As in any other internationalized industry, writing an academic paper is a highly competitive activity in all scientific fields. Management scholars in particular, however, have our own unique challenges. As with the rest of our colleagues, we struggle to find space in different publications to spread our insights, but very often we end up delivering our products—our ideas—with serious misspecifications given our customers’ preferences (i.e., the interests of each journal's potential readers).
To begin with, we can identify academic and practitioner-oriented journals that require different approaches to—metaphorically speaking—our product design, industrialization and delivery. Not only are problem statements different in both types of journals, but the language we use, the way we present our arguments and analyses, or even the paper length and format often have many disparities. So the Harvard Business Review is a great journal, but most of its contents would not find space in BRQ, and vice versa.
We can also find different goals, methodologies and even styles among scientific disciplines and management fields. One can easily identify diverse and frequently conflicting requirements, for instance, depending on whether a management journal reflects an economics, sociology, psychology or engineering perspective. Additionally, although generalist journals are usually open to methodological diversity, specialized outlets in finance, human resource management, marketing, etc., often have their own format. Even within the very same scientific tradition and management field, however, academic conversations are frequently different: no matter how original our problem statement is, different journals ask us to build our arguments based on different threads of research that may span several years.
The fact is that, over time, the number of submissions in BRQ (342 in 2017) has grown parallel to the rate of desk rejection, which is above 90%. To be sure, if the journal has the same publishing space for a higher number of submissions, the increasing rate of rejection should not be a surprise to anyone. A second explanatory factor certainly has to do with our editorial policy, which is based on low-risk revisions and short turnaround times to avoid imposing on authors the pain of second and third round rejections. There is a third reason, however, that is probably embedded in a common observation I have often heard about our journal: considering it is a Q3, BRQ has a reputation of being a tough journal.
We have given this reputation considerable thought. I, myself, have wondered if the type of editors for whom we search has provoked a selection bias with obvious consequences on rejection rates. This could be the case, but I truly do not see how an alternative strategy of choosing less skilled and thorough colleagues could be of any benefit for BRQ. I believe we do have a great challenge, nevertheless, in addressing the intersection of two phenomena: the evolution of the scholarly publishing business, on the one hand, and the recent presence of BRQ in the international market for ideas, on the other.
Let me put it bluntly: the indiscriminate worldwide pressure to publish in any of the numerous indexed journals has generated wrong professional practices, which mainly affect new entrants such as BRQ that maintain similar standards as those of the top incumbent journals. Note that there are hundreds of journals in management-related areas, and the correlation between scientific standards and impact factor may be statistically significant but not necessarily too high. This creates many misunderstandings for scholars around the world as to what exactly journals expect from authors regardless of their impact factor.
Along these lines, taking into account that BRQ is the flagship of an academic association (ACEDE-Spanish Academy of Management) whose goals and practices have likely by far exceeded its initial size and nature when it was established 28 years ago, our expectations are indeed not shaped by the impact factor we obtain each year. It is more a question of how we see the business of science. In fact, even if we realize these standards will not protect BRQ against the possibility of type II errors (i.e., accepting papers that should not be accepted), we are convinced they differentiate peer-reviewed “scientific” research from pure anecdote. This is what I mean when I refer to how we see the business of science compared to other businesses in the academic arena.
So next you will find a report on how we see our business, with specific suggestions—based on my own experience as Editor of BRQ for the last four years—on how to increase your chances of publishing in our journal. As a Q3 journal for the time being, we can show a certain tolerance with regard to the originality of the problem statement, the surprise of the hypotheses, sample representativeness or the novelty of claims and implications. However, the difficult equilibrium we must obtain among the different parts of a single paper cannot lead to any tolerance at all that minimizes the relevance or credibility of what we publish.
I am sure many readers will qualify some passages below with technical nuances or even contrasted views on how a paper should be built in general. It is good to remember, accordingly, that at BRQ we will cherish your freedom to be creative and appreciate your own ideas for their intrinsic value, regardless of whether you follow all of the following suggestions. Thus, if you think “it ain’t broke, don’t fix it”.
The relevance and originality of “your problem”A paper chance of being published in BRQ is not based as much on the paper itself but on the problem the authors chose to address—likely many years before they sent us their paper. What type of problems do we search for, anyway?
BRQ is an academic journal, which means that our readership is mainly composed of university professors whose main interest is the advancement of science, even in incremental measures. This advancement is our business. Therefore, even if we might truly enjoy reading papers about—say—Chinese crowdfunding, incentives in Colombian firms, marketing strategies in Spain or entrepreneurship rates in Poland, we will not consider them for publication unless they offer thought-provoking ideas that make Chinese, Colombian, Spanish or Polish evidence relevant to the advancement of our knowledge, respectively, on crowdfunding, incentives, marketing and entrepreneurship. To put it bluntly, the nationality of one's evidence per se does not make a paper worthwhile.
