1The first issue of the BMS was published in October 1983, 25 years ago (see Figure 1), and almost immediately became a bilingual English-French quarterly scientific journal. The present issue of the BMS is the number 100 issue.
2In January 1985, in the number 5 issue, the BMS published the first presentation of the Research Committee “Logic and Methodology” (RC33) of the International Sociological Association (ISA) (RC33, 1985), with the following sections: “Association Internationale de Sociologie”, “XI World Congress of Sociology”, “Report from the President of RC33, Dr. David J. Jackson”, “Call for papers for sessions of the Research Committee on Logic and Methodology – RC33”, and a report on the 3-5 October 1984 conference in Amsterdam of RC33 and the Dutch Psychometric Society (van Meter, 1985). That material constituted the first issue of the RC33 Newsletter, which the BMS has continued to publish twice a year. The Amsterdam conference became known as the first of a series of international RC33 conferences. The seventh such RC33 conference just took place in Naples on 1-5 September 2008 and the RC33 Newsletter in this issue includes several session reports concerning that rather successful international conference.
Figure 1: Cover of the BMS, number 1, October 1983
3In December 1993, the BMS finished its first decade of publishing and, in that number 41 issue, published in its “Brochures/Reviews/Reports” section (pp. 62-91) the full tables of contents of the BMS from issue 1 (October 1983) to issue 40 (September 1993). Those tables of contents were followed by an alphabetic index of authors of research articles (pp. 92-94) and an alphabetic index of the titles of research articles (pp. 94-98)). An index of key words existed at the time, but was too long and too cumberson to be published on paper. At the time, there was no attempt to to analyze this material, even with simple cross-tabulations of the most frequently occuring key words.
4Ten years later, in January 2004 more precisely, the BMS finished its second decade of publishing and issue number 81 included the complete list of tables of contents, the author index and the article-title index for all research articles published in issues 40 to 80. But this time, the BMS used its methodological expertise to analyze the contents of the journal (van Meter, Cibois and de Saint Léger, 2004; see also http://bms.revues.org/index1141.html). Trideux factorial correspondence analysis and Calliope co-occurrence of key word analysis were used to analyze the data base consisting of all key words characterizing each research article or ongoing research report published by the BMS from December 1993 to October 2003.
5The bidimentional factorial correspondence analysis diagram that Philippe Cibois produced (see Figure 2) showed a first or horizontal axis extending from the left, with “Chi-Square”, “Interviewing”, “Sampling”, etc. to the right where you find “Validity”, “Comparative”, “Collection”, etc. This seemed to be a “methods” axis going from more formal methods to more abstract methodological considerations (ibid.: 49). The second or vertical axis moved up from “Text”, “Language”, “Poverty”, “Well-Being”, etc., to “Careers”, “Time”, “Profession”, “Errors”, etc. It appeared to go from more social considerations and approaches toward more professional considerations. Not surprisingly, we found “Sociology” in the center with a closely packed collection of specific methods of analysis grouped around it. Although there are outliers, the collection of key words is organized in a relativly concentric manner around the origin of the graph with density increasing toward the center (ibid.: 49).
Figure 2: Bidimentional factorial correspondence analysis diagram of 1993-2003 BMS research articles
6The Calliope method of classification using co-occurrence of key words as a measure of similitude (de Saint Léger and van Meter, 2006) produces a bidimentional strategic diagram with a horisontal axis of “centrality” (relative strength of external ties) and a vertical axis of “density” (relative strength of internal ties). This divides the diagram into four quadrants with the first quadrant (central and dense) often referred to as “mainstream science”, the second quadrant (dense but not central) as “ivory tower” science, the third quadrant (non-dense, non-central) as “chaos” or “unstructured science”, and the fourth quadrant (central but not dense) as “bandwagon science”.
