Sunday, June 2, 2013

The Positivist Approach to Empirical Research - Dan Remenyi - Chapter Summary

The Central Role of Observation

The essence of modern ‘knowledge is that it is derived from observations made on the world.  All our research derives from, and ultimately refers back, directly or indirectly to our observations, our experiences and our measurements.  Things that cannot be observed either directly or indirectly through their effects of consequences are generally regarded as being outside the domain of science and thus not amenable to research.

There are several ways in which observations can be made of the world around us, including passive observations, observations of the consequences of uncontrolled interventions, or observations of the results of deliberate interventions.  These three types of observation are not mutually exclusive and a single research project could include any or all of these approaches.

Passive Observation is the method most frequently used in business and management research when the researcher is unable to conduct an experiment and has to rely on evidence that already exists.  The researcher collects evidence in the form of interviews, written reports, questionnaires, artefacts and so on.

2.   Uncontrolled Intervention   involves observing the effect of a major change in a driving variable  on one or more dependent variables.

3. Observation of Deliberate Intervention also involves observing the effect of a major change in a driving variable on one or more dependent variables, but in this case the researcher deliberately brings about the change in the driving variable.

Combinations of Approaches

In a research project it is possible to employ more than one of the above three approaches.  This is especially true in business and management studies at the doctoral level where the researcher might begin with a study based on passive observation, follow this by studying the effects of an uncontrolled intervention, and then try a deliberate intervention to see if it confirms the deductions made from the earlier studies.

Literature Review

Once the literature has been fully and critically reviewed, the researcher should be able to provide a narrative description of the current understanding in the field of study, including at least one area where there is incomplete knowledge which could be further investigated.

Assessment of the Established Theory

There are agreed facts and established (if sometimes conflicting) theories, and the researcher may have deduced a new theory by analysing and then sythesising ideas and concepts, already present in the literature of the discipline.

In business and management  studies established and accepted theories are unlikely to be available.  There is little grand theory and few, if any, authors who have developed seminal theories of the kind offered by Freud, Marx or Keynes.  As a result the business and management researcher needs to generate a grounded theory, a concept developed by Glaser and Strauss(1967), which they define as: ‘ an inductive, theory discovery methodology that allows the researcher to develop a theoretical account of the general features of a topic while simultaneously grounding the account in empirical observations or evidence.’              

The grounded  theory methodology normally relies heavily on the use of in-depth interviews with experts in the field of study for the collection of evidence that will be used in an inductive way to assist in the theory generation.

Theoretical Conjecture

The primary aim of the researcher developing a grounded theory is to describe the phenomena of interest accurately.  The grounded theory emerges through the process of concept discovery, within which the researcher develops abstract concepts and categories from the evidence.   It is important to note that in this approach to research, concepts and theories are regarded simply as more or less useful and not as more or less true or valid.

Once the grounded theory has been developed, the researcher in business and management studies is in a position to  make a theoretical conjecture or thesis, but there is no structured methodology for doing this.  Rather, this aspect of research or scientific study can be regarded as an art that relies almost entirely on the imagination and creative abilities of the researcher.

It is not acceptable for the researcher completely to invent a theoretical conjecture, and he or she needs to be able to defend how the concepts and ideas which exist in the literature and which arose from the evidence lead to the theoretical conjecture.

The following is an example of a theoretical conjecture developed through the use of the grounded theory approach for a doctoral dissertation in information systems management(Remenyi,1990a).

Strategic Information Systems (SIS) occur as a result of pressure or opportunities directly related to industry drivers.  The firm’s response to this pressure or opportunity is influenced by its strategy and by its critical success factors (CFS), and these issues determine the formulation of the SIS.  The decision to attempt to take advantage of SIS is made with little attention to detail concerning cost-justification and vendor selection, but with more attention to communicating with the staff, training appropriate people and setting up support facilities.

An important change occurs at this stage of the research.  Whereas the formulation of the research problem began as a description of the known facts from which a narrative theory was developed, this narrative will now be used paradigmatically.  That is to say, it will provide a set of logical conjectures as the basis on which to predict and explain observations.  

Whereas in the physical sciences the theoretical conjecture will frequently be expressed as a formula or as a series of simple propositions, in business and management research the theory or thesis will often be reduced to a diagram for the purpose of clarification

Hypotheses or Empirical Generalisations

When the new theoretical conjecture or paradigm has been developed the next step is to use it to derive hypotheses or empirical generalisations.  If the theory or thesis has been derived from a review of the literature without recourse to grounded theory then the term hypothesis is more generally used, while if the grounded theory approach was employed then the term empirical generalisation is usually more appropriate.  