There are, however, some caveats to this assertion. One could argue that, just as country specificities can be a source of biases that may compromise the generalization of results, they can also be a source of insights to the extent the institutional context can question long standing results that have generated a particular conventional wisdom. For instance, what we believe holds as a general rule in a particular field is sometimes the result of Western-based business evidence. In fact, political and cultural factors that are taken for granted can provoke particular results that otherwise, in a different political and cultural setting, could yield different outcomes. Let me use a trivial scientific allegory to illustrate my point:
Assume we are scientists trying to determine the boiling point of water. We have performed an experiment in a lab, for instance, in Boston, Massachusetts (USA), and we have reached the conclusion that water boils at 100 degrees Celsius (212 degrees Fahrenheit). As soon as we obtain this major breakthrough, we try to publish it in one of the top journals. I can imagine a brief and easy title: “The boiling point of water.” Now, since we have properly explained our experiment, many colleagues will surely have replicated the methodology in—say—Seattle, São Paulo, London, Copenhagen and Rome and have reached the same conclusion: water boils at 100 degrees Celsius. These new experiments and their findings can still be published shortly afterwards, although they will probably not find a top outlet since the originality of their results is much lower. After a short period, no matter how sophisticated the experiment is and the country in which it occurs, no journals will wish to publish a paper suggesting that water boils at 100 degrees Celsius. It is already conventional wisdom…unless someone finds something “rare” that can be linked to the conditions of the experiment. For instance, imagine we receive a paper from some colleagues in the Himalayas suggesting that, in their experiment, water boils at less than 100 degrees Celsius. They are not certain why, but they hypothesize that the pressure might have something to do with their different results given the altitude of their facilities. The title of their paper is along the lines of “The moderating role of atmospheric pressure in boiling water: evidence from the Himalayas.” These colleagues have made the “institutional context” matter in addressing an old problem statement, so they will surely find a good journal to publish their insights. Otherwise, merely replicating the experiment in the Himalayas, no matter how beautiful those mountains may be, is worthless in scientific terms.
Choosing a scientifically relevant problem statement is therefore a key issue. In a journal such as BRQ, we might not ask of you the novelty and relevance that the top five generalist journals in management may require; however, we will definitely be very demanding. We will not actually exchange methodological neatness for less-relevant problem statements. In fact, we would rather have simpler analytical methodologies that address more relevant managerial problems.
As an author myself, I realize that problem statements occasionally evolve from the beginning of our research to the moment we submit our papers. Sometimes one cannot find all the data we thought we would have available. On other occasions, we do find the data, but its quality in terms of missing values or outliers makes us choose a slightly different path. In addition, of course, even when we have all the high-quality data we desire, our analyses sometimes do not confirm our first intuitions. Indeed, many things can occur during a research project that may alter the initial goal of a specific research effort. My impression from the hundreds of submissions BRQ receives each year, however, is that many papers are still trying from the outset of the research project to prove that water boils at 100 degrees Celsius.
Constructing titles to predict content, offer novelty and increase your H-IndexEditorial offices are often the first to review submissions from a formal perspective, but there is no such thing as a tense flow between editorial offices and editors. For a journal such as BRQ, papers are sent to the editor approximately twice per week in batches, sometimes once per week. For this reason, it is good to remember that editors have their own work in progress including PhD students, teaching and academic service (even family). This means that when they sit down and start reviewing each batch, article titles can be compared to first impressions on a date. It would not be clever to base a decision on a first impression, but it definitely shapes an editor's perception.
Assume therefore that one is addressing an original and relevant scientific problem. How should it be phrased as a title? At BRQ we encourage you to use titles that help readers predict content, offer novelty and curiosity to catch their interest, and contain the appropriate key words so that colleagues can find your paper easily through a computer search (therefore increasing your chances of achieving a higher H-index).
If you send your work to a generalist journal such as BRQ, it is important to note that you are writing to a wide audience of researchers in fields such as finance, operations management, marketing, strategy, corporate governance, accounting or innovation. Any colleague working in different fields of management should be able to understand what you are trying to address. Also remember that the editor's research agenda might be far from yours, so please avoid abbreviations in titles and abstracts. I have found myself somewhat desperate reviewing submissions and trying to determine the meaning of a particular acronym in a title. If titles can be compared to first impressions on a date, I assure you that an inscrutable title will not have a positive impact on an over-burdened editor. Such a title might leave him or her wanting to run to safety; the safety of a fast desk-rejection.
Therefore, I would say that being clear in a title is a sound strategy to guarantee that the reviewing process gets off to a good start. Authors can choose to keep it simple and brief, as in “Friendship at work and error disclosure” (Mao and Hsieh, 2017) or “Family involvement and hotel online reputation” (Diéguez-Soto et al., 2017). However, one can still reflect simplicity with more informative and thought-provoking titles that often reproduce a question (frequently starting with an interrogative pronoun), as in “How brand post content contributes to user's Facebook brand-page engagement. The experiential route of active participation” (Gutiérrez-Cillán et al., 2017); “What makes firms embrace risks? A risk-taking capability perspective” (Tsai and Luan, 2016) or “Where do you want to take your family firm? A theoretical and empirical exploratory study of family business goals” (Basco, 2017).
To be sure, the lengthier the title, the higher the probability that it may result in a lack of focus and take the reader's attention away from the key message. In this sense, my advice is to avoid lengthy titles that include the specific statistical or econometric tools employed in the empirical analysis. I realize that sometimes it is worth highlighting the specific technique that one uses because it may be part of the contribution. In BRQ, you will find editors and reviewers with good skills in quantitative analysis who will help you to develop the paper in the right direction. Too often, however, authors send a negative signal by including quantitative techniques in paper titles. We find too many papers trying to hide the low attractiveness of a problem statement behind the name of sophisticated analytical techniques. No matter how sophisticated an author believes that his or her empirical analysis is, in the absence of an intriguing problem statement, papers are still boring, and they will tend to be rejected quickly.
