Sebastian Roché, Director of Research at the Pacte laboratory and lecturer at Sciences Po Grenoble, @sebastianjroche
Is researcher neutrality possible and desirable?
Objectivity and neutrality are two very different concepts. It is essential to distinguish between them. Value neutrality, or autonomy in relation to values, can be understood in two ways. On the one hand, neutrality relates to the choice of research questions and objects, or to the method we use to study them: for example, do we want to look at the social and economic causes of delinquency or deviance, or at the effectiveness of police patrols in deterring them? And are we going to do this with an interpretative approach, seeking to understand the logic of individuals, whether delinquents or police officers, or by verifying statistical laws (for example, the fact that children who are malnourished and receive less maternal affection have more frequent violent behaviour in adolescence?) On the other hand, neutrality means distancing oneself from positions taken in public debates, i.e. in political arenas where confrontations between parties or movements are organized around values such as equality, freedom, or even secularism, recently promoted to the rank of a value.
Is it possible to be value-neutral in both directions? Could a researcher not be influenced by the wider professional, social and economic environment in which he or she operates? It's simply not possible. There is no social science free of the inner predisposition, subjectivity and passion for their work among researchers. Value judgments, and therefore ideals, cannot be absent from the process of knowledge production. If we can assign neutrality as a value to an administration, for example, so that its agents treat everyone regardless of their social status or skin color, we can't impose that a researcher be unaffected by societal issues such as socio-economic stratification or differential treatment according to skin color. The reason is simple: in the social sciences, the social problems studied are problems that refer to the norms and values of society. The ideals to be achieved are objects of the social sciences, be they solidarity or the free market. When Durkheim studied forms of solidarity and suicide, he was concerned with social integration, given the impact of industrialization on France in the late 19th century. When Weber studied the formation of the German nation, he did so in the context of the affirmation of nation-states in Europe.
Nor is remaining aloof from public debate a desirable norm. Firstly, because researchers produce knowledge and have a duty to make it public, to publish it. This is necessary to fuel scientific controversy, so that arguments can be debated. Secondly, because researchers, regardless of their personal inclinations, are called upon to give their opinions, not only by public authorities, but also by companies and civil society. Researchers publish their work in specialized journals, but if they remain silent outside this framework, their work will be little known by decision-makers and the general public, and therefore of little use.
How would you define objectivity in the social sciences?
If value neutrality is neither achievable nor desirable, then making a value commitment to a cause we believe to be right does not make us right. We can denounce the lack of solidarity, the obstacles to the free market, the inequalities between men and women, but a commitment neither a priori gives reason to the fact that these various facts are attested, nor is it equivalent to having shown the causes. Likewise, a middle-of-the-road or moderate position is no better guarantee of objectivity than a clear-cut one, even if it may be more socially or politically acceptable.
The researcher's commitment is necessary, and today it is better governed by ethical rules. Wanting to influence a country's social and economic reality, or its police policies, is in no way contrary to the researcher's objectivity, provided that a distinction is made between facts and value judgments. Some authors have argued that there is no difference between facts and judgments - a debate too long to deal with here - but I'm not convinced by their arguments: differentiating them less does not make for better science. Objectivity rests, precisely, on the ability to separate facts and values. Others have argued that the researcher's job is not to improve society. This is strange: if it is legitimate to seek a vaccine against rabies and today against Covid to avoid suffering and death, it is just as legitimate to propose new voting systems that better represent the diversity of the population and therefore cohesion, or policing doctrines to prevent the police from maiming protesters.
Objectivity means putting interpretations to the test of facts, and convincing people that the way in which this test is carried out is convincing. If Darwinism, the theory of evolutionary biology, is true, then we must be able to organize the facts observed within the framework of this theory, at least better than within that of another theory such as creationism. I think the same applies to the social sciences. If, in the field I'm familiar with, we defend the thesis that the police fulfil a function of reproduction of socio-economic inequalities, and no other, we must be able to organize the facts within this framework. If we defend the idea that the police are a neutral public service, indifferent to wealth or skin color, we must likewise be able to account for the reality we observe according to this framework. But it seems impossible to say that either interpretation is completely right or completely wrong. There are other requirements. For example, the theoretical propositions tested must be sufficiently precise (for example, what is inequality of service?), the temporal organization of variables must be demonstrated (for example, does police violence come before violence against police officers?), and causality must be proven (for example, is violence against police officers the cause of ethnic discrimination among police officers?)
How important are methods to you as a researcher?
