Writing the Research Articles

1. Structure

The principal medium by which studies reach other researchers is the research article, published in an academic journal. The structure of these articles is very often identical, even where guidelines are not precisely defined. The structure that I outline below, which is almost universal for quantitative research, is equally well adapted to qualitative research. For example, the article present­ing the quantitative research of Zajac and Westphal (1996) adopts the same structure as that of Barley’s (1996) qualitative research published in the same issue of Administrative Science Quarterly.

This structure usually consists, after the abstract, of an introduction, an analysis of existing literature, presentation of the methodology, presentation of the results and discussion of the results. A fundamental point in writing an article is that a logical sequence is essential, through which the reader can understand the argument and follow its development. This is achieved by con­centrating on research aims, avoiding digressions and taking care to clearly explain transitions.

The abstract is a difficult exercise, all too often replaced by a paragraph taken from the introduction. A good abstract must present in a few words the context of the study and its principal results.

The introduction is generally fairly short. The interest the topic holds, theore­tically and/or practically, should be made clear.

The literature analysis enables the study to be situated in relation to previous research. The author should underline any divergences, and may establish links between related fields. One must, however, stress the importance of stay­ing focused on the research question, and so guiding the reader. This analysis may, where appropriate, lead to the formulation of hypotheses.

The methodology presentation explains the different stages of the research, and the order followed. Here we find descriptions of any samples used or of the case or cases studied, operationalization of the concepts, and discussion of the analyses conducted. It is important to recount exactly what was done, without any rationalization after the fact (Daft, 1995).

Results are limited to the main findings. As this part is usually fairly short, the results can be presented in a summarized form, as in the case of quantita­tive analyses, where tables are frequently used.

A discussion of research results puts them in the perspective of existing studies. Agreements and discrepancies are pointed out, and explanations for these are explored. In this section, the author will also talk about the implications of his or her research, in both theoretical and practical terms. We will also find here an analysis of the limits of the study, and often propositions are made for future research arising from these new results.

2. Form: Figures, Tables, References and Footnotes

2.1. Figures

It is very difficult to produce an interesting diagram that will have a real impact on the reader. A too-complex diagram will be illegible, and a too-simple dia­gram will not provide any information. Two essential types of diagrams are used: those presenting relationships (I will refer to these as models), and those presenting data (graphs). Models allow us to illustrate a process that would be too cumbersome to describe, or visually represent a group of relationships among variables. Similarly, graphs are visual representations of quantitative data. Graphs are an effective way of drawing attention to aspects considered to be important.

As Daft (1995: 178) points out diagrams can be adapted to every type of research. ‘For traditional hypothesis-testing studies, a visual representation strengthens the early part of the manuscript and can be revised in the conclu­sion section. For qualitative studies, a visual representation at the end of the manuscript is an ideal way to crystallize the theory developed from personal observations.’ As regards presentation, figures should be referred to in the text, and should be self-explanatory.

2.2. Tables

Tables, which are used very often, offer readers a visual summary of a discus­sion. They can present a synthesis of a literature review or of research data, whether quantitative or qualitative. As with diagrams, tables should be referred to in the text, and must be comprehensible in their own right.

2.3. References

References are important in that they situate the research within a body of theory and give credibility to the results. They must therefore be selected with care. For this, three basic criteria should be used: relevance, credibility, and accessibility.

A reference will be relevant if it gives something to the reader. It has to be strongly linked to the problem concerned. An article should be cited as a refer­ence only because of its results, and not for some minor point that it may only mention in passing.

The credibility of a reference depends essentially on its medium. A top­ranking journal will be regarded as very credible. It is important that references have the same level of credibility. Credibility also demands that authors follow up their sources: it is preferable to cite an article in which an idea originated, rather than a more recent work which has only reworked this idea. This can, however, pose a problem in the case of a work which is out of print.

In this regard, accessibility relates to the ease with which readers will be able to obtain the references cited. It is preferable to make use of works published in journals with a wide distribution.

A common error is to introduce too many references. In some manuscripts every phrase, or nearly every one, is backed up with one or several references, even when it treats only a minor aspect. Such a superabundance causes the force of the main argument to be lost, drowned under a mass of minor information. Campion (1997) conducted an inquiry among reviewers to determine criteria with which to select references to use in research articles (see Table 17.3).

2.4. Footnotes

Most journals recommend limiting as much as possible the use of footnotes, which break up the reading. Some journals simply forbid them, on the basis of the following principle: if the point is important, it should be in the body of the text, if it is not important, it should be removed.

3. Style and Qualitative Research

The style of writing to use for qualitative research articles has been addressed in numerous articles and books. The form of such writing too is crucial, particu­larly for two quite different reasons: the difficulty of presenting qualitative research and the realization that ‘data’ is already very strongly a matter of interpretation.

On the first point, Richardson (1990: 53-4) remarks:

Shaping qualitative material for mainline journals requires telling your research story in a way that meets the expectations of the editors/reviewers of those journals. Quantitative writers have an easier time with this, for their work is already strongly encoded as positivist-empiricist (through such rhetorical devices as tables and path diagrams) and because the criteria for judging the work is more precise and shared among the community of believers. Qualitative writers have a more difficult time, I think, because their papers have fewer strong encodings, and their reviewers have fewer and less precise agreements about significance and other such matters . . .

