1The open access movement initially aimed at facilitating the sharing of research results, and subsequently extended to the data themselves (Arzberger et al., 2004). Beyond the simplicity of the principle according to which publicly-funded data should be available to all, organizing data sharing and access raises all kinds of questions (epistemological, methodological, technical, political). Based on experience feedback, this article addresses these questions through a retrospective study of the practices developed by the French social science qualitative data bank, beQuali, whose first years (2006-2013) were characterized by division and conflict. As direct participants in this venture, we now try to understand the extent of the tensions that marked its foundation.
- 1 For more on the project, see: http://reanalyse.hypotheses.org/projet-anr. See also Duchesne et al. (...)
- 2 We discussed the idea of making these documents available online but the impossibility of anonymizi (...)
- 3 In order to depersonalize analysis, it was agreed with the publishers of the issue to refer to the (...)
2Achieving the critical distance necessary for this analysis required the introduction of many safeguards. What mattered the most for us, as authors, was to implement our investigative approach as part of a collective consisting of our tandem, combined with the ANR Réanalyse project team, with whom we have worked in recent years on experimenting and questioning the objective of open access to social science survey data and on conducting secondary analyses of the data.1 Discussions among the team members and the results we have produced have profoundly modified our understanding of what we experienced with beQuali. The second ingredient in generating critical distance, besides drafting and revising many versions of this text, was documentary analysis. We immersed ourselves in the study of the mass of documents accumulated over the years. We provide, in the annex, a list of the most significant of them and place them at the disposal of those who would like to develop further the analysis of this case.2 Finally, we took a disciplinary detour and took the risk of stepping out of our field of competence by familiarizing ourselves with notions and concepts borrowed from the sociology of infrastructures. Our use of these notions will possibly be considered as too instrumental, and so, we let the sociologists of science decide whether, in the case we present here, a more in-depth analysis of these concepts is called for. Informed use of the latter has enabled us to expand on our first, spontaneous interpretation of the events we have experienced. We hope, in short, that our study brings some insight into the construction of an infrastructure project and the tensions that marked its early years, by going beyond a chronicle of the possible antipathies and other affects that characterize the running of a project or organization.3
3In the first section of this article, we will discuss the escalating promises affixed to the increasing number of funding applications, and the growing difficulties in ensuring their credibility, and will address a question that is central to the construction of data banks, in a context of heated international debate: that of the transition from the expression of promises to the standardization of research practices, in this case qualitative methods in social science research. The second section will provide an in-depth discussion on the ambiguity of the concept of data archiving within the beQuali team and how it is reflected in a number of intermediate objects aimed at building an infrastructure as a boundary-object, that is to say, incorporating a constitutive interpretive flexibility. The third section will analyze the organizational changes made in relation to the growing disconnection between the functioning of the infrastructure in construction and how it is used. In the final section, we will examine how beQuali’s management has been confronted with the so-called “excellence-driven” policies through which it has been financed and their consequences on how the databases were shaped.
4BeQuali (Duchesne and Garcia, 2014) emerged from an initial project aimed at developing a system for archiving interviews performed by the researchers of Cévipof, one of France’s oldest research centers in political sociology. The Cévipof researchers have played a great part in the development of French electoral sociology, under the impetus of psychologist and sociologist Guy Michelat. In addition to contributing to the development of the quantitative approach to voting, G. Michelat popularized a model for analyzing electoral behavior combining non-directive interviewing and questionnaires. Following in his wake, many Cévipof researchers carried out “campaigns of interviews” to study how voters related to politics in general and the choice of vote in particular. The idea of grouping together the transcripts of those interviews in order to analyze them diachronically was the rationale behind launching a project for the creation of a French bank of qualitative survey data, especially as the Cévipof had set up a documentation center whose manager was eager to collect and preserve the researchers’ archives, thus “safeguarding” the data from many surveys, for patrimonial purposes (Gaultier-Voituriez, 2013).
5The idea began to materialize following the symposium on secondary analysis4 organized in 2005 in Grenoble by the CIDSP and GRETS, an Electricité de France EDF laboratory. This symposium brought together protagonists from the pioneering British databank, Qualidata (Corti, 2011; Scott, 2006), and French researchers, including the organizers of Verbatim, EDF’s pilot project for the exploitation of data from interviews conducted at EDF (Dargentas and Le Roux, 2005, Le Roux, 2005). The first note relative to what was to become beQuali (dated February 2006 and addressed to the directors of Cévipof and Sciences-Po) proposes to create, in collaboration with the CDSP (socio-political Data Center),5 a qualitative survey data bank of CEVIPOF. The intention was to extend it, if successful, to the other Sciences-Po laboratories. This note suggests setting up an archiving committee which will decide which surveys should be integrated into the database, and what procedures should be implemented for the processing of documents (inventory procedure, methods of anonymization, digitization and the embryonic phase of metadata development) and for their dissemination, particularly to Master’s students.
