Article A5.2 Doctoral Summer School 2008: Models in Research D. P. Dash Xavier Institute of Management, Bhubaneswar, INDIA dpdash[at]ximb.ac.in Amar Patnaik Doctoral Scholar, Xavier Institute of Management, Bhubaneswar, INDIA amar_patnaik[at]yahoo.com 1. Doctoral Summer School: A Model for Research Education Xavier Institute of Management, Bhubaneswar (XIMB), India conducted a week-long Doctoral Summer School (DSS 2008) on “models in research,” during June 7-12, 2008. The event was part of XIMB’s broader project of improving the quality of research education in the country and producing competent and innovative researchers in the field of management. As in the previous years, the doctoral summer school (DSS) was targeted at doctoral scholars in management and related fields who are in the proposal and pre-proposal stages of their research. This year’s theme was to help prospective doctoral scholars understand the use of models in research and prepare for innovative designs of model-based research in management and related fields, with some emphasis on quantitative modelling. In particular, DSS 2008 promised the following learning outcomes to the participants: (a) awareness of some of the key issues and challenges of doing management research, (b) examination of research as a process, guided by alternative standpoints, (c) appreciation of the diverse roles played by models in the research process, (d) approaches and methods followed in model-based research, and (e) various practical elements important in doctoral research. The topics discussed in various sessions spanned a wide range, covering areas such as orientations to research, role of models in research, characteristics of management research, models in management research, academic writing and peer review, managing doctoral projects, and research beyond science. The emphasis was on the use of models in different areas of management research. The event attracted participants from different parts of India (see Appendix), mostly full-time doctoral scholars. Among the participants, there were also academicians, government employees, and other professionals intending to start their doctoral studies. They were introduced to the world of research with an emphasis on the issues of doing research in an applied field such as management. 2. Everyday Learning and Scientific Learning: Role of Models In the opening session, the participants drew pictures of themselves as individual researchers, representing their images of the research activity they were about to begin. Explaining what they had drawn, the participants revealed diverse images of their thinking, though there were several common perceptions such as a long journey with several obstacles, twists and turns, and paths that might lead to blind alleys, confusion about the course to be taken at a fork in the road, focus, tenacity, open-mindedness, creativity, flexibility, intuition, the ability to turn failures into opportunities, to learn and grow, and so forth. To some, doing research seemed a delicate balancing act among family demands, hobbies and other interests, and doctoral study requirements. Further discussions on this indicated that the themes and images of research which had emerged in the exercise were popular opinions, although they represent research only partially. For instance, not all researchers start from a state of confusion as some of the images implied; some researchers also start with a great deal of clarity on what they intend to establish when they begin their research. Discussions moved on to the process of learning, as it occurs in different contexts, including learning in everyday life and scientific learning through research. David A. Kolb’s model of experiential learning (learning by doing) provided a way to describe the process of learning in everyday life (Kenney & Reid, 1986, chap. 5). The model involves a cycle of four elements: concrete experience (feeling), reflective observation (watching), abstract conceptualisation (thinking) and active experimentation (doing). The model suggests how individuals develop their learning styles by habitually focusing on some parts of the cycle rather than utilising the entire cycle. The work of the theoretical biologist, Robert Rosen (1934-1998) was discussed next, according to whom “modelling is the essence of science.” In Rosen’s framework that describes the process of scientific learning, the notion of modelling relation assumes an important place. It suggests that scientific learning happens through a scientific model, which is a formal system that represents some natural system, such that (a) observations on the natural system are encoded in the formal system and (b) implications of the formal system can be decoded into observable conditions in the natural system. It was suggested that Kolb’s and Rosen’s ideas were analogous -- both seeking to capture the basis of how systematic learning can happen, whether at the level of the individual or at the level of a scientific community. Different kinds of model are used in research, for example: (a) iconic or scale models (the phenomenon to be studied captured in a simplified form, (b) analogical models (models based on analogy, e.g., the “digesting duck” and a hydraulic model an economic system), and (c) symbolic models (models represented through easily manipulable symbols). Common types of symbolic model are theoretical models (expressed in mathematical form) and data models (expressed in statistical form). Management research seeks both explanatory and actionable knowledge. Therefore, explanatory models, which are models of some pre-existing phenomena, are as important as prescriptive models, which are models for bringing about some nonexistent state of affairs. Both of these pose challenges of their own. Explanatory models may not be sufficiently representative (i.e., they involve approximations/idealisations) or they may not be sufficiently general (i.e., transferable to new contexts). Prescriptive models may be used in unexpected ways with unexpected results. Both kinds of model can have unintended side effects. 