Marie-Estelle BinetProfessor of Economics at Sciences Po Grenoble and the GAEL laboratory
How would you define objectivity in the social sciences?
To be objective, a teacher-researcher must conduct his or her research and teaching without making judgments based on personal preferences. They must also seek to describe the facts accurately. Research in economics has evolved considerably in recent years. At every stage of their careers, teacher-researchers must comply with very strict, pre-established evaluation rules aligned with international research practices, in order to obtain a position, promotion or funding for their research. These rules leave little room for their personal and political opinions, and this process favours, at least in this sense, the objective production of knowledge.
In most cases, and in our discipline, obtaining a position as a lecturer or university professor requires publication in the best national and international journals in the field, according to the " publish or perish" rule. These academic journals are listed and classified at French level by theHCERES and internationally by Sismago in particular. Their selection process is based on "scientific excellence". Of course, this scientific dogmatism has its limits, the first of which is the lack of plurality in the approaches and methods used. What's more, despite this drastic selection process, the articles that are published and recognized can sometimes describe the facts inaccurately and thus lack objectivity, particularly when science is wrong.
The work of renowned economists such as Reinhart and Rogoff (2010) provides a recent example. Using a sample of 44 countries over the last two centuries, they showed that countries with debt levels in excess of 90% of GDP are condemned to low growth levels of less than 1%. Later, replicating this study using the same database, Herdon et al. (2013) highlighted a series of errors, calling these initial results into question.
Secondly, by definition, knowledge is renewed through a process of creative destruction, and the very nature of research is to question and therefore sometimes challenge the results obtained by peers. Citizens wishing to gain a better understanding of reality may not only find it difficult to comprehend research work, but also to form a precise idea, given the divergent nature of the results. All in all, while academic economic research largely eliminates the subjectivity of personal opinion, it can sometimes struggle to describe reality.
Is researcher neutrality possible and desirable?
The researcher's neutrality is desirable, but difficult to achieve during the research phase, due to the hyper-specialization of economic researchers and the "tyranny of the significant result". On the other hand, in a second phase, these various productions, in the form of articles, books or the doctoral thesis, must be disseminated and valorized, and to be useful, be capable of influencing future academic research (being read, quoted and sometimes contradicted), but also the debate and public policies. In these cases, neutrality is not desirable.
Economics researchers are often highly specialized in the research methods they use (see next paragraph), and thus exploit their comparative advantages. As a result, they can be criticized for lacking neutrality in their choice of methods. However, this difficulty can often be circumvented by an appropriate choice of research topics for which their preferred methods are relevant. Moreover, in most of the academic publication media mentioned above, but also in national or European calls for projects, authors must justify the relevance of the research methods used, on pain of seeing their article or project rejected.
In applied economics, on the other hand, the "tyranny" of obtaining a statistically significant result is likely to call into question the researcher's neutrality. Suppose you obtain funding for a research project with the aim of studying the persistence of nudges to reduce energy consumption, and the underlying psychological mechanisms. In their simplest form, nudges (see Thaler and Sunstein, 2010) take the form of small visual messages (smiley face or green color to signal "low consumption" behavior and red color to signal overconsumption). To carry out this type of research, you will construct an experimental protocol and observe the energy consumption behavior of consumers subjected to these different treatments.
If your results ultimately show that consumers subject to the green nudge consume as much as those who are not, you will conclude that these nudges do not influence energy consumption and that their influence is therefore statistically insignificant. In this case, of course, you won't be able to analyze whether their effects are persistent, or study the behavioral biases at work, which the nudge could correct. Nor will you be able to publish your "non-results". As a result, you'll probably have to revise your experimental protocol in the hope of obtaining statistically significant effects this time. Is this neutrality? Yet this problem could be easily solved if non-significant results, and the search for their causes, could be published in the best journals. In addition to promoting the researcher's neutrality, this type of analysis would also be very useful (see last paragraph).
As in most other disciplines, the results of the work of economics researchers are not intended to be neutral, whether intentionally or not. One of the criteria for the evaluation and promotion of an economics researcher is his or her ability to influence other research, measured in particular by his or her impact factor, as well as public debate. Even if this is not always their initial intention, and sometimes not even their doing, the work of economists is also likely to influence public policy. The aforementioned study by Reinhart and Rogoff (2010), highlighting the negative impact of high levels of public debt (in excess of 90% of GDP) on an economy's growth rate, had a considerable impact when it was published, after the 2008 crisis and against a backdrop of runaway levels of public debt. This study undoubtedly made a major contribution to the implementation of austerity policies, particularly in Europe.
How important are methods to you as a researcher?
Quantitative methods, and in particular econometrics or the estimation of multiple linear or non-linear regression models, play a major role in my research work. Many of these studies, carried out alone or in collaboration, have led me to analyze existing databases. I have carried out a number of studies using data from the Direction Générale des Collectivités Territoriales or INSEE (the French National Institute for Statistics and Economic Studies), concerning the identification of tax competition between communes, the determinants of business start-ups within French regions , or the economic effects of intercommunality, and more generally of the superposition of territorial authorities in France. We have also analyzed macroeconomic data from 104 countries over the period 1973-2007, in order to describe post-crisis recovery patterns and their determinants, through 276 analyzed crisis episodes. In a research project currently under evaluation, we propose to work on medical and administrative data from teams of caregivers at Grenoble University Hospital, in order to model the teams in the form of a social network.
Inother studies, however, we have had to produce our own data before analyzing it. For example, we planned and carried out an island-wide survey on Reunion Island to identify the determinants of drinking water consumption and propose innovative water-saving policies. In line with this work, I am currently conducting laboratory experiments with experimental economics specialists, in which participants are subjected to various complex pricing schemes and informational price nudges. In these ongoing studies, participants (students in the laboratory or real households in the field) are confronted with different treatments, and the comparison of their consumption choices, which we collect, makes it possible to identify the nudges most effective in reducing overconsumption of water, then to analyze the persistence of their effects, and to study the underlying psychological and behavioral changes.
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?
As a teacher-researcher specializing in quantitative studies, I have naturally been confronted with the problem of insignificance mentioned above. There are two possible scenarios. Firstly, if the result is truly insignificant, in which case, if I take the example of nudge, it has no influence on consumption choices in "real life", whatever the context and socio-demographic characteristics of consumers. In this case, there's no point in conducting further research into this environmentally ineffective policy. The problem is that this result is unlikely to be published. For this and other reasons, I have given up trying to publish the "non-results" of some of my research.
Secondly, non-significance can result from the lack of precision or quality of the data used (their inability to describe reality), or from the use of biased statistical data processing methods (i.e., leading to false results, and thus leading to the erroneous conclusion that the effect is non-significant). In such cases, the researcher must instead try to identify the cause of the problem and find a solution. This is the daily task of most researchers working with quantitative data. This intervention generally leads to a deeper understanding of the methods involved, and ultimately improves the quality of the research carried out. But this iterative process is extremely costly and generally delays the dissemination of research results.