7Rather characteristically for a social science journal published in France and in French (as well as in English), the terms “Social Science” and “France” figure in quadrant one of the strategic diagram (see left-hand side of Figure 3a and Figure 3b). Both terms are central (see the external ties in the strategic diagrams on the left), and dense (see the strength of the internal ties in the diagrams on the right). Note that the strength of these ties is relative to an initial threshold of 50. Below 50, ties are drawn as dashed lines and not as continuous lines of varying thickness proportinal to the strength of tie (ibid.: 51).
8The “Social Science” class shows the importance of the ISA Research Committee 33 (RC33) and its October 2000 conference in Cologne (BMS, 2000; van Meter, 2003). Not only was the Cologne conference reviewed extensively in the BMS, as the recent Naples RC33 conference is in this issue of the BMS, it was also the source of several articles published in the BMS. This central sociological methodology class, which appears to be thematic, is complemented by a more geographic but international class centered around the term France but including Germany, Great Britain and the Netherlands, all countries wher RC33 is well-represented and producing sociological methodology articles published in the BMS. It is interesting to note that a specific type of sociological methodology is also associated with these four countries: social network analysis. Indeed, these four countries are active in that field and several have hosted one of the international “Sunbelt” social network conferences.
9This year, the BMS not only celebrates a quarter of a century of publishing but also its membership in Revues.org (visit http://bms.revues.org). At Revues.org, BMS has a three-year “moveable barrier” which means all research articles and ongoing research notes are free and fully availble on the Web three years after their initial publication in the BMS. However, for all BMS issues, the titles, author names, abstracts and key words are fully available and the titles, author names and key words are indexed in both English and French. Although the entire series of the BMS articles is not yet been published on Revues.org, it should be in the not-too-distant future.
Figure 3a: Strategic Diagram of BMS 1993-2003 BMS research articles
Figure 3b: Strategic Diagram of BMS 1993-2003 BMS research articles
10The three types of indexes are among the many Web-based tools made available by Revues.org to search through the collection of BMS articles. Like most Web-based tools, the searches can only be univariate and not multivariate cross-tabulations, but Google-type searches can use several terms at a time. But this remains short of Boolean searches or a full multivariate analysis of the contents of BMS research articles. For that purpose, we have constituted a complete collection in English and French of all BMS research articles and ongoing research reports from issue 1 to issue 99. That collection has been divided into ten different files: BMSAbs1-9, BMSAbs10-19, BMSAbs20-29, BMSAbs30-39, BMSAbs40-49, BMSAbs50-59, BMSAbs60-69, BMSAbs70-79, BMSAbs80-89 and BMSAbs90-99. These files should soon be available on the BMS Web site at the Ecole Normale Supérieure in Paris: http://www.cmh.pro.ens.fr/bms/Welcome.htm
11Using these files, the abstracts and key words in French were extracted and divided into three corpora: articles from 1983 to 1993, articles from 1994 to 2000, and articles from 2001 to 2008. These three files were used by Mathilde de Saint Léger to perfom an analysis by Calliope presented in this issue in her article “Comment ont évolué les thématiques des 99 premiers numéros de BMS? Analyse à l’aide de logiciel de fouille de texte”. Gérard Boudesseul used the same three files to perfom a correspondance analysis and both an ascending and a descending classification analysis.
12Calliope produced a strategic diagram for each of the three periods. The dense and central first quadrant of these diagrams included the clusters: "Qualité donnée" (data quality), "Entreprise" (businesses), and "Enquête" (surveys) for the first period; "Questionniare", "Grande-Bretagne" (Great Britain), and "Qualitatif" (qualitative) for the second period; and "Corpus", "Enquête", "Statistique" (statistics), and "Individu" (individuals) for the third period.