3. Model-Based Research in Management 3.1. Models in Context In management research, it is particularly important to specify the contextual space in which a model is developed, because the variety of contexts involved in management tends to be quite high, with each context associated with a large number of influencing variables. A well-specified contextual space indicates the specific assumptions and conditions under which a model is developed. A model developed under certain conditions may not be readily applied under some other conditions. The experience of Apple Computers in developing new products can be a case in point (Balachandra & Friar, 1999). The product development process for Apple 3 computer followed a model not appropriate for the context relevant to that product. Similarly, models developed for understanding (or managing) the so-called developed economies may not hold good for the so-called emerging economies. One way to start model-based research would be to identify a puzzle, calling into question some of the assumptions of a well-known model that is available in the academic literature. Another way would be to identify a context not anticipated in the academic literature, which nevertheless arises in practice. To do either, an extensive survey of the literature is required, with a focus on culling out the missing links or gaps. The process was illustrated using an example of research in international business (Nayak, 2007). Despite a rigid regulatory regime, British American Tobacco flourished in India during the period 1906-2004, by investing more in businesses other than its main line of business businesses, although such complementary investments were not a common practice during much of that period. The international business literature at that time was based mostly on the context of developed economies and it did not capture the importance of complementary investment. This was the gap in the literature that led to the formulation of the research hypothesis in this case. 3.2. Modelling Human Action Management being an applied field, studying human action becomes important. But, action turns out to be quite an elusive thing to study. Although, sometimes, it might be studied in term of stimulus and response, the variations are often too large to be contained within this framework. In this context, Vygotsky’s activity theory was discussed. The theory introduces a third element, apart from stimulus and response, an element that captures the context of a particular human activity. In the conceptual framework of this theory, the three elements are subject (stimulus), object (response), and tools (mediating artefacts). These three elements seek to capture the common form of all human activities, thus constituting a basic model of human activity. Of course, the model has been extended in various ways and applied in different research contexts, including doctoral research (see, for example, Vakkayil, 2008). 3.3. Understanding Variations Through Statistics Statistics provides a particular way to speak of variations at some level of description in terms of relatively more stable properties at a different level of description, by referring to random variables, samples, populations, and so forth. Random variables are variables whose values cannot be predicted accurately, as they assume values at random, but usually following some constraints, expressed statistically in terms of probability distributions. Different kinds of probability distribution were discussed, including the ubiquitous normal distribution. The idea of hypothesis testing was also discussed at length, including null (or status quo) hypothesis, alternative (or research) hypothesis, level of significance, and Type I and Type II errors. 3.4. Multivariate Models In many areas of management research (e.g., market research), the variables of interest often depend on a large number of other (explanatory) variables. These explanatory variables may not be entirely independent of each other. In such cases, based on a set of research hypotheses, data on selected variables are collected from an appropriate sample of respondents, using well-designed research instruments. A multivariate factor analysis then helps in identifying the independent factors underlying these explanatory variables. The independent factors are arrived at by grouping together similar variables. In another application of multivariate factor analysis, a set of hypothesised factors can be tested as to how well they fit the given data--this is called confirmatory factor analysis. Various software packages are available to do factor analysis; SPSS is one such software package. 3.5. Modelling Time-Series Data In many areas of management research, the main research task involves understanding secondary data that are available as time-series data, cross-section data, or panel data. Secondary data must be “prepared” suitably before being analysed. Such preparation often deals with problems of specification bias (excluded variables and incorrect functional form), multicollinearity, heteroscedasticity, auto-correlation, and so forth. Specifically, while dealing with time-series data, if the generative mechanism underlying a time series can be understood and specified in a mathematical/statistical form, it can be used as a sound basis for forecasting, assuming the stability of the underlying generative mechanism. This stability condition, popularly referred to as the stationarity property of the underlying mechanism, requires that its mean, variance, and autocovariance (at various lags) are time independent. Contemporary research interest is focused more on nonstationary time series, which has time-varying mean or variance, or both (Phillips, 1995). Besides stationarity, other important issues with time-series include cointegration and causality. EViews is a software package for doing time-series analysis. (The interested reader may also look up an earlier seminar report on time-series analysis [Report R5.10] published in this issue of Research World: Doing Research With Time-Series Data.) 3.6. Modelling Panel Data The term panel data refers to multi-dimensional data (e.g., some macroeconomic indicator of 20 states collected over 10 years--here the 20 states constitute the “cross-section” and the 10-years’ data for each state constitute a “time series”). The use of panel data brings in useful contextual information into a study. Analysis of panel data involves comparisons among the “pooled” data (i.e., all the data-points taken as a whole) and various components of the pooled data set. Different kinds of analytic model are used, such as constant coefficient model, fixed effects model, random effects model, and so forth (Yaffee, 2005). The EViews software package is also used for doing panel-data analysis. 