6Between 2007 and 2009, two series of events occurred that were revealing of changes in HSS research. Those events were to prevent the implementation of this initial project for the exploitation of the surveys produced by a “historical” laboratory and to contribute to its transformation into a national database project. Firstly, a schism occurred within the Cévipof following the renewal of its director6 and resulted in the departure of the project’s initiator. Secondly, the new scientific director of Sciences-Po launched a major operation for the bottom-up development of priority lines of research for the institution. He asked researchers to submit proposals for inter-laboratory projects to be evaluated for funding by an international scientific advisory board (SAB); this constituted a break from the framework whereby research projects were financed by research centers. The opportunity was thus taken to propose the construction of the “DIM” framework (Data, infrastructures, methods) within which the qualitative survey data bank would be developed. Thus, the DIM framework was created and the BDQuali project was selected. The advisory board was pleased to see Sciences-Po following in the footsteps of Qualidata. However, the board noted that, unlike Qualidata, the data bank, would not be able to force the authors of surveys to archive them and would, therefore, have to find other ways of encouraging them to do so.7 The funding provided for the project made it possible to employ a part-time research officer (starting in September 2009) whose task was to finalize a feasibility study,8 to build a network of Sciences-Po researchers interested in the project, and to find additional funding.
7The first funding was provided through a contract with the National Research Agency (ANR).9 Entitled “Réanalyse”, this project consisted of a series of experiments in secondary analyzes in order to evaluate their potential and limitations and to be in a position to determine the appropriate conditions for disseminating survey data. In reference to Qualidata, the project has focused on the need to avoid any de-contextualization of data. Two systems were envisaged. The first, developed in collaboration with the recently created Medialab of Sciences-Po,10 consisted of storing the surveys not only as a series of indexed files but also, according to a “holistic” archiving approach, by taking advantage of web technology and creating “survey sites” to facilitate users’ exploration of all documents. The aim was to link together, in user interfaces, all the information pertaining to each survey, namely the “data” and documents relating to the context in which they were produced. The second system consisted of designing a “survey of surveys”, combining analysis of publications, of the documents pertaining to the survey and of the interviews with the primary researchers. A collective of 12 researchers, including the organizers of the 2005 symposium, was formed for this purpose.
- 11 The Qualidata experience had proved the difficulty of involving researchers in the operation; It se (...)
8While the project was under evaluation, the French government launched, via the ANR, an invitation to tender for “Equipex –equipment of excellence”, in the framework of the Programme of Investments for the Future (PIA). A joint proposal by the CDSP, Medialab, and beQuali was submitted, combining three equipment projects: a University polling institute in the form of an Internet panel, Medialab and a bank of qualitative survey data. The central idea was to use the possibilities offered by the digital revolution to help France to catch up with other countries in terms of reflections and publications regarding the tools and practices for creating and analyzing data. The Equipex project, entitled DIME-SHS, was to be coordinated by the Director of the CDSP, since he was the chief scientific officer of the Internet panel, which was the most extensive infrastructure of the three. It was also the infrastructure that offered the best self-financing potential. Indeed, an Equipex project is granted funding for a period of 10 years, after which it must be able to generate the resources necessary to sustain itself.11
9Both the ANR and Equipex project proposals were selected a few months apart and the two projects were, therefore, launched in parallel.
- 12 In this article, beQuali refers to the whole project of construction of the data bank and the peopl (...)
10The various project proposals were accompanied by mounting promises which the projected equipment had to fulfill in order to urgently remedy a situation (Audetat, 2015, Joly, 2015); this phenomenon of performative anticipation has now been well examined in the literature, including in social science (Vinck, 2015). The transition from the Cévipof’s modest documentary center project to that of a national infrastructure, soon to be called beQuali12 led to an escalation of promises of a radical break from what was presented as the French lag in methodological reflection, particularly on qualitative methods. BeQuali, like Qualidata, would have to make it possible to “safeguard” the surveys, use them more intensively, allow for qualitative, and in particular diachronic, comparisons, and help to better understand the research-making process. But it would also have to improve analytical methods, particularly by facilitating the exploration and reading of documents and by developing new tools. Moreover, it would have to contribute to the reflection that was developing in the United States on active citation (Moravcsik, 2010). Finally, it should be the ambition of beQuali to reconcile the debate initiated by Qualidata by finding ways of archiving surveys without decontextualizing the data.
