Abstract
The increasing development of research mobilizing “mixed” quantitative and qualitative methods is associated with the search for more permeability and innovative paths between these approaches. In this double thematic dossier of the BMS, we wish to continue in this direction of decompartmentalizing methods by moving their boundaries of application. Exploratory analyses are using methods of data extraction and processing which “transgress” the traditional correspondence between materials and methods of treatment. The six articles of this double dossier to be published in issues 124 and 125 of the BMS offer original designs combining these qualitative and quantitative dimensions. They are based upon surveys focusing on temporal processes and life courses. They seek to open up new avenues for research methods, without undermining the validity of existing standard methods.
Linking Quantitative and Qualitative Methods to Study Temporal Processes and Individual Trajectories
Since the beginning of social sciences, qualitative and quantitative analyses sometimes confront, sometimes complement each other in a variety of ways. Their respective positions and their roles in the administration of evidence have evolved greatly and their boundaries of application seem now to be increasingly porous. More and more young researchers are competent in both methodological dimensions, and “mixed methods” have developed and been disseminated in a manner to articulate and clarify more systematically the respective contributions of qualitative and quantitative methods at different times and stages of a research.
Calls for relaxing of barriers between these methods are not new. The BMS in the last thirty years has published articles defending the idea that qualitative and quantitative methods are complementary and interdependent rather than conflicting, and that researchers should look for practical consistency between them rather than applying strictly “the right” procedure and reinforcing their compartmentalization (Wilson, 1986; van Meter, 1994).
“Mixed” empirical research has increasingly expanded, along with methodological and theoretical clarifications advocating greater permeability between these dimensions (Sale et al., 2002; Maxwell and Loomis, 2003; Hall and Howard, 2008). Research designs are more and more closely integrating qualitative and quantitative perspectives, sometimes by different analytical approaches applied to the same data set. They show the added value of these combinations which provide a more comprehensive view of the phenomenon under study, and also raise new analytical dimensions.
More recently, proposals have emerged which also tend to overcome the sequential use of these methods so as to reach greater synergy between types of data and analysis, in particular by examining qualitative data with quantitative methods or vice versa.
Nowadays, demographers are interested in life histories, textual analysis resorts to modeling methods, interviews are recoded and submitted to statistical measures, questionnaires are reformulated as narrations, etc. These various forms of crossover analyses are currently being developed and have shown to be very creative and promising (Small, 2011).
These openings and movement between methods appeal for the clarification and the relativisation of some conceptions and oppositions too often “automatically” associated with the quantitative/qualitative division: objective vs. subjective data, positivist vs. interpretative sociology, large vs. small sample size, statistical validity vs. comprehensive approach, discrete variables vs. contextual configurations, mathematical formalization vs. thematic association, representativeness or monographic coherence, including the debates on forms of causality.
The six contributions to this methodological exploration will appear in the number 124 and the number 125 issues of the BMS and a French version of this Introduction and perhaps other elements will be made available on the BMS-RC33 email list (https://listes.services.cnrs.fr/wws/arc/bms-rc33) and the BMS Blog (http://karlvanmeter.wordpress.com/).
Exploratory Propositions for the Transgression of Boundaries
In this double dossier of the BMS, we wish to further reduce the tension of division by “playing” with the boundaries between these methods. The articles presented here have in common the fact that they offer explorations of original data by a sort of “back-and-forth” between qualitative and quantitative methods, or by adopting an “in-between” position borrowing from both at the same time.
These approaches in fact use procedures for collection and analysis that are a priori not directly fitted to the data under study. They “transgress” in a certain way the usual correspondence between the material and the method employed. They sometimes lead to “quantifying” qualitative information (collected through in-depth biographical interviews) by appropriate coding and analysis operations, and sometimes result in extracting comprehensive life stories from quantitative data (collected with a large-scale questionnaire survey).
