François BonnetDirector of research at the Pacte laboratory and lecturer at Sciences Po Grenoble, @frbonnet
How would you define objectivity in the social sciences?
Objectivity means taking into account facts that are embarrassing to one's opinions. All academics have opinions and are tempted by confirmation bias, i.e. they are tempted to be positivist with the facts that go their way and constructivist with the facts that are embarrassing for their theories and values. Another technique for ignoring non-cooperative results is ultra-positivism: it consists in demanding an extraordinarily rigorous threshold of proof, which of course is never applied to favourable results.
To claim objectivity, then, is to submit one's ideas, conjectures and intuitions to empirical tests, and to heed the verdict of these empirical tests. A good way of claiming objectivity is to confront one's work with the malicious gaze of someone who shares neither our opinions nor our values, but who uses the same logical and methodological tools to assess the seriousness of an assertion. But as long as we take into account results that are embarrassing to our opinions, and don't alternate postures (positivist or constructivist) depending on how the facts fit in with our values, we can reasonably claim to be objective.
A second problem linked to objectivity concerns the choice of subjects in the social sciences. The decision to devote time, energy and taxpayers' money to one subject rather than another cannot be innocent. If we live in a society that has been disrupted by environmental catastrophes or is in the throes of civil war, and we choose to study the length of a dog's leash, we can do so in a perfectly objective way, but we must not lie to ourselves: not to work on the central problems of our time is to deny that these problems are important, and to take refuge behind an aesthetic posture that is as contrary to the scientific approach as variable-geometry objectivity.
Is researcher neutrality possible and desirable?
On an individual level, the notion of neutrality in the academic world primarily concerns teaching. The pedagogical relationship is doubly unbalanced: firstly, because the teacher grades the students, but above all, because most of the time, he or she decides on the entirety of what will be discussed in the course (themes, readings, authors, etc.). Insofar as the teacher is virtually omnipotent over the content of his or her course, honesty requires exposing students to the widest possible range of facts and interpretations, so as to avoid the course becoming a one-sided reflection of the teacher's inclinations. This requirement is important both from an intellectual point of view and for the legitimacy of the academic institution.
Of course, it's unrealistic to expect teachers-researchers to be neutral: they are human beings formatted in a given ideological space-time. On the other hand, I think it's very important for the academic world to be perceived as globally balanced in terms of diversity of opinion. If everyone thinks that economists are vile neo-liberals, it's bad for the legitimacy of economists, even among audiences a priori sympathetic to the cause of neo-liberalism.
Indeed, the legitimacy of scientists rests on the general public's belief that the academic community respects the norms of theethos of science: disinterestedness, organized skepticism, the idea that knowledge is a common good, and more generally the general principles of positivism - namely, the primacy of explanation and quantification, the idea that the production of scientific truth is independent of the personality or opinions of academics, the independence of science from morality or power, the principle of the cumulativity of knowledge- . Anyone involved in the academic world knows the distance between reality and principles. But I repeat: our legitimacy depends on our ability to keep the general public believing that we subscribe to their theory of science.
A good way of meeting the expectations of the general public is to conform to the norms of theethos of science. Conversely, when an academic community openly defies these norms, when it shows itself too openly in the media in the service of a well-identified segment of the ideological space, it destroys its legitimacy. It is for this reason that the academic community would do well to respect these norms, and at the very least to ensure that forms of ideological diversity are maintained within each discipline.
How important are methods to you as a researcher?
In my opinion, the question of methods is first and foremost one of research design. At the start of my career, I tended to evolve within epistemic communities oriented towards inductive approaches: we take an interest in phenomenon X, we go and see it "in the field" (i.e., using qualitative methods), we carry out observations and interviews, and from these research operations emerge an interest in phenomenon Y, which we realize is far more interesting, relevant and evocative than X. And we write a dissertation, a thesis, articles on X and Y. However, I had the good fortune to be oriented very early on towards the comparative (qualitative) approach, and I was very struck by the benefits of observing X in two different countries. Unless you're very clever, when you observe X in a single context, many aspects of X appear to be trivial, banal, self-evident and so on. Comparing X in different contexts very effectively reveals variables (and invariants), and thus enables us to move beyond the merely descriptive.
The moral I drew from this (after many others) is twofold. On the one hand, there is an intrinsic epistemological superiority of research designs built to adjudicate between competing explanations over simple inductive approaches (of the "I'll go and see and I'll tell what's interesting" variety). And on the other hand, over time, I have also become convinced that there is also an intrinsic epistemological superiority to approaches that aim to provide statistical proof of a causal relationship. I'm writing this in the full knowledge that it wasn't this kind of approach that drew me to the social sciences as a student, that my favorite books are more in the realm of historical sociology, and I'm well aware that bad quantitative designs can be made. But if I reason on the scale of an academic community, I can see that the average (and I stress "average") quality of production increases as one moves towards hypothetico-deductive designs and sophisticated methods, and that it is generally preferable to train students in this direction.
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'm currently working on an indirect empirical validation of the principle of less eligibility. The principle of less eligibility is a notion introduced by English economists in the Victorian era, according to which welfare payments cannot be more generous than the low wages in a given society, otherwise the poor will prefer to receive welfare payments rather than accept low-paid work. An extension of the principle of less eligibility, formulated by Georg Rusche, a German Marxist criminologist before the Second World War, is that the treatment of convicted criminals also depends on the situation on the labor market: crime cannot pay more than the lowest wages, otherwise the poor are incited to crime.
In my book The Upper Limit to combine these two ideas, the Victorian principle on the relationship between the labor market and social policies, and the Marxist principle on the relationship between the labor market and penal policies: the "upper limit" is the ceiling effect exerted by low wages on social and penal policies. In Scandinavian countries, the high standard of living of low-wage earners enables relatively generous social policies to be pursued, and in turn, very restrained penal policies (incarceration rates of around 60 prisoners per 100,000 inhabitants, with the police killing virtually no-one). In Brazil, on the other hand, the very low standard of living of the most marginalized fringe of the labor market means that the welfare state is in its infancy, and penal policies are very repressive (350 prisoners per 100,000 inhabitants, 2,000+ people killed annually by the police).
Between the two - and I know there are other legitimate analyses of these phenomena - the contrasting cases of France and the United States: the standard of living of low-wage earners in France is lower than in Scandinavian countries, but still allows for rather generous social policies, and relatively restrained penal policies (100 prisoners per 100,000 inhabitants, 15 people killed annually by the police). In the United States, which is a much richer country than France, low wages are lower, the welfare state is less redistributive and penal policies are considerably more severe (650 prisoners per 100,000 inhabitants, 900 people killed annually by the police).
The principle of less eligibility implies that relatively high wages and relatively more generous social policies have a downward effect on crime (traditionally left-wing ideas); but also that low wages form the limit of the generosity of social policies, and that penal policies also have a downward effect on crime (traditionally right-wing ideas). I try to assess these proposals by the highest possible standards of proof, because I am aware of the normative stakes involved in these issues, and also because,in the final analysis, it is theory that must conform to fact, not the other way round.