The quality of data collection methods and reflection on tools and experimental designs to test interpretations are two concerns. For a long time, I probably focused on the first dimension, data collection, with the aim of measuring phenomena. Later on, I was more concerned with the discussion of general interpretations, i.e. whether the data could be used to decide between different theories. And, incidentally, I note that the problem in the human sciences is as much about producing theories as it is about producing data.
Given the demise of the great interpretive paradigm of Marxism, whose comet tail I only saw as a young researcher, and the fact that no other has replaced it, there are now many partial theories, but no overall theoretical framework. Contrary to what critics say about the supposed ideologization of the social sciences, and therefore of political science, the truth that best fits empirically observable facts, the trend is in the opposite direction. Sensitivities are certainly evolving, notably on the question of equal rights for individuals, with attention being paid to inequalities other than those of socio-economic status. But there are more empirical tests of theoretical propositions using different methods than in the past in the social sciences in France. The number of publications and the prominence given to the presentation of results in scientific journals bear witness to this. Objectivity has progressed, as have methods. From this point of view, it is worrying to observe that politicians may value the words of a Tourcoing butcher as a source of information on delinquency, rather than INSEE studies. Which made the director of this venerable organization cringe (and tweet).
As far as methods are concerned, I was interested in two things. Firstly, the collection of information on delinquency, and contacts with police officers. In both cases, the idea was to find reliable counting methods, and this meant breaking away from administrative statistics (delinquency recorded by the police, contacts with police officers counted by police officers). Then, to understand why police officers are more violent in France than in Germany, I looked at the systems for regulating police behavior (whether they have a constitutional court or not; whether they are organized centrally or locally, for example). While the notions of variable configuration and institutional networks seem promising for explaining differences, I don't have the answers. Methods for counting or describing institutional environments are making clear but slow progress.
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 going to take a look at the highly polemical debate on the existence of "facies" identity checks by the police. On the one hand, political leaders and police headquarters claim that the police are not "racist", in the sense of discrimination, i.e., the more unfavorable treatment of people who do not belong to the apparent majority ethnic group (physical appearances). The fact that these managers have no evidence to back up their assertion casts doubt on its veracity. However, we need to be able to test this proposition rigorously, taking into account the neighborhood where the checks are carried out, the type of behavior of the people checked and, in particular, whether they have committed an offence.
It is also necessary to conceptualize the question in order to organize the verification. For example, should "availability to police control" be considered when measuring the possibility of ethnic discrimination by the police? Does the fact that a person is on the street more often raise his or her risk of being stopped, and if this practice is more frequent in the minority group, does over-policing reflect police discrimination or the minority lifestyle? Similarly, is a police officer being discriminatory if he carries out the checks requested by the public prosecutor in a poor area where many people belonging to the ethnic minority live? Using quasi-experimental testing methods (with comparisons between groups under the same conditions), or by checking the types of identity checks carried out, we can verify that whether or not a person is available for a check, or whether or not the check is requested by a judge, we can verify that the officer's discretion, i.e. his or her ability to choose the target of the check, remains intact, and that they exercise it to the detriment of minority groups.
Empirical tests carried out by colleagues and myself over the last ten years in France have shown that people of African, sub-Saharan or North African origin are more subject to spot checks, whatever their age or gender, even if the phenomenon is more marked for young boys. Facies-based checks are also carried out both inside ZUS (sensitive urban zones, poor suburbs) and outside ZUS, particularly in the city center. The discrepancy between checks on whites and non-whites is true in poor areas, and is even greater in places where the minority is least represented, as in Paris when getting off the Thalys from London. To ensure that delinquent behavior is taken into account in the dynamics of identity checks, we have grafted modules on the experience of identity checks onto so-called "self-reported delinquency" surveys, which provide a good measure of young people's delinquent involvement. These variables were taken into account in the analyses. It appears that children, from the age of middle school onwards, are subject to over-control by the police. The POLIS and UPYC surveys in Marseille, Aix-en-Provence, Lyon and Grenoble have demonstrated this. All in all, over a ten-year period, all national and local surveys have shown ethnic discrimination in all police departments, regardless of location.
The results of studies carried out by advocacy organizations, universities, the CNRS, INED and Sciences Po Grenoble are not, however, accepted by public authorities. The rigor and repetition of the tests fail to convince. The argument put forward for rejection is a moral one: one is either with the police or against them. Observing illegal practices would therefore be immoral. It is therefore in the name of a moral commitment that some people give themselves the right to ignore objective results. This is an interesting illustration of the tension between values and objectivity.