13We can summarize these results by saying that "Questionniare" swallowed "Qualité donnée" and added cognitif pre-testing during the second period. "Grande-Bretagne" is largely a geographic thematic cluster (France, Germany and Europe), similar to that seen above (de Saint Léger and van Meter, 2006), and "Qualitatif" is, of course, a cluster associated with “qualitative” sociologie. In the third period, "Enquête" is a continuation of "Questionnaire” and "Statistique" a continuation of "Qualité donnée". With decreasing maximum values for density, the thematic clusters experience a relative loss of internal coherence, which may be the counterpart of increasing interconnectivity. With this development, and the example of the thematic evolution of "Qualité donnée" toward "Statistique", de Saint Léger presents a hypothesis: sociological methodologies were clearly delimited in the early 1980 but, with the increased use of computers, these methodologies became increasily interconnected. “Thus the themes centered around ‘Qualité donnée’ in 1983-1992 can be considered included in the more global framework of ‘Statistique’ during 2001-2008.” In a similar vein, the cluster "Individu" in the third period recovers some of the themes of the cluster "Qualitatif" in the second period.
14To support her hypothesis, de Saint Léger also cites the increasing maximum centrality values since this corresponds with the increasing interconnectivity and openness of specific methodologies. Also, the “emerging” terms show the increasing important of computer tools and the objects of study, relative to particular methodologies. Inversely, the “declining” terms show the deceasing presence of particular methodologies and data collection, relative to the objects of study. De Saint Léger ends by asking: Can one say that the various methodologies are clearly delimited? Is that evolution due at least in part to the development of computer tools?
15In summary, with Calliope, one can see continuity in the major thematic concentration on both formalized survey research and its associated statistical analysis (traditionaly labeled "quantitative" sociology) and indepth interviews and text-based research and analysis (traditionally labeled "qualitative" sociology). These themes are systematically found in the dense and central first quadrant, and regularly change their relative positions and importance in relation to each other. They in no way can be construed to be "poles" or "opposites", but are mutually associated central themes evolving together over time.
16In this issue of the BMS, Gérard Boudesseul, in “Un quart de siècle de méthodologie - L’émergence du longitudinal, des réseaux et de la statistique textuelle”, like Mathilde de Saint Léger, uses the titles, abstracts and key words (in French) of the 318 articles in the BMS, issues 1 to 99, as a corpus for analysis. Boudesseul uses hierarchically ascending and descending classification programs and also a factorial correspondence analysis program. With the results of these three methods, he describes in detail five major thematic classes: Survey Situation, Longitudinal Analysis, Social Network Analysis, Theoretical Debate and Textual Analysis. The interrelationship between these five classes is described and their evolution over time is examined.
17According to Boudesseul, “two research traditions seem to be relatively atemporal and therefore constitute an integral aspect of sociology and the methods debates that extend throughout it. One is the methodological precautions to be taken during the crucial moment of data collection. The other is the debate concerning paradigms, neighboring disciplines and epistemology. A second major result is that there are three particularly important thematic currents in the BMS, each with its own particular time period: longitudinal analysis and its associated categories, particularly memory, recall and retrospective data; social network analysis with its specialized and formalized vocabulary of graphs, nodes and links; and finally textual statistics which try to reconcile the reasonable use of numerical tools and borrowed material from semantics, and linguistics in general.”
18In short, identifying survey methodology, social networks and textual statistics as major continuous and evolving themes of BMS research articles reinforces our previoius statement that these themes “in no way can be construed to be ‘poles’ or ‘opposites’, but are mutually associated central themes evolving together over time.” Quite clearly, Boudesseul’s Survey Situation class corresponds with de Saint Léger’s Survey (first period), Questionnaire (second period) and Survey again (third period). Boudesseul’s Textual Analysis corresponds very closely with de Saint Léger’s Qualitative (second period) and Corpus (third period). And social networks has systematically been a key member of dominant clusters in de Saint Léger’s results over the three periods, even if its has not been a dominant cluster by itself.
19As for the dispersed and “atemporal” nature of Boudesseul’s Theory Debate, it is surely a reflection of BMS editorial policy in that the BMS doesn’t accept “theoretical” articles which do not have an application included in the article. This means that “theory” is always associatted with some other topic associated with an application and a method.
20In summary, four different methods of textual analysis arrive at surprisingly similar results concerning both the thematic content and evolution over time of BMS research articles and research reports. These results can of course be developed further and readers wishing to do so can find the complete corpus on the BMS Web site.