11An escalation of promises always involves a difficulty in ensuring their credibility, or even their coherence. In this case, the context created by the developments in the international debate around questions of digital infrastructure and digital humanities would also challenge those promises. Indeed, Qualidata elicited strong and well-argued reactions (Duchesne, 2017, Gillies and Edwards, 2005, Mauthner and Parry, 2009, 2013, Mauthner, Parry and Backett-Milburn, 1998, Moore, 2005, 2004, Savage, 2005). Some of them rejected the approach to secondary analysis which the British archive sought to disseminate because it promoted research practices based on an epistemology which was, if not actually positivist, at least borrowed from the “good practices” and validity criteria set by studies in social science that only considered survey data. The directive requiring researchers to entrust their surveys to the bank and encouraging them to use the data stored in the databank has come to be interpreted as a result of the neo-managerial policies imposed on research and social sciences. Furthermore, the notion of active citation –which requires that scholars make the data they have used for their research accessible to the readers at the time of publication of their works– was introduced in order to facilitate the validation or rejection of the analyses published (Moravcsik, 2010). This introduction gave rise to a debate on transparency and the need to allow for the replication of the research results (Moravcsik, 2014). The debate was accompanied by the creation of repositories in which researchers were invited to deposit their data. It culminated in 2015 with the adoption of a joint declaration by some 30 international political science journals, in which they committed to requiring that the authors who submitted articles certify that the data analyzed were actually available13. This declaration, in turn, prompted an outcry from American qualitativists, who warned of the effects of scientific standardization which the application of such a commitment would produce14.
- 15 The Archipolis consortium, originally called Archives of the Social Sciences of Politics, was added (...)
12In other words, the de-contextualization resulting from the archiving of survey data no longer appeared as the main risk attached to the implementation of a data banking facility: it was the archival process itself that seemed to entail significant risks of standardization of qualitative survey data. This change in context was all the more significant for the beQuali project that the information campaign conducted among researchers showed that they were less interested in the possibility of working on surveys performed by others than in the fact that they could make their own data accessible at the time of their works’ publication. To fulfill the need to make surveys available and to significantly increase the capacity and speed of data accessibility, self-archiving seemed to be the only solution. Thus, the beQuali project, which involved a third consortium set up to facilitate the collection of the survey data to be archived (Archipolis), initiated a reflection on the subject.15 But in the context of growing controversy over the effects of transparency, self-archiving appeared increasingly risky, in that the promise to avoid standardization seemed less and less credible. This led to growing tension within the team, torn between the will to honor this promise and a plethora of standards and formats it was elaborating in order to address the important technical challenges, poorly anticipated by the creators of the data bank. Pressure grew within the project’s team around the importance of archiving in all dimensions of the project, a tension we shall discuss in the next section.
13While the promises made by beQuali were largely developed by social scientists, the infrastructure itself was constructed by actors operating at the frontier between different professional spheres: researchers but also archivists, documentalists, computer scientists, and administrators. The concept of boundary-object was developed to help study those situations of collaboration between different worlds –particularly in the context of data bank projects (Meyer, 2009)– on the basis of observation of the actors’ practices. This concept is relevant to our analysis. To avoid an anecdotal use of the concept (Trompette and Vinck, 2009), we shall explore the primary dimension of interpretive flexibility that characterizes the concept of boundary-object (Vinck, 2009).
14Indeed, beQuali is subject to different interpretations by the actors linked together through their involvement in the project, particularly as regards the very notion of archiving and the practices associated with it. The team’s difficulty in agreeing on compatible interpretations, in a context of intense controversy, became clear during the construction of a whole series of intermediate, often digital, objects (filing plan, inventories, metadata, Website, graphic charter, copyright assignment and usage contracts, specifications for IT development, etc.). These intermediate objects were intended to both help to implement the project actions and to enhance convergence and compatibility between the various interpretations of the team members.
- 16 As an indication, for one of the two authors of this article, the computer-based record correspondi (...)
15Like its model, Qualidata, beQuali is a project that aims at the preservation and dissemination of survey data, in particular, to contribute to and provide insight into the history of social science and the teaching of research methods, and to make them available for reuse. The construction of the infrastructure first required that the documents pertaining to surveys be archived. But in the case of Qualidata, the archiving process was intended to preserve survey data in their original form, whereas in beQuali, the objective was to preserve them in a format that would allow for new treatments (Both and Cadorel, 2015). The “polyphony” generated by this ambivalence was brought to light by the team members in an article (Both and Garcia, 2012), the drafting of which was accompanied by intense negotiations, in particular, because the article put the visibility of the project on the line. In fact, the practical effects of this ambivalence on the archiving system that beQuali aimed to develop quickly became the daily concern of the team members, who, during the first three years of the project, worked on developing a plethora of documents16 –those describing the intermediate objects particularly. All these documents were intended to define procedures for managing the process of archiving and disseminating the survey data. The ambivalence regarding the function of the archiving tool weighed heavily on the development of those documents. We shall examine a single example here; that which the team called the “documentary framework” of the survey, in other words, the more or less formalized procedure that governed the selection and/or retrieval of the documents to be archived for each survey.