The aim of this double dossier is therefore twofold. It is of course to move towards broadening the diversity and validity of the methods by going beyond the routines and orthodox practices. But these explorations are not only for a playful or provocative purpose. In opening new avenues, they provide original insights concerning the topics we are interested in, and enable us to extend or refine our results and interpretations. We can then discuss their precise conditions of validity and, in a modest and cautious way, pursue the new paths that have been opened. Indeed, the aim is neither to draw radical conclusions about their excellence, nor to put into question the “standard” methods and practices that have proven their relevance and efficiency. The interest of these displacements lies in experimentation and opening frontiers. They represent a means of freeing up research design, at least temporarily, from the consistency between the methods used in data collection and those used for data processing and analysis, from the probationary aspect of the measures, from the need to certify their generality (as expected from quantitative methods), or their proximity to the reality (as typically required from qualitative methods).
The six contributions in these two issues have also in common their reliance on surveys that make it possible to study temporalities, with calendars of events, life histories and institutional frames, all from a diachronic point of view. Such questions concerning time, processes and sequences are indeed at the heart of many research programs in social science (Mendez, 2010). The different types of longitudinal material seem particularly suitable for the methodological explorations proposed here. Our questions often focus on the complex processes on which life histories, business stories or the history of science are based. In continuation of the theoretical framework developed by Andrew Abbott (2001), analysis of social processes is performed by searching for typical sequences of events and temporal patterns by which, instead of getting caught up in the singularity of individual trajectories, researchers look for regularities and possible generalizations.
By displaying the procedures used to build such typologies, to stylize individual trajectories or ideas, to track relevant criteria, to obtain empirical categorizations of paths or themes, these original contributions attempt to demonstrate precisely the tools they implement. The boundaries between objectivity and subjectivity can also be blurred, considering that the contributions combine the collection of objective biographical events (for instance, occupational or residential mobility, or the experience of an historical period), traces of these events (for instance, publications related to the event), and individual perceptions and opinions about them. Consequently, the underlying causalities are related to complex configurations of conditions rather than attributed to isolated factors.
Several contributions use statistical methods but without assertion of representative ness, and sometimes without ensuring the significance of the differences identified. Others reconstruct individual histories without having interviewed people. They also use methods of comprehensive analysis of subjective representations, losing (temporarily), during the recoding and clustering operations, part of the meaning of the events and the complex arbitrations at play at key moments.
These attempts at crossover methods may therefore be based on a double frustration with respect to both the “genuine” qualitative methods and the “genuine” quantitative ones. But, far from avoiding their full deployment at other points in the research process, these explorations have the merit of trying to link them in an innovative way. This certainly comes at the cost of violating of boundaries. This may also offend the supporters of “commensurability” that claims a strict consistency between the epistemological postures and the methods. But the research presented here does not intend to advocate the discovery of a new methodological paradigm on which everyone would agree, nor does it involve theoretical postulates. This research is limited to a few proposals for opening “canonical” methods to transpositions, switches in scales and units of analysis, new forms of recoding or reformulations that can enrich the analysis and provide avenues for achieving, afterwards, more relevant and reliable techniques and tools. Other paths for investigation and analysis may be stimulated by these attempts, without damaging the need for coherence between theory and methods, but simply by separating them temporarily.
In summary, we can expect some innovations and improvements in the ways to see, to enumerate to measure and to analyse certain research topics in social science.
The Six Contributions to This Double Dossier on “Crossover Analysis”
These introductory remarks provide the conceptual framework to which the contributions of this double dossier refer. We briefly present their contents below.
The three contributions presented in the 124 issue of the BMS were developed as part of a global research project, Bipaje, acronym for Bifurcations in Youth Pathways (“
The next two contributions have in common that they suggest methodologies not directly “consistent” with the collected data: subjective well-being appreciations submitted to quantitative treatments to describe and clarify what is at the heart of the “quality” in life course; statistical and qualitative analyses of qualitative data coming from multiple sources regarding the identification of “activity patterns” in creating a new business or “entrepreneurship”. These two works both develop typologies. In one case, the typology is about life courses depending on the length, order and tone of periods of life; in the second, it is based on changes in the “activity pattern” during the process of entrepreneurship, also taking into account the chronology of these patterns and the time spent in the different regimes.
