reflection on the multiple purposes that the human sciences aim at or can aim at - even if this is less necessary in the sciences of matter, whose benefits are more readily thought of in terms of technology. And on what is at stake in the choice or elaboration of concepts, which are never "neutral" or scientific in the true sense of the word - an issue pointed out by Durkheim (too complacently), then Bourdieu, or Foucault in a different way.

How important are methods to you as a researcher?

Methods play an important role. I became critical of my initial approach (without abandoning it), which was quantitative. This gave rise to epistemological and methodological reflection, which led me to invest in different methods. The statistical approach can be favored when it comes to reasoning "on average", or in a representative way. In some fields of research, statistical studies are lacking; in others, their domination has, in my view, become problematic. 

When it comes to ordinary citizens' perceptions of the police, for example, international studies are mainly quantitative. The focus is on the explanatory variables of a phenomenon that seems to remain, in itself, poorly understood - we don't really know what the respondents are talking about, or what they have in mind when they make judgments about "the police". What are they thinking about? And how circumstantial is the answer to this question - is it very different in so-called "working-class" neighborhoods, something that quantitative research is currently unable to determine? If there is sometimes a lack of statistics or quantitative studies (as was the case in the same field of study in France), here there seems to be a lack of observations that would enable us to get to grips with the phenomenon under study - to describe or document it. In certain research traditions, it seems to me that the idea of "description" could be re-evaluated, in connection with an epistemological reflection, which would show that the distinction between describing and explaining is not so obvious.

Surveys of individual and group interviews have led me to ask a different kind of question: how can we analyze and objectify the implicit in discourse? Often, very powerful things are said in hushed tones: how can we first be attentive to them - they are sometimes things we might not pick up on - and then decipher what is said, by "connecting the threads" or elements that may echo each other in the course of an interview - which makes it possible to objectify, to a certain extent, this work of interpretation. The more time we spend on this, the more the things said in the course of an individual or group interview - or a public meeting, for example - reveal themselves to be rich in latent meaning. Typically, we realize that a banal remark is in fact a response to something said earlier, or to a type of discourse circulating in the social arena.

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?

I've worked and worked again on the "racial question" in different ways, each time by chance. My motivation is primarily intellectual - but it doesn't really matter - to contribute to the understanding of power relations in the broadest sense, and to transform my own view of the social world. In particular, I study the phenomena of identification with categories such as "whites", "blacks and Arabs" and so on. The implicit and explicit use of these common-sense categories in France, in various fields, by ordinary individuals as well as those involved in public action, is a proven fact. But the recognition of this fact, and research on the subject, is met with resistance. 

I believe that the people who have tried to prevent this research, in the name of political and ideological preferences that may be respectable (such as a Marxist vision of society, a "color-blind" republican ideology, the choice of a form of ignorance that would avoid opening a "can of worms", etc.), see themselves as "more neutral" than researchers who conduct empirical research on these subjects for the purpose of knowledge...

As a researcher, I choose the notions that seem to me to be heuristic, or that can reflect my observations. In the academic world, the word ethnicity has met with resistance. The sociological notion of race - the meaning of which, though different, is relatively clear, provided one is informed - is considered even more sulphurous. Let's say I'm "politically indifferent" to all this: I still have to use or not use this word (which is useful to me in this case). In either case, whether I like it or not, I'm positioning myself in an academic "field of forces" that is no stranger - here again, whether I like it or not - to power relations that transcend and run through it.

The choice is yours. It's worth remembering that this choice is largely based on the need for knowledge. Notions such as racialization or whiteness, for example, which are polemical and politically divisive, have an obvious heuristic significance. By drawing attention to phenomena that have remained largely unnoticed and little studied, they make it possible to renew the "terrain" and understanding of the mechanisms of racism, and even to act on them. I am by no means indifferent to the political implications of my research, and they are an integral part of the taste I have for my work.