16In order to build the data bank prototype, its founder, on one hand, convinced two colleagues to entrust him with the documents pertaining with their survey, and on the other hand, made available the data relating to a survey he had previously coordinated and which the first of researcher officers recruited by beQuali had participated in. The archiving of this survey’s data and documents, carried out by the second research officer recruited permanently soon after he had joined the team, prompted a vigorous dispute over which documents should be selected for archiving. The research officer selected a large number of documents, including documents relating to the organization of meetings and trips, drafts, etc. For the coordinator of the survey, this selection of documents –drawn exclusively from his own archives– gave a falsely comprehensive view of what the research work had been. Above all, it far exceeded the amount of data which secondary researchers would take the time to consult and complicate even more access to the important documents.
17In the fall of 2013, the team members gathered for a two-day meeting in order to take stock of the difficulties they were facing with the selection process. In an attempt to rationalize the choice of documents and to depersonalize the debate, the team tried, among other things, to define what the “documentary framework” of a survey should be. The minutes of the meeting specify that the process of collection and dissemination of the documents is not intended to be all-encompassing: it aims to select what will enable other researchers to use the documents provided in a valid manner. The founder of beQuali, and author of the minutes, added: “The researcher(s) will have to be consulted once the survey has been filed, preferably before being digitized, and in all events, before the documents can be made available on the repository of documents we propose to put into service”. But the central questions of what constitutes a survey and of the part played by the researcher-author in its definition were ultimately not considered in the practices subsequently displayed by the team.17
18Many intermediate objects were the bearers of the tensions within the team, particularly that concerning beQuali’s objectives in terms of archiving. They should have enabled the team to better manage the different purposes attributed to archiving, that is to say to reconcile the patrimonial and the instrumental, the ends and the means. The table below provides a description of the main ones.
Table 1. Synthetic presentation of the intermediary objects produced by the beQuali team that put the interpretative flexibility pertaining to the archiving process at play
Intermediate objects
|
Objectives
|
Requirement for its existence
|
Tensions at play
|
Inventory grid
|
To organize the collection of the information pertaining to surveys which can potentially be archived in the bank
|
To produce data that will make it possible to create an inventory, i.e. easily accumulable data
|
Over the specification and standardization of the data required by the permanent members: How much leeway should be given to the research officers (ranging from students/doctoral and post doc students to documentalists) in engaging with the authors of the surveys in a dialogue regarding their survey? And therefore, what place is there for the categories used by the primary researchers?
|
“Survey on the Survey” (EsE)18
|
To prevent the risk of de- contextualization of the data
|
Offering secondary researchersa way to understand the archived survey
|
Is the aim to contextualize the survey in order to avoid false interpretations, or is it provide the person who is going to re-analyse the survey with information that would otherwise be inaccessible – in particular the intentions and impressions of the researcher- author? Doesn’t the large document produced for each of the three surveys interfere with the work of the person who is going to reanalyze them? Worse, does it not tend to freeze the meaning of the survey?19
|
Filing plan, document naming principles and a so called analytical inventory model
|
Managingthe files corresponding to the documents collected from the researchers, and filing them for archiving
|
Enable the users – the secondary researchers – to navigate the documents more easily
|
Are these three objects intended for archivists and documentalists or for researchers? Should they take into account the plethora of reasonings – or even the disorder – that researchers apply when conducting an survey?
|
BeQuali Website
|
To ensure visibility of the project and serve as an interface for accessing survey data
|
To draw researchers’ interest, faciliate contact with the team and access to the surveys
|
A wide gap between the principles stated on the site – a holistic approach to the survey, making the boundary between data and context more flexible, among other things – and what the tool actually allows: separate consultation of the list of documents and of transcripts of interviews, for example
|
Metadata
|
To articulate the grids used by international social science, libraries and archive services (DDI 2.2, EAD, DC, METS)20
|
To allow for automatic equivalence between the different systems and to minimize the need to re- enter data
|
Ambitious attempt by the team to develop – without including the researchers in this reflection – and sustain a system of equivalence between the pre-existing grids, instead of conducting a joint reflection on the categories that are relevant to its project21
|
- 22 Great care is taken to develop and apply a graphic chart, to ensure that those objects are made vis (...)