The last contribution in tracing the geographic location of the emergence over time of a scientific specialty justifies the need of combining quantitative treatments (bibliometric analysis of keywords, graphics and other methods) and interviews with key players in the field, to overcome the biases and shortcomings that would occur if a unique methodological approach had been employed.
The Three Contributions Related to the Bipaje Research Project
The first article, by Arnaud Dupray and Dominique Epiphane, uses a quantitative survey with predefined responses but which also includes a module of questions about the respondent’s opinions concerning the experiences of career change, in order to take the measure of “occupational bifurcations” in early working life. The likelihood of such a “bifurcation” is studied over a seven-year period after graduation, but the analysis takes into account on a monthly basis, with a calendar, the situations available over a ten-year duration.
In attempting to identify and measure the importance of “occupational bifurcations”, the authors resort to a “qualitative reading” of the data by going back to the employment labels extracted from the calendar. Returning to the raw material of the survey enables the authors to work in fine detail on the delimitation of the phenomenon in question. Once having identified all individual potential pathways constituting a bifurcation, a common quantitative treatment – logistic regression in the present case – is implemented to uncover conditions that affect them. The authors derive combinations of factors leading alternately to a high probability or a low probability of bifurcation. They then try to identify individuals whose initial context corresponds to those conditions, and thus display how these conditions combine and intervene in real life courses. Afterwards, their intention is to present “quasi-life stories”, taking advantage of all the information collected in the data base. They thus show that bifurcations are configured differently depending on time, psychological and financial costs engaged. The combination of qualitative and quantitative analyses firstly allows to identify a population responding to multiple and complex criteria; secondly, to achieve patterns in the contexts at the origin of occupational bifurcations; and thirdly, to show how conditions can adjust a life course and translate into a new career orientation.
Claire Bidart and Catherine Gosselin use a qualitative material, in this case biographical interviews, that are encoded in a monthly calendar of events and situations reconstituted over a 59-month period, in a preliminary phase. The authors then multiply the analytical points of view: from the quantitative one, by analyzing the calendar of events with a qualitative treatment by returning to the interviews to understand the episodes of strong “time coincidence” of events, while considering the international comparative framework in which life courses are embedded in institutional systems and societal contexts. The back-and-forth of analysis between individual units and units of episodes allows the authors to outline the national peculiarities of duration of these school-to-work transitions and clarify the meaning of time coincidence of events going back to the life courses interviews.
The authors derive from these analyses various attitudes toward change that are not independent of the societal context in which these pathways are embedded. In doing so, the authors show that these stories acquire a new intelligibility and a more general scope, going beyond their singularity. The perceptions and behaviours of these young persons, registered by interviews, are thus related to macro-sociological phenomena, such as those emerging from the distribution of individuals in social situations over time (employment, education, inactivity, education with employment, unemployment), and according to their country of origin.
The intention is to show that qualitative and quantitative methods not only provide additional insights on the question under study, but also influence each other: firstly, guiding the choice of life stories that will be told and developed; and secondly, enriching the understanding and interpretation of events belonging to a unique personal history. In this way, combining several levels of analysis allows a gain in generality by escaping from the singularity of life stories, and to show their consistency and resonance with the macro social environment.
The contribution of Sylvain Bourdon, María Eugenia Longo, Eddy Supeno and Camila Deleo is based on the same material employed by the preceding authors, but over a shorter period of time (24 months), and grants a major role to the subjective perception of events by respondents. Indeed, the authors propose a method of recognizing and describing “high points” or “key moments” in the time during which young people enter adult life and access employment. The authors construct an “intensity index” of the “key moments”. The intention is first to synthesize a wealth of information and point out sensitive and critical moments that could have destabilized or changed the orientation of a life course. In addition to their analysis of “key moments” (types of events, frequency, temporal concentration, link with career changes, etc.), the authors in a final step return to the qualitative material of interviews to contextualize these eventful moments and their emotional intensity. Life stories then underscore the sequence of events and the genesis of the “key moments” in time. They also provide a means of testing the measurement scale retained to calculate the weight of events and to define their influence in subsequent months. The “intensity index” of “key moments” presents itself as a heuristic tool for synthesizing information and concentrating on particularly eventful periods. In addition, it allows one to highlight the interdependencies between events belonging to different spheres of life (emotional, occupational, residential, family, etc.).