Compose your abstract to get 30 additional minutes from EditorsLet us move forward to abstracts. Imagine yourself carrying out an online search on the Internet or in specific databases. When we roll the mouse wheel to scroll down the list of papers, we may find in some titles particular keywords or a sense of novelty and curiosity. Whenever this happens, the next step is to take a look at the abstract to determine if it actually fits our interests. Therefore, abstracts are part of the “first impressions” package of each paper and serve as a selling tool. We must convince the reader to download our paper and keep reading. In the case of a submission, we must convince an editor that our work is at least worth 30minutes of his or her time, which is approximately what it will take for them to obtain a glimpse of the introduction, hypotheses, data and conclusions. So what does BRQ expect from an abstract?
These recommendations should not be seen as a sort of straitjacket to stifle your creativity, but I strongly suggest that you begin by establishing your problem statement. It should be something along the lines of “We address/explore X,” “This paper analyzes/examines Z.” Go for the jugular. You have only 150 words and less than 30seconds to make an impression on editors and reviewers, so right afterwards tell us what you have found: “We show/argue that…” Whatever you write in these two first phrases, try to be incisive and make it as surprising as possible. Then, in a third phrase, explain how you have reached your results: your methodology. Say something to the effect that “We test our propositions on a wide dataset composed of XX workers/firms.” You can of course change the order of these three phrases or integrate them. For instance: “Based on a 10-year dynamic panel data (DPD) from two official sources (X number of observations), our results show that…,” or “Based on comparative longitudinal case analyses of seven stat ups in the IT industry, this paper explores how X affects Y. We find…” Finally, tell us how your results may impact theory, practitioners and/or policy makers: the “so what” question, the implications, the reason why your results are important. Do not take this last step for granted, as if it is obvious for colleagues reading your abstract, especially in a generalist journal whose editors may not be specialists in your field.
Lastly, be perfectionist throughout the text, but especially in your abstract. It is fine if your problem statement or your findings are not as original or important as they should be to publish in journal X or Y. You will find journal Z. However, if you do not express yourself correctly in the abstract, if you do not make the word count, or if your abstract has grammar or spelling mistakes, do not expect to find sympathetic, friendly or forgiving editors. When authors commit those types of mistakes, it gives the reader a great deal of information about a lack of thoroughness: Do not expect an editor to continue to read a manuscript that fails to provide these essential elements.
“Introduce” your work to explain who cares, which conversation you are joining and what readers will learnOver the years in “meet the editors” sessions, I have heard some colleagues suggest that they often read titles and abstracts only, and with that information they get a very good idea of a paper's chances to make it through the revision process. Needless to say, the editors at BRQ cannot overemphasize the relevance of a good title and abstract; however, I do not recall rejecting a paper without reading the introduction. Thus, here we have our third important hurdle.
Introductions can be tricky and lengthy if authors forget their purpose. As Grant and Pollock (2011) have suggested, a good introduction must answer three questions.
To begin, “who cares?” It is important to explain the problem statement from the very beginning, with specific clarifications on why it is interesting and important from a theoretical or practical point of view. It is the authors’ option if they desire to precede their opening with an anecdote, a quote, or a statistic. Nevertheless, they must ensure that these additions are truly relevant to the topic and that they are far from being clichés. I would not advise authors to start with a definition, or by—say—reminding potential readers of such hollow phrases as “Today's turbulent environment is characterized by economic uncertainty and increased international rivalry,” or “In highly competitive environments, firms must develop products and services that meet their customers’ needs.” Authors have very little time and space to make an impact on editors and readers. We must use them wisely.
Second, “what is the conversation in which authors situate their study, what has it taught us and what has it neglected?” At BRQ we prefer a second paragraph in which authors tell us briefly what other colleagues have shown us theoretically and empirically regarding the topic, and especially what they might have overlooked, which is at the root of a paper's contribution. In this sense it is important to note that a common mistake is to highlight the existence of a mere gap as a sign of relevance. Simply because a gap exists does not mean it deserves attention, let alone publication space. Do not support the relevance of your paper with sentences such as “This has never been done before,” or “This fills in a gap.” Such statements may be accurate; however, the fact that it has not been done before or that a gap is being filled does not justify publication per se. We encourage authors to explain very clearly why the gap they have identified is scientifically relevant. Never take this for granted.
Third, “what will readers learn”? Tell us how you believe you will surprise your colleagues with your claims, perhaps highlighting the type of empirical study you use if it offers some novelty, for instance, facilitating the generalization of results over other studies that have a similar problem statement. Are there any implications for practitioners and policy makers? Building on the literature that inspires a particular paper, and once it is clear why it is important and how the literature is lacking, an additional paragraph could tell readers the exact nature of your contribution. To do so, you should also clearly explain how you will fulfill the paper's stated aims.