19We could go on with this long list of objects produced by the beQuali team during its first two years of existence. The proliferation of objects and the intensity of this production coincided with an increasing division of labor between the members of the team. In addition, the more the permanent staff endeavored to specify and refine these objects,22 the wider the gap became between the view of archiving as an end in itself and the view of archiving as a means to an end. In other words, the work conducted to build this collection of objects did not enable the team to develop beQuali as a boundary-object. It merely led to a reversal of priorities between a heritage-centered objective to a research-centered one (see Box). This was accompanied, as we shall see in the following section, by a change in the nature of the team and in the statutes of its members.
Table 2. Changes in the missions assigned to the infrastructure, according to beQuali’s website
BeQuali’s objectives in January 2013
|
Contribute to generating new forms of scientificity of qualitative surveys in social science by making the progression from data collection to analysis visible and therefore debatable; Promote secondary analysis and thus open up possibilities for new comparisons. Develop methodological reflection. Provide methodological teaching materials; Safeguarding heritage surveys
|
BeQuali’s objectives in January 2016
|
BeQuali offers a catalogue of surveys in social science and humanities, based on qualitative methods (observations, interviews, focus groups, etc.). This portal provides the scientific community, with access (subject to prior authorization) to raw data and a corpus of documents providing information on the context in which the data was produced. BeQuali is an initiative for the preservation of research data and provides access to common resources and tools for the teaching of methods and the production of new research knowledge
|
- 23 This did not happen by chance, even though it may have been experienced as such by the protagonists (...)
- 24 P. Flichy defines the frame of use “as that which defines the type of social activities proposed by (...)
20In order to fully understand why the question of the relation between the functioning and the usage of this type of infrastructure is raised, it is important to bear in mind that Qualidata was developed without having first experimented with secondary analysis practices. In this regard, the absence of any assessment of the uses made of the data provided by Qualidata was regularly highlighted by critics (Parry and Mauthner, 2005). It is striking that one of the most frequently cited texts on the practice of secondary analysis was written by one of the people at the head of the data bank, to respond to the users’ demands (Bishop, 2007). BeQuali, on the contrary, was founded with the aim to first experiment with reanalysis; but issues and uncertainties related to funding changed the priorities.23 In other words, the founders of beQuali wanted, as Patrice Flichy explained, “to build a frame of functioning of the tool based on how the users will use it”.24
21Thus, this section attempts to account for the specific moment when the implementation of the intermediate objects under construction should have been accompanied by utilization of the data; this would have helped to create links between designers and researchers, between the frame of functioning and the frame of use. This period was characterized by three mutually reinforcing dynamics: a change in the activity of beQuali’s team members, who gradually ceased to do social science; the end of the collaboration with Medialab; a distanciation from the potential users.
- 25 The IDEX –“Initiatives of Excellence”– are groups of tertiary education and research institutions t (...)
- 26 As indicated in the description of the team on the beQuali website: “Five research assistants, supp (...)
22BeQuali was created by a CNRS researcher, assisted by a young doctor in political science who, after completing several fixed-term contracts working on the project, was hired permanently by Sciences-Po in the framework of the Equipex contract. Convinced that the tool should be designed primarily by researchers to ensure that it is well-suited to their practices, they recruited another doctor –this time an anthropologist– who had extensively investigated in the field of archives. A few months later, having gauged the extent of the technical challenges facing them –both in terms of IT and archiving– they chose to supplement the team by recruiting an archivist trained in digital humanities. A computer scientist was subsequently recruited on a fixed-term basis as part of the permanent team. In parallel with the process of production of intermediate objects, the amount of time dedicated to research by the two doctors –who happened to be hired under the title of chargé d’étude or research officer– gradually decreased. The first recruited research officer underwent a training course in project management and was appointed “operational” manager of the team, leaving to the researcher the role of scientific coordinator. He ended his involvement in the Réanalyse project. The second research officer, who showed relatively little scientific interest in the first surveys archived by the bank or in the activities undertaken by his colleagues in Réanalyse, soon manifested more interest in archival issues than in those related to the re-use of these surveys. He invested himself in the development of Sciences-Po’s archive project and participated, as much as he could, in archivists conferences. He too ultimately contributed little to Réanalyse. Two post-doc students were subsequently recruited thanks to a fourth funding round, obtained following a call for tender from the IDEX25 in which Sciences-Po participated: From then on, the CDSP no longer relied on their own staff to experiment with secondary data use, but on these two new temporary recruits.26
23These developments were reinforced by the confirmation of the team’ status as a service unit. At the time of Equipex’ inception its coordinator, who was also in charge of the unit within which it was created –the CDSP– considered re-building the team around a methodological research project. Together with the beQuali’s founder, they jointly submitted a contribution to the “Higher Education and Research Conference”,27 raising the question of the consequences of project-based funding on the development of research infrastructures and re-affirming the importance of developing, in France, research on the questions pertaining to research methods. At the time of the CDSP evaluation by the AERES,28 its director even asked for the status of the unit to be changed from that of “Mixed service unit” to that of”Service and Research Unit”. This would have enabled the team to, among other things, employ researchers full time,29 to better exploit the research results obtained by members of the team and in particular, for beQuali, the work carried out within the framework of Réanalyse. Despite the favorable opinion of the evaluation committee, the umbrella institutions –the CNRS and Sciences-Po– decided otherwise. This led the Director of the CDSP to request that the full-time members of beQuali dedicate themselves fully to the archiving of the survey data.