Strategic Information Systems occur as a result of pressure or opportunities directly related to industry drivers.
The firm’s response to this pressure or opportunity is influenced by its strategy and by its CSFs and these issues determine the formulation of the SIS.

The decision to attempt to take advantage of SIS is made with little attention to detail concerning cost-justification and supplier selection.

More attention is given to communicating with the staff, training appropriate people and setting up support facilities.

Note that from the theoretical conjecture provided on the previous page, four hypotheses or empirical generalisations were developed.  This is regarded as a reasonable number of empirical generalisations with which to work at a doctoral level.

Some research studies stop at the stage of theoretical conjecture, perhaps having developed some hypotheses or empirical generalisations.  It is argued that at this stage a contribution has already been made to knowledge and this may well be so.  Certainly achieving a theoretical conjecture and producing empirical generalisation would normally be more than adequate for a masters degree and depending on the subject area it might even be enough for a doctorate.

However, if the newly discovered knowledge is to be more widely useful, then it is usually necessary to progress to a further stage in the research in which the new thesis is tested against a larger sample population.  In most circumstances this additional step would be required for a doctorate degree.

Measuring Instrument
In business and management studies the collection of evidence for the purposes of testing empirical generalisations frequently, although not by any means always, requires the preparation of a questionnaire which is sometimes referred to as a measuring instrument (Oppenheim,1966).  The researcher needs to ensure that the questionnaire is unambiguous, reliable and valid for the purpose for which it is to be used.

Testing and Analysis
To establish super-variables that simplify the interpretation of the evidence.

Confirmation and Refinement of  Theory
Testing and analysing the evidence may lead the researcher to confirm or reject the theoretical conjecture or to develop a fuller or more refined theory.
In addition, provided that the sample was representative of the broader population and the measuring instrument was valid and reliable, it should be possible, at least to some extent to generalise the theory.

Research is almost always too complex for each step to follow from the previous step in the planned or desired way the first time it is attempted, and sometimes a problem that arises in one step will only become apparent when a later step is in progress.  Thus the researcher may have to retrace his or her steps several time during a major project such as a doctorate degree.

Uncontrolled Interventions
The second research category or approach referred to as uncontrolled intervention arises as a result of a change that has been brought about by an agency external to and independent of the researcher.  This approach to research comes close to what could be regarded as a field experiment.

Literature Review
In this case however, the literature review might be used to identify and consider likely events which could be regarded as uncontrolled interventions and which thus could provide the focus for the research.

Evidence Collection Design
Evidence needs to be collected relating to the situation both before and after the uncontrolled intervention and, where possible, for a control group that is unaffected by the intervention.

Primary and Control Evidence
The use of control evidence is typical of an experimental design that will be discussed below under research based on deliberate interventions.

Testing and Analysis
However univariate stastistical analysis is more likely to be the central analytical tool in the case of this category of research.

Confirmation and Refined Theory
Thus this approach is much more focused in the type of research problem or question than research based on passive observation.

Deliberate Intervention
There are circumstances where some degree of experimentation in an organisation is possible and the controlled introduction of a computer system to a part of the firm on a trial basis might be just such an example.

Experimental Design
The design of experiments is an extensive and involved topic of considerable importance to the researcher.  The key issues involved are:
the form that the intervention will take.
What evidence will be collected before, during and after the intervention;
What sort of control group(s) is required;
How a sample may be chosen to represent the whole population;
Who will make up the experimental team;
Who in the organisation will provide the interface for the purposes of evidence collection.


Measuring the variables Ex-Ante
If an organisation is to be studied by way of an experiment, the first stage is to collect evidence concerning the performance of the organisation before the deliberate intervention occurs.  This evidence might be based on accounting or statical data or on opinion surveys.  This is the base evidence that will be compared to the corresponding evidence after the intervention has been completed.

If, as is desirable, there is a separate control group to which the intervention is not applied, the evidence for the control group should clearly be collected at the same time as the evidence for the intervention group.  The importance of a control group, which should match the intervention group in all of the relevant parameters, is that if the change is not in fact due to the intervention, it should be observed in the control group.  Conversely, it it is due to the intervention it should not be observed in the control group.

Testing and Analysis
It is worth pointing out that in this context modelling refers to the creation of simplified abstractions of reality which capture the critical or key features of the situation.

Relationship Between the three Categories
Passive observation is the most general category.  Here there may not be a prior theory and therefore the researcher may have to develop a grounded theory by induction from a priori observations.  The development of theories of gravitation, from Keppler through to Newton and eventually Einstein represents the most important example of this in the natural sciences.

These approaches are attempts to create knowledge in such a way that theory may be developed with some sort of predictive power.  Category one research actually has holistic undertones, while categories two and three are clearly reductionistic.



No comments:

Post a Comment