3.7. Structural Equation Modelling Structural equation modelling (SEM) involves a system of simultaneous equations, with some endogenous and some exogenous variables, where multicollinearity is believed to exist among some variables. Because of multicollinearity, the ordinary least square (OLS) method for estimating parameters is not applicable. The Amos software package can be used to specify, estimate, and assess a model using a “path diagram” that shows hypothesised relationships among variables. 3.8. Purpose Behind Models Researchers build models with different purposes in mind. Such purposes were discussed using examples from research in finance. At least, four kinds of models can be found in this field, namely abstract, theoretical, applied, and empirical. An abstract model maybe driven purely by a researcher’s intellectual pursuits, sometimes even for fun. A purely abstract model may not seek to explain any observed behaviour. But, if it does, it becomes an example of a theoretical model. A classical example of theoretical model in finance is the capital asset pricing model (CAPM), which is derived from the basic axiom of utility maximisation by individuals. Another example would be Spence’s job-market signalling model (Spence, 2001). An applied model is usually a specific model developed for a specific application. Altman’s Z-score model is an example of applied model which assesses the financial health of a company and predicts the probability of its bankruptcy. An empirical model is developed to study a specific event. How do prices of Indian stock market react when firms announce a dividend? Do Indian stock prices follow a random walk? These questions call for empirical modelling. All things considered, in general, simpler models are preferred to complex ones. As a matter of fact, many important results in the world of research have been quite simple in nature. Of course, simple results are easier to communicate, understand, and apply; this might be a reason for their greater appeal. 4. Research Writing and Peer Review Writing is a common mode of communication among researchers. Research writing takes many forms, each with its specific requirements. Participating in the community of researchers, one often reads other’s writings and provides “peer reviews.” A mock peer-review process was conducted during DSS 2008. An anonymised article was given to the participants and they were required to write their own peer reviews of the same. These reviews were discussed and compared with the five actual peer reviews received on that article. This led to a discussion on the do’s and don’ts of peer reviewing (see, for example, Lee, 1995). The reviewer must be honest and sincere in writing the review. It is often a good idea for the reviewer to reveal one’s own background, expertise, and theoretical/philosophical standpoint if these have a bearing on the review being written. Some reviewers start a review with a summary of the material under review. Overall, it is important for the reviewer to be kind, encouraging, and constructive in one’s review; any advice given should be as specific and actionable as possible. 5. Research Beyond Science I never do a painting like a work of art. It is always a search, I’m always seeking and there is a logical connection throughout that search. This is why I number them [the works]. I number and date them. Maybe one day someone will thank me for it -- Pablo Picasso (1881-1973), quote found on the Web. Research, that is systematic and open inquiry, happens even outside the realm of science. Discussions on the nature of inquiry (or research) in creative disciplines, such as art and design, brought out some of the broader issues in research, which might be relevant for any researcher working in any field. Balachandra, R., & Friar, J. H. (1999). Managing new product development processes the right way. Information Knowledge Systems Management, 1(1): 33-43. Kenney, J., & Reid, M., (1986). Training interventions. London: IPM. (Chapter 5, “Learning and training,” pp. 115-148). Lee, A. S. (1995). Reviewing a manuscript for publication. Journal of Operations Management, 13(1), 87-92. Retrieved July 14, 2008, from http://www.people.vcu.edu/~aslee/referee.htm Nayak, A. K. J. R. (2007). Does direct investment in complementary businesses make business sense to foreign companies in an emerging economy? Case of British American Tobacco in India, 1906-2004. Global Business Review, 8(2), 189-204. Phillips, P. C. B. (1995). Nonstationarity time series and cointegration. Journal of Applied Econometrics, 10(1), 87-94. Retrieved July 11, 2008, from http://korora.econ.yale.edu/phillips/pubs/art/a108.pdf Spence, A. M. (2001, December 8). Signaling in retrospect and the informational structure of markets [Nobel Prize Lecture]. Retrieved July 14, from http://nobelprize.org/nobel_prizes/economics/laureates/2001/spence-lecture.pdf Vakkayil, J. D. (2008). Boundary crossing: A study on the dynamics of inter-activity interactions in an Indian software development firm. Unpublished doctoral dissertation, Xavier Institute of Management, Bhubaneswar, India. Yaffee, R. A. (2005). A primer for panel data analysis. Retrieved July 11, 2008, from http://www.nyu.edu/its/statistics/Docs/pda.pdf
Xavier Institute of Management, Xavier Square, Bhubaneswar 751013, India Research World (ISSN 0974-2379) http://www1.ximb.ac.in/RW.nsf/pages/Home |