24Medialab played a large part in the first reflections on the construction of the infrastructure, using the “survey-sites” interface, which was supposed to facilitate the shift away from the traditional approach to archiving and to develop original research tools. Sciences-Po’s Medialab had recently been created and, as it was, had been a full partner in the first Réanalyse funding round. A computer engineer was hired for one year under this contract to build the infrastructure for the dissemination of survey data. But what he developed was finally quite different from the specifications set for the original project: the idea of a survey-sites interface was abandoned and would not be incorporated into the infrastructure. The members of beQuali only became aware of this at the end of his contract.
25Meanwhile, Equipex was financed, and Medialab was part of the overall project. Its team soon had their hands full working on this project as well as all those they submitted at the same time and for which they obtained funding from several sources and for different projects. Consequently, beQuali had to make do without the expertise of Medialab’ engineers and had to rely on CDSP’ small team of computer engineers, who were already well occupied with the Internet panel project. A young computer scientist was recruited under a fixed-term contract but he was still in training. The experimental dimension of the tool was consequently eliminated with the end of collaboration with computer engineers.
26The implementation of the objects designed for archiving and making survey data available was accompanied by procedures that tended to keep potential users at a distance. In particular, the coordinator of the Equipex decided that the qualitative surveys would be made accessible via the “French portal of access to data for humanities and social science” –Quételet portal30– which disseminates the quantitative surveys archived by the CDSP. This forced the beQuali team to design a complicated system for depositing and accessing survey data, since the procedures in force on the Quételet portal, developed for making very large surveys available (those of public statistics institutions in particular) were highly formalized. Susan Leigh Star has highlighted the fact that a developing infrastructure is strongly influenced by the pre-existing infrastructures it is intended to replace: it “wrestles with the inertia of the installed base and inherits its strengths” (Star, 2010, p. 29). In this case, beQuali’s “catalog” consisted of only three titles, and everything still had to be done to find ways of convincing colleagues to take an interest in it. The complexity of the access procedure that had to be invented did nothing to simplify matters.
- 31 To prevent a researcher, for example, from changing his mind and withdrawing his survey once the te (...)
27The formalization of the procedures of submission and use of the surveys’ data did not stop there. Using the services of the lawyers’ firm that the quantitative team of DIME-SHS had had to hire, the DIME-Quali team had a number of contracts drawn to protect their work31 and to release themselves (or rather the umbrella institution, Sciences-Po) from any liability in the event of a misuse of the data. In this case, again, the system became complicated in that some of these documents would have to be countersigned by Sciences-Po’ scientific director.
- 32 For example, while putting the surveys online is intended to make them accessible more quickly and (...)
28The formalization of relations between the researchers (whether they be the authors or users of the surveys) and the team created a distance between both parties; a distance that grew wider over time. From the outset, the team was careful to avoid a process of invisibilization of the technical work of construction, data input, and maintenance of the infrastructure; a process described by Dagiral and Peerbaye (2012) and Flichy and Parasie (2013) in their study of similar cases. In reality, however, there developed, if not an invisibilization, then at least a form of domestication (in the sense meant by Callon, 1986) of researchers-authors, who were relegated to the status of users and reduced to silence by the interface.32 This was something of a paradox considering the infrastructure was intended to enhance the exploitation of qualitative research results. In this respect, it should be noted that the anonymization of the surveys –which had initially been put on hold– became a priority project for the team immediately after the departure of the data bank’s founder. For the latter, anonymisation negatively impacted the integrity of the surveys and limited the re-use of the data, and was unnecessary provided they were re-analyzed by researchers: mutual trust between colleagues (the authors of the surveys and their users) and a shared professional ethic ensure the protection of the respondents. Conversely, the bank tended to insert itself between researchers: the steward of trust, it is now the bank that guarantees the value of the data. This change is currently taking place in Qualidata, which has set up evaluation committees to assess self-archived surveys (Corti and Eynden, 2015).