Use the literature to develop theory and synthesize your thesisThe primary purpose of a theory section in a typical BRQ paper is to ground propositions (comprising concepts) or hypotheses (involving measures), which is something quite more demanding than merely showing that you are acquainted with your colleagues’ work. Hence, if your paper really embodies an original and relevant thesis, you will have to challenge and extend the existing knowledge, not simply rewrite it as in a mere literature review (Grant and Pollock, 2011).
It is worth noting that a scientific management journal such as BRQ has a different purpose and targets different readers than other outlets inside what we could call “airport literature.” The latter is meant for the general public, especially executives, and has an informational nature. Nevertheless, scientific outlets such BRQ are meant for management scholars and aim at providing managerial theory advance. This is why the analyses we publish cannot be based exclusively on a simple portrayal of what others have done, descriptions of good practices from a strictly empirical point of view or reports on anecdotal evidence. In the absence of theory, such approximations often lead to the identification of specific financial, marketing or organizational practices and tools as the essence of management. Many of these practices and tools may even be presented as different alternatives when they are simply different packaging covering up a lack of novelty in many concepts. By contrast, it may take more time and at times it may even become a tedious effort, but building theory allows us to clarify the nature of causal relations and to identify which elements come first and when. Otherwise, we will not be able to determine the many differences among the recommendations provided by consultants and the analyses that should be performed by scholars. Both are needed; however, at BRQ, we are only interested in the latter. Let us therefore discuss what we mean by “theoretical contribution” (Whetten, 1989) and explain how we would like you to address it (Sparrowe and Mayer, 2011).
A complete theory must contain three essential elements that seldom reflect a linear sequential process despite my sequential description of them:
First, we should ask ourselves which variables should be considered to be part of the explanation. To do so, we must keep in mind the trade-off between comprehensiveness (are all relevant factors included?) and parsimony (should we skip some factors because they add little additional value to our understanding?). A second element of any theory is to clarify how these variables are related to disentangle causality.
These first two issues, deciding the variables and clarifying their relationship, can be regarded as the descriptive part of the theory. As explained above, however, description is not enough. We need a third element in the theory: explanation. Authors must tell readers the underlying psychological, economic, or social dynamics that justify the selection of variables and the novel causal relationships they are proposing. In doing so, they must present valid, sound and relevant arguments.
Valid and sound arguments are those in which, if all the premises are true, the conclusions necessarily follows. As editors of BRQ, we very often find non-valid arguments for which, despite the truth of the premises, not all of the conclusions necessarily follow. On other occasions, one can easily recognize that an argument is unsound when it is based on at least one false premise. Finally, a third very common type of mistaken argument is tautology, which presents assertions that cannot be falsified (Priem and Butler, 2001). For instance, imagine that an author tries to justify a relationship between a particular resource and business competitiveness, as in conventional resource-based view reasoning. If the argument supporting the relationship defines valuable resources such as those that increase efficiency and/or effectiveness, and competitive advantage such as achieving increases in efficiency and/or effectiveness, we would be facing a clear tautology.
In restraint, we recognize the master: quantitative empiricsIn regard to collecting and analyzing data, we could basically say that we need authors to keep it simple and meet high standards of analytical thoroughness (Vázquez, 2017).
As in any journal and despite our rate of desk rejection, the level of scientific contribution of each paper we publish can be disputed. Scientific progress is precisely about discussing each other's work, even after having published it. What cannot be disputable is that the results and implications of BRQ's papers are compelling, which depends on how we help authors develop their empirical analysis. Let me start by giving you a hint of what we desire to see in quantitative analyses, and I will later turn to qualitative studies.
Even if we do not follow hard rules to develop a quantitative empirical section, the search for simplicity and thoroughness at BRQ represents the two sides of the same coin: credibility. Whatever the statistical analysis an author is performing, the sampling, the collection of data and the analysis must be explained clearly so that our editors, reviewers, and readers, can rule out alternative explanations.
The first issue to review when investigating causal relationships is whether the sampling is random, which is at the root of most of the serious problems we find with sample representativeness (Bono and McNamara, 2011). For instance, if we desire to study people's religiousness in a country, we would not place our interviewers at the door of a church on a Sunday morning. It would generate a sample selection bias, which requires specific econometric solutions that are not always easy or feasible to implement. Thus, even if some element of sampling error is unavoidable, we often face papers in which problems with the representativeness of the sample make the authors’ effort pass unnoticed. I am thinking about samples of students who are surveyed as if they represented the general population for consumer behavior purposes, samples made up of innovative firms—and not the general population of firms—that are surveyed to delve into several strategic issues, or samples of firms listed in financial rankings in which the data are truncated because firms with financial ratios below the minimum allowed to join them do not appear in the sample at all. I am obviously not trying to provide a complete list of problems that can be produced by a wrong sampling (see, e.g., Maddala, 1983). The bottom line of these reflections and examples is that sampling is often an underestimated issue in research design that later leads to complicated econometric analyses. In the worst-case scenario, we will not be able to perform the econometrics we must to avoid biased results; in that case, our paper can lose most if not all of its value.
Data collection is a second issue that deserves attention. Except for the papers that belong to the special sections of BRQ (Reviews and Research Agenda, Methodological Insights or Counterintuitive Ideas), the rest use primary or secondary data. Primary data are collected directly through some form of interaction with people or organizations, using interviews, focus groups, surveys and participant observation. Secondary data are not collected by the researchers themselves; they resort to an existing source instead of interacting with people. This is the case of archival data, which generally falls under the following categories: publicly available data sets, on the one hand, and private data sets and records, on the other.