In a somewhat provocative article, Sutton (1997) presses the argument, and identifies four situations in which it can be preferable to hide or at least mini­mize the role of qualitative data in the development of a study: ‘when weak qualitative data lead to good insights, when describing the qualitative research reduces the quality of the writing, when an outlet does not publish “empirical” papers, and when writing for an audience that remains biased against qualita­tive research’ (Sutton 1997: 97-8).

The second point stresses the fact that no description can be neutral, that all texts have their biases (Denzin, 1994; Mitroff, 1972). This takes on special impor­tance in a qualitative project. As Geertz (1973: 9) points out in ethnography, a description is thick with the interpretations of the researcher: ‘what we call our data are really our own constructions of other people’s constructions’.

Consequently, the distinction between data and analysis becomes at best difficult, perhaps impossible. The whole chain of evidence of a positivist approach is put into question. To convince the reader of the sound basis for the study, one can, as Johanson (1996) advises, set up a network of presumptions.

Drawing an analogy with the judicial process, she advises authors to try to convince the reviewers in the way that a prosecutor tries to convince the jury. In the absence of formal ‘proof’, it is a question of winning over the jury by a body of elements that mutually reinforce one another.

3.1. Case studies

A case study can be presented in different ways, each of which have advantages and drawbacks, and each of which favor one or another purpose. The first pos­sibility is to set out an account which is as neutral as possible, so that readers are left to form their own opinion. To achieve this a chronological account may be adopted. However, we should be aware that the apparent neutrality of the account is deceptive: certain aspects have been chosen for discussion, and others not, the structure of the text implies a particular logic, and so on. The difficulty of this choice lies in its structure, which does not allow for the topics covered to be grouped according to type. This makes comprehension difficult, and weakens the arguments. The solution most often adopted is to follow the account with a discussion of the analyses according to themes, which necessitates repetition and lengthens the document. By contrast, another possibility is that of breaking the case study up into the topics discussed. Here the advantage is in having an article that is more structured, more focused, but the elements of the context are minimized. This kind of structure can also invite criticism as it can seem that the author chooses to describe examples that support his or her theory. Glaser and Strauss (1967) call this exampling and rightly condemn it.

All the same, if several cases are to be presented, one can arrange a presen­tation in a case-by-case order of priority, or transversely, in order of topics. The presentation of individual cases has the advantage of giving an overview, but to the detriment of comparisons, which can mean that the argument is less con­vincing. The transverse presentation is, on the contrary, more analytical, but makes it very difficult to give an overall picture.

3.2. Style

The style of writing that is adopted takes on a special importance in qualitative research. The writer has to convince readers without using figures. An ethno­graphic text involves three dimensions (Golden-Biddle and Locke, 1993): authenticity (assurance that the researcher was there and was genuine about the experience in writing up the account), plausibility (does this have meaning; does it contribute to our knowledge?) and criticality (does the text impel readers to reconsider ideas and beliefs they have taken for granted?). Authenticity is obtained by particularizing everyday life, by delineating the relationship of the researcher to the field, by describing the methods used in collecting and analy­zing data, and by qualifying personal biases. As for plausibility, meaning is gained by encoding the text so that unorthodox methodologies are normalized, by legitimating the atypical and smoothing the contestable, and by presenting oneself as an expert. The contribution the work makes is emphasized by point­ing out gaps in existing studies, and in leading readers to feel that something new has been presented.¹ Finally, criticality is obtained by challenging readers to reconsider the assumptions that underlie their own work.

Still in the area of ethnography, Van Maanen (1988) distinguishes three princi­pal styles, and illustrates them by examples drawn from his own work:

• The realistic style aims to be neutral and impersonal. It is characterized by the absence of the author in the final text, by relying on precise details arranged redundantly into categories, by the point of view of the subject being pre­sented through quotations, and by the interpretative omnipotence by which the researcher assumes the right to interpret and describe reality.
• The confessional style, by contrast, is very personal. Here the researcher recounts in detail any difficulties that arose, and incidents that occurred in the field. The three conventions that characterize this style are: the researcher is brought to the fore; the researcher is strongly involved in the field, and his or her point of view is taken into account; and a final distancing of the researcher helps to return a degree of objectivity to the account.
• The impressionist style focuses on the process, rather than on the results of fieldwork or on the researcher. This style is characterized by: a raw account that aims to make the experience come alive to the reader; by the fragmenta­tion of the results because of this linear narration; by the personal way in which the subjects and the researcher are described, which makes the story more lively; and by its dramatic flavor, which imposes standards of literature rather than standards of the appropriate discipline, that is, of ethnography.

It is clear that the choice of style is not a neutral decision. It reflects the epis­temological position of the researcher, his or her conception of research itself. This link between epistemology and writing style explains the homogeneity in tone observed in any given journal. Given this, authors are wise to identify the journal whose style best corresponds to their own, or otherwise to conform to the style of the journal in which they seek to be published.

Source: Thietart Raymond-Alain et al. (2001), Doing Management Research: A Comprehensive Guide, SAGE Publications Ltd; 1 edition.