29Two years after the launch of DIME-SHS, the number of norms and procedures governing the archiving of the surveys seemed to increase irreversibly. In the meantime, experimental work continued in Réanalyse but became totally disconnected from DIME-Quali; and the researchers working on the first archived surveys made no use of the online exploration tools, which were far inferior to other tools they had access to. The infrastructure’s user framework, which in the project, was to be defined by engaging the participation and responsibility of designer-users (Flichy, 2008), ceased to be the focus of the team, now entirely devoted to the operating framework of archiving. The team members diverged on a central question: Was DIME-Quali the alpha version of an instrument that should be tested without further delay to verify that it corresponded to the promises that accompanied the project; Or was it already the instrument that the team should now “stock” with surveys, while waiting to know who will want to use it? The last section of the text explains how the question was resolved.
30BeQuali is supported by the institutions involved in the definition of the scientific policy of so-called “excellence” implemented by the Programme of Investments for the Future: TGE Adonis / TGIR Huma-Num, SAB Sciences-Po, ANR, IDEX, as well as the contributions of the experts employed by Sciences-Po and the CNRS. Its scientific project responded to many of this policy’s objectives, including the construction of a major infrastructure for social science in line with the requirements for internationalization, for better dissemination and exploitation of research results and research equipment. This political orientation is in line with a European, if not global trend (Heraud and Lachmann, 2015, Ravinet, 2011), it is financed on a project basis and is dependant on heterogeneous resources the management of which is complex, to say the least: temporary and limited-term contracts; ex-ante project evaluation and administrative monitoring of goal achievements based on indicators and/or “deliverables”; Centralized coordination and organization of competition between project bearers and their institutions.
- 33 This would also have helped to simplify certain procedures, including the signing of each deposit o (...)
31The designer of beQuali sought means of countering some of the effects of how the project was financed. He constructed it as part of a network of projects –Réanalyse, DIME-Quali and Archipolis– comprising participants with different statuses and from various institutions, a network he headed even though he was not affiliated with the institution that contributed the largest share of the funding. The growing number of contracts entered into aimed not only at accumulating resources but also at giving the project a national dimension and at ensuring that it would not become an infrastructure exclusively at the service of Sciences-Po. The founder several times considered creating an association, which would give the infrastructure a legal status33 while reinforcing the cohesion of the project. This proposal was, each time, rejected by the CDSP’s Director and Equipex’ coordinator.
- 34 For readers unfamiliar with French social sciences, Sciences-Po has long embodied a dominant and po (...)
32And, as months went by, the project’s ties to Sciences-Po grew tighter. First, more and more of the laboratories operating under the supervision of the National Foundation of Political Science joined the Archipolis consortium. Then, two of beQuali’s members became involved in the development of Sciences-Po’s new “Archive Mission”. Above all its library, which was intended to develop into a data library (Cavalier, 2015), turned to the CDSP for support. The DIME-Quali team, in search of additional resources, and in particular, of permanent staff, agreed to discuss, with the library team, the possibility of developing a version of beQuali dedicated to the archiving of surveys conducted by Sciences-Po researchers. It was, incidentally, during the first meeting aimed at setting up this collaboration, that the CDSP’s director publicly presented himself as the head of beQuali for the first time, although its founder and scientific animator was attending the meeting. BeQuali’s strong ties to Sciences-Po accentuated the tensions regarding the risks of standardization because the surveys conducted in this institution’s laboratories were far from being representative of the variety of qualitative social science surveys concerning political objects.34 However, the prototype was developed on the basis of this type of surveys. It was originally intended to be modified to enable it to integrate other types. Failing which, to characterize it as a “national” infrastructure is tantamount to imposing the formats developed to archive these surveys on other fields of research.
- 35 Let us note that this raises the question of intellectual property in this type of project.
33In the annual report to be submitted to the National Research Agency, the team had to indicate the number of surveys archived. After two years of work, only three surveys were listed in the catalog. According to the Equipex’ coordinator, who reported to the ANR on the state of advancement of the project, those inadequate “results” were due to the fact that the team spent an excessive amount of time reflecting on the uses of the data bank and experimenting with reanalysis. The founder of beQuali –who had no administrative authority over the people involved in the project– believed, on the contrary, that those results indicated a need to rethink and rebuild the instrument, which, besides causing him concerns in terms of scientific standardization, caused the archiving process to require too many manual interventions. Sciences-Po’ scientific director then attempted to reach a compromise: the authority of the CDSP’ Director and Equipex’ coordinator would be confirmed, but meetings would be held on a regular basis, enabling the founder of beQuali to continue giving his opinion on the strategy to be followed. Finding that it would be impossible to reorient beQuali, the latter refused the proposition and left the project arguing that, as it stood, beQuali did not fulfill the purpose for which he created it. A few members of DIME-Quali’s Scientific and Technical Committee also resigned, including its chairman; resignations which, however, received little echo within the Equipex steering committee.35 On the contrary, a form of institutional escalation occurred (Ross and Staw, 1993: 724): at a time when the project seemed to be failing, the relevant institutions invested further resources and two more people were recruited into the team in the months following the split.