Primary data offer authors the opportunity to demonstrate the greatest research talent, which is something we certainly value at BRQ because it involves thoroughness, determination and considerable effort. One of the toughest tasks all BRQ editors face, in fact, is to reject the work of researchers who have proved to be thorough, full of willpower and devoted workers because their papers show serious problems of common method variance. We often receive submissions based on surveys with multiple predictor and criterion variables completed by a single individual who is supposed to answer questions such as “what level of just-in-time has your firm reached (on a scale from X to Y)?,” or “what is the level of your firm's commitment to the environment (on a scale from X to Y)?” All these research efforts suffer from numerous and very important common method biases, such as social desirability or cognitive dissonances, which compromise the validity of results (Podsakoff et al., 2003). In particular, when participants in a survey are asked about novel issues that reflect their role in society (how efficient they are, how clever they are, how environmentally friendly they are), they will feel compelled to answer when they do not actually have an opinion, misjudge their own views, avoid the type of answers that would make them look bad, or even consciously lie to comply with accepted behavior (Bertrand and Mullainathan, 2001). Matching problem statements with the correct data is crucial, so thinking over them simultaneously from the beginning is not a trivial recommendation.
Regarding secondary data, the main problems in our submissions have to do with the operationalization of variables. Operationalization seeks to define and measure abstract concepts, which are difficult to observe directly in terms of observable and measurable effects (Nedon, 2015). The challenge with proxies is to minimize the “jump” between the proxy and the conceptual variable, and this requires their validity to be justified rigorously with coherent arguments supported by theoretical and empirical literature. Take the case of medicine as an example: phenomena such as health can be operationalized by one or more indicators such as “body mass index” or level of “smoking” (McLaughlin and Hinyard, 2014). When researchers use such proxies, they deliberately omit many factors that may affect a person's health. In fact, we know that an overweight person may be healthy (or at least healthier than a thinner individual), and a particular smoker may live for 90 years in good physical condition. However, what matters is that in aggregate terms individuals with a lower body mass index who do not smoke are more likely to enjoy better health. This is why these indicators can be used as proxies. They do not represent such a big “jump” in aggregate terms and to the extent the data are drawn from publicly available databases, they offer the possibility of replicating methods and verifying results, which represents an increasingly valued byproduct in physics, biology, medicine, and—of course—management (Collins and Tabak, 2014; Marcus, 2014; O’Boyle et al., 2017; Vázquez, 2017).
A third empirical issue that often gets authors into some trouble is the econometrics. To be sure, if we have clear understanding of the stated problem, a random sample with a large database in terms of observations and variables, and a decent construct operationalization, econometric techniques can generally be simple and effective. However, this is not always possible, and in addressing the construction of more sophisticated econometric models, several complications may arise.
To begin with, be aware of incomplete, as well as inappropriate, model specifications. These are two related but different problems that we regularly find in our submissions.
The first one, incomplete model specification, often generates omitted-variables bias; that is, the specification would be incorrect in that it omits an independent variable that is correlated with both the dependent variable and one or more independent variables. This results in over-estimating (upward bias) or under-estimating (downward bias) the effect of one of more other explanatory variables. Suppose, for instance, that an author desires to study the factors that determine workers motivation. She could run a multiple regression to estimate it, including the independent variables related to leadership style, communication, content of work, and promotion opportunities. However, somehow she forgets to include the influence of the reward system. This causes her regression's results to be biased, because workers with highly similar values for the considered variables can have drastically different levels of motivation for different types of reward systems.
An important omitted variable in a model is probably the most important source of endogeneity, but other sources, such as a loop of causality between the dependent and one or more right-hand side variables, are not less common. Assume, for instance, that we want to regress average salaries from the top 100 MBA graduates on variables such as the percentage of economics-based courses, management-based courses, ethics courses, the relative size of the MBA core curriculum, student's previous work experience, academic quality, and some others (Arruñada and Vázquez, 2013). Academic quality can be proxied by the average score obtained in the Graduate Management Admission Test (GMAT). However, just as we expect higher GMAT scores to be associated with higher salaries, we also presume that good students with the best GMAT scores tend to choose the best schools. Therefore, their average quality will also depend on the school's average expected salaries. We should consequently treat the average GMAT as endogenous. In certain situations, we can use the facts of the situation ingeniously to make reverse causation from the right-hand side variable to the dependent variable unlikely. When—as in the case just explained—this is not possible, one of the most frequent recommendations is to use instrumental variables (IVs). In this case, authors must provide compelling arguments to support the assertion that the instrument is related to the endogenous variable, that it is correlated with the dependent variable only through the endogenous variable and that it is not itself endogenous (Bettis et al., 2014). Other solutions to isolating causality, such as matching a treatment group with a control group or performing a quasi-natural experiment, are less frequent because they generally require a radical reconstruction of the paper.