Two Methodologies Not Directly “Compliant” with the Collected Material
The following work by Eva Lelièvre and Nicolas Robette breaks with these medium-term temporalities to consider an annual calendar of life over a period of 50-70 years. The aim is to address quantitatively a highly subjective topic: individual perception of quality of life. The original material of the Biographies et Entourage survey conducted in 2001 by INED and used here allows the authors to connect collections of factual situations (related to family, emotional life, employment, housing, etc.) with subjective perceptions by respondents of this organized story of their lives, carried out through the data collection framework. Indeed, people were asked to split their lives in different periods, to characterize them and to make a judgment on the overall “tone” of each era as years that were from “very difficult” to “very good”, with three intermediary descriptive terms.
The recourse to a quantitative assessment of the corpus is thus provided immediately by this material that lends itself to the confrontation of factual events and judgments by the respondents. Applying adequate statistical treatments provides a way to reconciling life course milestones and biographical turning points (the passage from one era to the next). The authors show in particular that men and women do not attach the same importance to the same events: family events are more often associated with biographical turning points for women than for men, while the opposite is true for business events. An Optimal Matching method followed by ascending hierarchical clustering is then applied to annual records of tone to synthesize in a few classes this extreme diversity of life courses. A presentation in the form of time charts (chronogrammes) identifies each class of life course by its characterization in terms of tone. The classes are then compared with individual socio-demographic characteristics (gender, occupation, educational level) and life events (separation, migration, death of a close relative, etc.).
To study the process of business creation, Jean-François Barthe, Nathalie Chauvac and Michel Grossetti deal with a rich qualitative material consisting of interviews, conducted in two waves, with business managers and entrepreneurs, and of written materials. These sources were used initially to form “narrations” of stories from which information is systematized and formalized for statistical processing. From these narratives and a few quantitative indicators, the authors are capable of distinguishing between five “activity patterns” (exploratory, in crisis, survival conditions, project, marketable) which correspond to possible stages of evolution of these entrepreneurial activities.
After describing qualitatively these “activity patterns”, each of the 66 companies studied is encoded according to the annual “activity pattern” which describes it, offering the possibility of building a comprehensive classification tree presenting all successions of patterns. The corresponding trajectories or pathways can afterwards be described quantitatively to yield a typology of evolutionary paths divided in six classes.
Despite the efforts to objectify criteria distinguishing between “activity patterns” and the passage from one to the other, the authors do not circumvent the difficulties of such a systematization which is often obliged to return to the interview material to confirm affiliations. As for the analysis of individual calendars of situations that allows changing the unit of study, the suggested approach is adapted to capture temporal processes involving multiple actors and, where it is permissible, to diversify the type of units under study: people, relationships between them, businesses, etc. In addition, the quantification of information is made possible “by pulling up” the analysis of qualitative material attesting to the mutual enrichment of quantitative and qualitative treatments achieved through this coupling.
Bibliometrics and Interviews with Key Players of a Specific Field
The last contribution to this double dossier relates to the sociology of science and combines quantitative (descriptive statistics, graphs) and qualitative methods (individual interviews) to explore how a scientific specialty (DNA repair) emerged in real geographical areas by taking into account the location of the first researchers in the field and their subsequent geographical mobility. The methodological interest of this contribution is in how it shows the limits and deadlocks of any approach that would stick to a strict bibliometric analysis or, inversely, any approach that would only give credit to the testimonies of actors in the field. A mixed method is proposed by Marion Maisonobe, which combines a bibliometric analysis and a “prosopographic” approach.
Counting publications based on key words in data bases, the frequency of publications per authors and per laboratories (to which the authors under study belong), and finally the spatial location of these units allows one to detect the points of concentration of research production in this field. Interviews with major authors on this topic in the years following the onset of items comprising “DNA repair” as a key word, helped trace the geographical locations from which the spread of related academic contributions took place. Several bursts of production and diffusion are thus identified, particularly through the institutional mobility of actors in the field.