34Yet, two years later, only six surveys were listed in beQuali’s catalog. Moreover, they had all been produced in the laboratories of Sciences-Po and were far from representing the diversity of qualitative data which beQuali was intended to provide access to. The project-based financing mechanism, in which teams and staff were made to compete with one another, the institutions’ lack of openness to the outside, with indicator-based evaluation and the transfer of scientific authority to the administration, seem to have largely contributed to the project becoming a Science Po instrument designed to embody and promote the practices of an institution of excellence.
35Written by the founding members of the French qualitative survey data bank –beQuali– this article has given an account of and attempted to make sense of the tensions that accompanied its creation, between 2006 and 2013. Four angles of approach were chosen, inspired by studies in the sociology of science dealing, in particular, with the process of safeguarding data into a data bank. The first approach consisted of relating the genesis of the promises that accompanied the creation of a “Qualidata à la française”, a project which was very –if not too– ambitious. BeQuali was inspired by the British model (Qualidata) –which was itself strongly invested in the promotion of “best practices” in the field of data collection, management and analysis– and was nevertheless supposed to prevent the risks of standardization associated with the conversion of survey responses into data, and, on the contrary, to contribute to diversity in qualitative research. The incompatible promises on which the project rested constituted a breeding ground for misunderstandings and tensions within the team. We then focused on the process of construction of the infrastructure and on a large number of objects developed by the team. These objects were intended to help deliver those promises, and as such should have embodied the different conceptions of the project: conservational in its archiving function and instrumental in its function of dissemination of surveys, in particular. Yet we found, on the contrary, that as a result of the team’s work overload, these objects were appropriated separately and facilitated neither the expression nor the articulation of these different conceptions of the infrastructure. The third section of the article describes the changes that occurred in the team in terms of professional positioning (i.e. in the statutes or activities of the team members). We have shown that these transformations were as rapid as they were extensive and contributed to keeping researchers at a distance; the latter were increasingly seen as users, and the design of the operating framework did not take into account how they used the tool. Within the team, this development resulted in growing dissension between the employees with the status of researcher and the others, with regard in particular to the sustainability of the infrastructure or the pursuit of experimentation. In order to understand how this issue was resolved, the fourth and final section of the article examined the governance structure in beQuali –a project of “excellence”– and the disconnection between scientific responsibility on the one hand, and administrative and financial authority, on the other. As the tensions grew within the team, the crisis resulted in the departure of beQuali ‘s founder and most of the fellow researchers.
36BeQuali was indeed created and continues to operate, but its survival has necessitated a re-definition of its objectives. Above all, the bank is struggling to archive more than one survey per year and so far, the uses made of it have been limited. Does comparing beQuali to Qualidata, whose success in terms of catalog and influence is undoubted, help to better understand this failure, however relative it might be? Can it be considered that this failure is due to beQuali being positioned as part of a so-called “excellence” mechanism characterized by competition, narrow institutional boundaries, project-based financing and short-term funding? The British team clearly did not burden itself with so many contradictory promises: without any difficulty, it has been able to deal with the effects of standardization resulting from a centralized archiving system and supported by a powerful national incentive mechanism. Criticisms came from outside the team and their effects were partially mitigated through very effective networking activities. However, these criticisms continue and fuel extensive debate about the construction of databases and how they are used for research in humanities and social science. Referring to the necessity of maximizing the returns on public investments in surveys, particularly qualitative ones, those criticisms emphasize the low usage made of archived surveys and the high cost of these infrastructures. They highlight the discrepancy between the stated objectives of improving transparency and enhancing good practices on the one hand, and the complex reality of the process that leads to the discovery and permanent transformation of research standards and practices, which, should they be fixed, would have a long-term and negative impact on scientific activity. In other words, while the so-called policies of “excellence”, which beQuali is part of, probably explain the intense tensions that this project has experienced, its relative failure may be more broadly related to the archiving itself and to the fact that this type of instrument possesses political properties which must be accounted for. To support this conclusion it is necessary to explore more closely the data banks that have succeeded unquestionably and to go beyond the mere celebration of their success (Bowker, 2000). For the sake of symmetry with our account of the history of beQuali, we believe, like Geiger, Moore, and Savage, that a similar study of the genesis of Qualidata should be undertaken (Geiger, Moore and Savage, 2010).
The authors are particularly grateful to the publishers of this journal’s issue for their reviews of the various versions of this article and for their many suggestions; Claire Dupuy, Camille Hamidi, Alexandre Jaunait, Thibaut Rioufreyt and Virginie Van Ingelgom for their support, as well as Magali Vautelin for her help in formatting the text. Many thank also to Delphine Silberbauer who translated this paper.