Endogeneity can therefore be the reflection of an incomplete, as well as inappropriate, model specification. Another form of inappropriate model specification that we often see is what has been colloquially termed “kitchen sink” regressions. We obviously expect to see variables in the right hand of the model that have theoretical relevance. However, I would say that 20% of our submissions include models in which one can easily tell that the authors have thrown into the model every predictor variable they had available. The way I see it, this causes two fundamental problems in the papers we receive: on the one hand, it is more difficult to convince readers that you are addressing a relevant management problem instead of a somewhat random collection of relationships that, luckily enough, present many asterisks next to their coefficients. On the other hand, adding some variables arbitrarily simply because they apparently yield significant coefficients can cause many things to go wrong. For instance, one of the most common complications arising from this empirical strategy is sign flipping due to high multicollinearity (although it may certainly occur anyway, even when one does everything correctly). When two variables convey the same or similar information, the computer uses slight differences between them to fit a few outliers. Depending on how these outliers correlate, we can obtain a large positive coefficient on one of the predictors and a large negative coefficient on the other. Sign flipping is always a problem, but it is especially so if one of these variables occurs to be one of the key variables in a hypothesis. To summarize, trial and error in model specification is acceptable if you are searching for potential proxies that reflect the variables with theoretical relevance. Otherwise, be careful.
Finally, there are many problem statements that address causality or change that require panel data or, at least, the introduction of lagged variables (Bono and McNamara, 2011). We can tolerate a certain “tension” between a dynamic objective and a static analysis; however, we strongly advise authors to think it over before addressing with cross-sectional data such problems as the effect of R&D on innovation, how changes in organizational leadership impact a firm's investment patterns, or how stock options influence financial strategy. Panel data offer many advantages and possibilities over cross-section analyses (Baltagi, 2008), but it is important to keep in mind that many causal relations “need time” to be tested because—among other reasons—the influence of one variable over another does not simply occur at once.
Qualitative empirics: a matter of suitability, narrative and credibilityBRQ is a multidisciplinary journal inspired by methodological diversity and open to inductive and deductive approaches. That is what our mission states, and we even mean it: BRQ is interested in qualitative research, particularly in case study analyses. We do not merely say we are interested and then search for highly sophisticated econometrics. If you have understood the philosophy of the previous section, you have now realized that we value interesting questions and simple, rigorous research, so any author can do that with qualitative research methods. The challenge is implementing them appropriately: that is, (1) deciding correctly when they suit the problem statement that an author is trying to address; (2) developing a structure and a narrative that fit the—mostly—inductive nature of qualitative papers; and (3) addressing the methods section with the same analogous credibility concerns and safeguards that one must use when presenting a quantitative analysis.
When do we expect to find a qualitative empirical study? Eisenhardt and Graebner (2007: 26–27) put it simply: these type of studies are better suited to confront questions such as “why” and “how,” rather than others such as “how many” or “how often,” which are the typical goals of quantitative analyses. Regardless of whether the sample is small, no one will deny the academic interest of a single “talking pig” (Siggelkow, 2007). In fact, the field of neurology obtained a great deal of knowledge about the functions of frontal lobes from the behavioral changes of a single construction foreman whose head was perforated by a tamping iron. Nobody asked for a representative sample of the population. The problem is that very few of us have the opportunity to work with “talking pigs” or foremen with localized brain damage, so a descriptive analysis is usually not enough. The discussion of case studies, ethnographic observations or industrial anthropology reports, for instance, must provide a conceptual insight. This is an absolute must.
At this point it is worth noting that qualitative studies can be used either from a deductive or an inductive perspective (Pratt, 2009). Bitektine (2008), for instance, explains how a prospective case study design can be implemented to test theory driven hypotheses. However, social research is organized around two main empirical strategies in which BRQ is most interested: deductive studies with quantitative econometric approaches, and inductive analyses with qualitative research methods. Therefore, if you follow an inductive approach to build and advance theory, the structure and narrative of your paper should follow some general rules.
Despite the multiple data sources for a qualitative project (interviews, archival data, survey data, ethnographies, observations), you should display the raw data to readers and not simply pick a part of it and provide a mere description of the selected data in the text. Instead, you should show a clear chain of evidence reflecting how you moved from your data to your interpretations (Pratt, 2009). Additionally, this effort will help readers to more easily connect their interpretation of facts (which may or may not coincide with yours) with the emergent theorizing (Bansal and Corley, 2012).
In terms of formal structure, note also that the front end of your paper should be shorter than that of a typical quantitative article (Bansal and Corley, 2012). The reason is simple: you are pursuing a different goal. Instead of going from the more general to the more specific, inductive reasoning will move you from specific observations to broader generalizations and theories. Front ends in inductive qualitative papers should therefore establish the ground to hook the reader and identify the gaps in the literature that make your qualitative study worthwhile. In other words, they should contextualize the empirical study and act as a catalyst for the emergent theory that will come afterwards.
Hence, compared to quantitative studies that provide theoretical advancement when they develop their hypotheses at the beginning of the paper, inductive qualitative papers present their most important theoretical contribution in their back end, after they have presented their empirical evidence. As explained above, there are few “talking pigs” that make describing a case study worthy of a paper. That is usually not enough. We need a good discussion that integrates data, the emergent theory, and the literature in which the paper seeks to fill a relevant gap.
