Florent Gougoulecturer in political science at Sciences Po Grenoble and the Pacte laboratory, @FlorentGougouand Simon PersicoProfessor of Political Science at Sciences Po Grenoble and at the Pacte laboratory, @SimPersico
How would you define objectivity in the social sciences?
Objectivity in the social sciences is an ideal. This ideal is approached when the evidence used by researcherA to analyze the social world is independent of her moral and political preferences. This ideal means that a researcher B whose political preferences are opposed to those of researcher A would achieve the same results if she used the same protocol to collect and analyze the same data.
In this sense, we define objectivity as a form of intellectual honesty that we propose to evaluate according to two criteria: the systematicity of data collection and the transparency of the survey protocol. But what do these two criteria actually mean? The criterion of systematicity means that research aiming for objectivity must not dismiss facts that are not appropriate, and the second means that such research must not sweep any weaknesses in the survey system under the carpet. It is these standards of scientific work that mark out our path to objectivity.
The ideal of objectivity does not mean that social science research should be a standardized activity, using only one type of protocol. On the contrary: insofar as social facts are not strictly reproducible, objectivity demands that the social sciences be pluralistic. Faced with Researcher A's findings, Researcher B must be able to pose and test alternative hypotheses, use other investigative techniques, mobilize other data, extend the scope of the investigation in time or space. In short, it must be able to advance knowledge and possibly provoke controversy, provided it does so with the same systematic approach to data collection and the same transparency in its investigative protocol. Scientific debate has its place in the social sciences if it is based on the confrontation of robust evidence.
This conception of objectivity leads us to prefer certain epistemological approaches to others: those based on data rather than on a simple conceptual apparatus, however elegant, critical, philosophical or mathematical. From this point of view, the social sciences must, in our view, avoid two pitfalls. The first is to throw objectivity out with the bathwater of neutrality, by considering that scientific discourse, because it is socially or politically situated - which it is - can be satisfied with subjective, normative or militant analyses, based on a fragmented and/or poorly exploited field of inquiry. The second pitfall is that of privileging complex equation models over the quality of the data mobilized.
That would be a shame. Indeed, when based on relevant, clearly presented data, the social sciences produce objectively convincing and socially important elements of knowledge. Let's take the example of discrimination and prejudice against certain groups, for which we now have a wealth of evidence.
For many years, questionnaire surveys carried out on representative samples of the population have shown that Jewish, Black and Muslim people are victims of numerous negative prejudices. Field studies, whether experimental, ethnographic or based on archives and interviews, show that these prejudices have objective consequences on the lives of these people: in their dealings with the police and the authorities, as well as in their access to housing or employment. Meta-analyses that compile statistics from existing surveys confirm these results. Other studies show that these discriminations can be reinforced by the framing of these issues in public debate or legislative decisions: a recent article in a reference journal for global political science shows that the 2005 law banning the headscarf in French schools has adversely affected the educational success of Muslim girls, their trajectory on the job market and the composition of their families.
All these results obviously deserve to be replicated, debated and refined, but they already provide an objective measure of the scale and nature of the inequalities that structure our society. They cannot be dismissed out of hand, as they too often are in public debate, whether by academics or politicians.
Is researcher neutrality possible and desirable?
If objectivity is an ideal, we believe that the neutrality of female researchers is a chimera. There's no point in denying that we're both researchers working on democratic political life, and citizens with preferences about how that political life works. In fact, this is what makes the quest for objectivity all the more important: it's to guard against the risk of our preferences affecting our results that we insist on systematic data collection and a transparent survey protocol.
Every year, in one of our research methodology courses at Master's level, we have students read and discuss the first two chapters of a major work on the epistemology of the social sciences by Pierre Favre, a researcher who taught at Sciences Po for many years (and whose courses Simon was lucky enough to attend in the early 2000s). The book's manifesto title, Comprendre le monde pour le changer (Understand the world to change it), shows just how undesirable neutrality on the part of femaleresearchers can be.
Is neutrality possible when it comes to Pierre Favre's first mission: to understand the world? It's unlikely.Any scientist who embarks on a social science research project, who chooses a research subject and a research question, is guided by his or her intellectual interests, if not by his or her preferences. While notallscientists share the same relationship with their subject, behind the great names of the social sciences often lie those who are passionateabout their subjects, convincedthat the scientific method offers the appropriate tools to objectify, understand and analyze the social world they are interested in. As we've already said, the quest for objectivity is a necessary condition for a scientific method that can never be completely neutral. This is what Max Weber was talking about when he called for the axiological neutrality of scientists.
If neutrality isn't really possible, reading Pierre Favre, we understand that it isn't desirable either, since he invites us to use the knowledge produced and accumulated by the social sciences to change the world. Today, we are convinced that the results of research deserve to be communicated more widely in order to serve public debate, even if this means taking a stand. In this context, we consider that arguments based on science have a particular value: that of objectivity and refutability. From this point of view, we differentiate scientific arguments from the arguments of scientists: an academic title does not guarantee the robustness of the results mobilized.