Finally, credibility concerns and safeguards for the way in which the data are collected, analyzed and discussed should be a major issue in qualitative research. Authors can decide to report data chronologically or by reflecting a specific pattern, but the following guidelines should be kept in mind:
You should explain your methodological approach from the very beginning of the empirical section, specifying its diverse stages and describing the logical sequence that connects everything: the stated problem you are addressing, the research design and the case selection, the data obtained, and the results and conclusions you will deliver. This action plan (Yin, 1989), which you can even present in a figure or exhibit, reflects the scientific approach of your research and grants more or less compelling arguments to support the coherence and generalization of your results. Once this is done, you should address the reliability and validity issues.
With regard to reliability, include the protocol with the working rules, data collection tools, timing, and procedures for analyzing the evidence (Yin, 1994). The protocol is useful as a working guide (internal reliability), and it allows other researchers to reproduce the study (external reliability). It covers the objective and motivation of the research, the field procedure (arranging for visits and interviews, experiments) and the guide for reporting the case. The information obtained from different sources can be arranged by categories (Miles and Huberman, 1994) and recorded in a single database to guarantee the chain of evidence.
The accuracy of the results (internal validity) and the extent to which they can be generalized (external validity) is, nevertheless, a different matter. Since interviews of one type or another are the main data source in almost every qualitative paper, retrospective sense-making bias is probably the key issue. The criticisms that our editors and reviewers often raise refer to the memories of plausible images that individuals often develop to rationalize what they have witnessed (Weick et al., 2005, p. 409). The point is that the people we interview may not remember accurately what occurred in the past, and they may even fall into self-deception to construct events in a positive light (or at least present them in a manner that is coherent with conventional accepted behavior). Under these circumstances, it is key to capture different dimensions of the same phenomenon, designing the empirical study to include the triangulation of data, researchers and methodologies (Voss et al., 2002).
In terms of the triangulation of data, it is worth noting that data can be collected from secondary sources of a documentary nature, primary sources such as direct observation (on-site visits, direct participation in decision groups) and interviews with numerous informants. Such informants should view the phenomena of interest from diverse perspectives to tap into potential differences in perspectives and thus mitigate biases and lapses (Pratt, 2009). They should thus preferably be situated on different layers of the hierarchical chain, functional areas, geographies, or even outside in different organizations (Eisenhardt and Graebner, 2007). In any case, triangulation may also be implemented by using several researchers to obtain possible relevant points missed by each interviewer, thus containing ‘observer bias’ (Voss et al., 2002). Finally, in addition to the triangulation of data and researchers, authors can also seek to mitigate bias by combining different research methods. Experiments or quasi-experiments can complement interviews (Sartal and Vázquez, 2017). Another option, for instance, is to combine retrospective and real-time cases (Leonard-Barton, 1990).
Discussion (or why does your paper exist?)Every section has its own irreplaceable purpose, but remember that your discussion probably reveals the reason for your paper's existence.
You have already explained your results. It is now time to state what your results mean. This section should demonstrate your ability to think critically about your problem statement so that readers can obtain a deeper understanding compared to what they have already read elsewhere. To do so, you must argue why your analysis is important and how it changes the conversation: either by challenging the extant theory, by offering thought-provoking insights on the topic you are addressing, or both.
Theoretical implications generally come in the form of insights on the underlying patterns, principles, and relationships associated with each major finding (Geletkanycz and Tepper, 2012, 257). They often involve delving into when and why such patterns, principles and relationships emerge, as well as how they allow the refinement of the interpretation of the theory you developed in the theoretical section. It is usually very helpful to refer the reader to a figure or table enhancing the interpretation of the data. For instance, Fernández-Olmos et al. (2016, 131) provide a theory on why the relationship between internationalization and firm performance apparently follows a W-shaped curve in family businesses (FB). They suggest “In a first stage, FBs expand within their home region, following a traditional internationalization pathway (Graves and Thomas, 2008). At this point, the FBs are inexperienced in foreign markets and lack financial resources, managerial capabilities and external networks. (…) In a second stage…”
Sometimes part of the contribution does not intend to be theoretical but simply focuses on illuminating a specific topic that is the subject of broad and current interest (Bergh, 2008). For example, in attempting to determine how marketers can foster the relationship with a brand through virtual communities, Gutiérrez-Cillán et al. (2017, 269) estimate a model of relational efficacy for a firm-managed Facebook brand page (FBP). They suggest that “(…) the particular case analyzed does not seem to indicate that entertainment value motivates engagement. It may thus be concluded that perceived utility encourages active participation, whereas entertainment produces receptive users but not active participants. One feasible explanation for this result could be related to the content of the posts analyzed…”
Lastly, remember to highlight unsupported hypotheses and theoretical arguments, but do not merely warn readers about them: explain why you believe they might have occurred. Is it an empirical issue related to your sample or analysis, or do you find alternative arguments that the literature has underestimated? A simple example is Carbonell and Rodríguez-Escudero (2016, 36): “Counter to our expectations, the study's results reveal that process-based rewards are detrimental to new product quality. A plausible explanation for this effect lies…” Very often, it is more interesting to see a paper in which thought-provoking ideas are presented about why an author has not found what he or she expected than one that simply confirms that “we have verified all the proposed hypotheses.” In any case, the BRQ editors dislike papers in which authors present post hoc hypotheses based on empirical findings, as if they had been developed a priori. This has actually been a concern for many leading journals for many years, and it still is (Locke, 2007; Leung, 2011).