However, the job of a social scientist cannot be confined to speaking at podiums or on TV. It implies respect for the formats and codes of the sciences, in particular evaluation and criticism by other scientists. That said, the corollary is true: publishing only in journals with a high h-index [an index that measures the dissemination of scientific work via the number of citations by other works] cannot be satisfactory. The readership of such journals is very limited, and hyperspecialization often reigns supreme.
To be of use to the greatest number of people, the social sciences must therefore leave (the comfort of) their traditional arenas of production and discussion. And once in the public arena, scientists must accept that their positions are subject to critical evaluation, from both a scientific and a political point of view. And then, their positions are invitations to respond with arguments, based on data, to advance knowledge.
How important are methods to you as a researcher?
Methods, in the broadest sense of the term, play a central role in our approach. This is reflected, first and foremost, in our concern to collect, produce and make available to the scientific community the data we need to answer the questions we pose: election results, responses to questionnaire surveys, analyses of electoral programmes, the content of legislation, the sociological and political characteristics of politicians, etc. This diversity of data reflects the diversity of the research questions that have guided us since the start of our careers. This diversity of data reflects the diversity of the research questions that have guided us since the start of our careers: are the working classes in favor of the radical right? Is ecology an issue that transcends existing cleavages? Do politicians keep their promises? What changes have taken place in the French party system? Each question requires us to find the most appropriate data to answer it, and to choose the most appropriate method for collecting and analyzing it.
Our choices in this area are guided by a number of convictions, forged in the course of our various research projects: often, it is our strategy of systematic data collection that has enabled us to come up with original results. We have also affirmed these convictions through the many methodological courses we have run, which have given rise to numerous discussions between ourselves and with our students.
These convictions revolve around a key principle that we now apply in all our work: to begin with a thorough and precise description of our objects of inquiry. We are now convinced that the primary task of the social sciences should be to describe reality as effectively as possible. Obviously, the choices we make in categorizing reality are never neutral: they must be made explicit and remain open to scientific controversy. The fact remains, however, that our mission is first and foremost to make intelligible the formidable complexity of the social world. From this point of view, we are careful not to place ourselves exclusively on the side ofexplaining political phenomena.
To this end, we favour the collection of data selected because they seem to us to be the most relevant to shed light on the phenomenon of interest to us and to test the hypotheses we have formulated - in this context, we are clearly situated in a hypothetico-deductive epistemology. To achieve this objective, we use exhaustive data whenever possible, as well as representative samples or samples that come as close as possible. Most of our data are quantitative, as they offer the advantage of being able to describe the frequency of the phenomena we observe, even if they suffer from the simplification of reality they may imply. Finally, we are adept at making comparisons, both in time and space. This has implications for the way we organize our work, which must be collective in order to draw on large, robust databases.
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?
Our respective work on the evolution of the French party system, on elections and on political ecology has led us to closely follow the recent success of French ecologists in European and municipal elections. The choice of this subject is anything but neutral for one, who readily declares himself a fellow traveler of the Greens (Simon). For the other, who has never been involved in a party (Florent), it doesn't have the same emotional dimension.
Irrespective of our personal preferences and these differentiated relationships to political ecology, we were able to measure, from the electoral data initially collected by the French Ministry of the Interior - cold and unusable data without our lengthy recoding and cleaning work - the extent of the ecologist dynamic and show that it is not confined to the big cities. We were also able to identify a number of regularities in the ecologists' victories in recent municipal elections - being able to rally other left-wing forces behind them from the first round, not facing an incumbent mayor - and thereby highlight the limits of this electoral dynamic.
These results are not without consequences when it comes to thinking about the near electoral future, whether we're thinking about the departmental and regional elections of 2021, or the presidential election of 2022. The problem, and this is another of Pierre Favre's major arguments, is that while the social sciences can explain what has happened, their ability to predict what will happen is more uncertain and limited. It is impossible, therefore, to derive from the analysis of past elections certain predictions for future elections. We have to confine ourselves to hypotheses about more or less probable scenarios. This is how we came to the idea that, in a situation where the electoral space for the left and the ecologists is profoundly weakened, a strategy that would see the ecologists go almost alone in the first round, in competition with Communist, Insoumise and/or Socialist lists, would diminish their chances of success. This applies to the departmental and regional elections, of course, but not only.
What are we to make of such an idea, based on an objective study of what has happened, the better to envisage what might happen according to this or that alliance scenario? Our answers then diverged. After having produced an analytical note on this question together, one of us (Simon) engaged in a fierce defense of the option of the broadest possible rallying of the left behind the ecologists in the partisan spheres, without ultimately succeeding, while the other preferred not to get involved in this battle (Florent).
Describe, understand, communicate, act: that's what we want our research to be used for. The same objectivity and the same methods, different forms of commitment, but in the end, the same conviction: social science findings can enlighten public debate and offer valuable arguments. As scientists, we are obliged to pass them on.
We decided to take part in this four-handed exercise, as we do almost all our research together (and teach half of our courses together). We have thus shared our similarities, starting with our desire to make the full complexity of the workings of contemporary democratic political systems intelligible to as many people as possible. We have also come across a few differences here and there, the most obvious of which concerns the way in which we disseminate our results.