To summarize, in general terms, authors should use the discussion section to highlight the relevance of their findings, and the strategy to do so is to fill the existing gaps in the field, while simultaneously revealing new relevant gaps that had not been previously exposed. If we address this challenge correctly, the theoretical conversation with BRQ readers will be rich enough to—hopefully—help to shape the direction of the research in your field.
Play the final solo to end the song: conclude your paperA paper's conclusion is similar to the final phrase in a blues song. It is not that you cannot tell the song has finished, but its absence makes the listener feel that the piece is incomplete. You must use this section to deliver leads that other colleagues, practitioners and policy makers can follow in the future. Under this philosophy, and even if you have in fact answered many of the key issues in the discussion section, you should try to address the following questions as succinctly as possible:
First tell us what your research has shown. Give us a very brief description of your main results as a mere aid to memory; however, instead of repeating the same sentences you have used in the abstract or introduction, restate your insights and arguments in a way that ties them all together. Give us a “complete thought”, not a collection of random and vaguely related ideas. For instance, “This paper examines how real activities manipulations by bond issuers affect their yield spread. Extending the findings of Cohen and Zarowin (2010), we find that similar to equity issuers…” (Mellado-Cid et al., 2017).
After you have given busy readers a first punch to quickly remember what you have done, explain how your analysis has added to conventional wisdom in your field, and tie your arguments to the research you have described in the literature review. You can give a preliminary hint, and then address it more broadly, as in Sanchez-Marin et al., 2017, 237): “this study contributes to the literature in several ways. First, it extends our knowledge about the real effectiveness of SOP as a mechanism for adjusting and aligning CEO compensation (Krause et al., 2014; Mangen and Magnan, 2012). Second, it shows how important the internal governance mechanisms, such as the board of directors and the ownership structure, of companies are, as elements exercising an indirect influence on the effectiveness of SOP in terms of CEO compensation arrangements (Stathopoulos and Voulgaris, 2016). And third, it highlights the peculiarities of SOP in a particular context of Spanish listed companies.”
Third, acknowledge the shortcomings of your work. This entails highlighting the technical problems you faced in your data and methods sections, as well as the unanswered questions that your research has left behind. Regarding the technical problems, report them with an honest but positive attitude. It is not a question of enumerating all the technical deficiencies of your work, which will only create doubt in editors and readers about the credibility of your effort. You must do whatever you can to compensate them. For example, (Diéguez-Soto et al., 2017, 161), “we do not have the information available to run a longitudinal analysis but, meanwhile, we have tried to show literature regarding several measures of robustness (Michiels et al., 2013), which…” Thus, acknowledge the limitations, but at the same time put them in context and explain why you still believe in your claims. If editors perceive that you do not trust your results, you should not ask them to have more faith in your work than you do. A similar but different situation is caused by limitations regarding unanswered questions, which authors can easily address to inspire other colleagues (Sanchez-Marin et al., 2017, 237): “this study only analyzes the indirect effects of the board of directors and ownership structure in the relationship between the SOP and CEO compensation. Future research should consider other internal and external mechanisms of governance that might interact with SOP…”
Finally, although BRQ is a scientific journal whose main target is university scholars, our editorial policy (Vázquez, 2015) also includes a concern to enhance our impact on economic and social players outside the academic world (Vermeulen, 2005). We therefore value authors’ reflection in their conclusions on how their results can be of any use in the real world, either for management practitioners or for workers, unions, policy makers, or NGOs. Again, we are especially interested here in insights from unexpected results or that at least contain some degree of surprise, as in Ruiz-Jiménez and Fuentes-Fuentes (2016): “One unexpected finding of our analysis was that the results show no direct effect of gender diversity in the top management team on either product or process innovation. A possible explanation…”
A final thoughtI would say that the generation of new knowledge is based on two key issues that are not always easy to balance: the freedom to create new ideas and explore them at will, and the need to follow certain methodological standards that allow readers to distinguish science from anecdotes or mere beliefs.
Some years ago (Vázquez, 2015), BRQ expressed its choice of editorial policy to make compatible both of the above issues. Our goal was to emphasize that our editors would not merely be paper-pushers (forwarding the reports of our referees) or try to re-write papers in the way they would have written them themselves. Once an article's scientific originality and potential impact have been verified, the feedback for authors must be largely editorial, underpin credibility and allow for replicability (Bettis et al., 2016). Thus, following the example of journals in other disciplines (New England Journal of Medicine, Science, Nature) we hoped—as has finally occurred—that this strategy would save authors and BRQ considerable time and effort, while at the same time maximizing the chances of achieving the proper balance between creative freedom and methodological standards.
This editorial is just a step forward; a specific path to guide in a particular journal the balance between an author's right to think freely for the sake of thinking, and the heritage we have received from many centuries of thought on scientific progress—often forgotten—starting with Aristotle, Ibn Sina (Avicenna), Bacon, Descartes, Popper or Khun. In any case, if you feel these guidelines are too limiting, remember we will assess the value of your ideas for their own intrinsic value. After all, as Borges would put it, our nothingness as authors and editors differs little. It is a trivial and arbitrary circumstance under which I am the reader of your papers and you their author.