2. RESEARCH METHODS
2.3. POST-POSITIVIST RESEARCH
2.3.1. Case study research
A case study method involves studying a phenomenon in a real-life situation, often used when research and theory are at an early stage of investigation, and informative descriptions of the phenomenon are required (Bonoma 1985). They are useful for questions of “how?” and “why?”, where there is limited control over the environment, and the focus is on events at a particular point in time, to identify patterns or features (Yin 1994:6). In particular, information systems research is concerned with exploring and explaining information systems in an organisational context, concentrating on the social implications rather than technological issues, so a case study is an obvious choice (Benbasat et al. 1987).
Case study is used to contribute to knowledge of individual, group, organisational and other phenomena, and can take a number of forms: theory seeking, theory testing, story telling, picture drawing or evaluative case studies (Bassey 1999). It is commonly used in the social sciences and in education, as well as information systems.
Case studies involve multiple sources of data, so results are often found from triangulation of data. For example the study by Kaplan and Duchon (1988) combined qualitative and quantitative methods in a case study investigation.
In carrying out case study research a general strategy for analysis is suggested at the outset. This might be:
• Following a theoretical proposition that led to the case study, to answer how and why questions and lead to alternative explanation to be discussed;
• Thinking about rival explanations, resulting from other influences, and discussing these;
• Developing a case description, which might highlight relevant causal links to discuss (Yin 1994:21).
Selecting a case for investigation depends upon the type of investigation, paradigmatic viewpoint of the researcher and availability of subjects (Saunders et al. 2003:139), or a case study might be used in combination with other methods. Although selecting a confined case to study may limit the generalisability of any findings, post-positivists are more interested in forming theories, so the exploratory nature of case study makes it a suitable interpretive method. Findings from a case study are bounded within a specific context, for example the paper by Markus (1983) demonstrates the importance of context on an investigation. But putting a boundary around the context does not preclude proposing a form a generalisation from the findings, such as:
“In some cases it may be found that…” (Bassey 1999:12)
The approach adopted for case study research can be just as rigorous as other methods, because it includes stages of identifying the questions to be asked, stating any propositions, considering units of analysis to provide boundaries, linking the data to the propositions and identifying criteria by which the findings will be evaluated (Yin 1994:20). There are four strategies for case study: single case or multiple case and holistic case or embedded case. Whereas multiple cases are used to establish whether the findings of one case are relevant to other cases, single case research is used where this particular case has some unique quality about it, which is of interest to the wider community as well as the researcher. A case study is holistic if it is looking at the organisation as a whole, but is embedded if it is concerned with a particular sub section of the organisation (Saunders et al. 2003:140).
Case studies may be subjected to the same criteria as other forms of research for determining the quality of the end result, validity and reliability. Indeed four tests for evaluation are advocated by Yin (1994): construct validity, internal validity, external validity and reliability. Construct validity has been criticised because it is difficult in case study to establish criteria for testing results, and data collection may be subjective. Construct validity can only be applied if it is possible to select specific changes to measure, and to demonstrate that these measures do reflect any changes observed. Internal validity is shown if there is “a coherent and illuminating description of and perspective on a situation that is consistent with detailed study of that situation” (Schofield 2000:71). In case study there is a need to take care with
making inferences, as there may not be sufficient data to state facts, so cause and effect cannot be stated. Yin suggests eliminating all rival explanations, in order to be left with those inferences remaining. Internal validity indicates the extent to which the interpretation of the researcher is consistent with the findings (Yin 1994:34).
External validity is more widely discussed in terms of whether the findings are generalisable to a wider population, from the particular case being studied. Critics say that a single case cannot be generalised, but critics are often relying on analytical (or statistical) generalisation, where particular results generalised to a broader theory. Bassey (1999), on the other hand, talks about “fuzzy generalisations”, which gives an outcome along the lines “it is possible, it is likely or unlikely that findings from a particular case will be found in similar situations elsewhere”. Reliability is a measure of whether later studies would generate the same results as the case study, showing that there are minimal errors or bias in the methods used by the researcher.
Klein and Myers (1999) advocated case study as a form of interpretive research and suggested a set of principles for evaluating a case study investigation:
1. The fundamental principle of the hermeneutic cycle – switching between an holistic and partial view;
2. The principle of contextualisation – the importance of the context of the situation under investigation;
3. The principle of interaction between the researchers and the subjects – social construction of data between researchers and subjects;
4. The principle of abstraction and generalisation – interpretation of the details of the research to general concepts;
5. The principle of dialogical reasoning – awareness of contradiction between theory and actual findings;
6. The principle of multiple interpretations – sensitivity to different interpretations of data;
7. The principle of suspicion – awareness of possible bias or distortions in the subjects’ narratives.
On closer inspection, these do correlate with the four tests for evaluation of case study, suggested by Yin (1994).
Critics of case study research point to possible bias as researchers interpret the data in their own way, and consider this research as only an exploratory part of other research methods, such as field studies. But within limited domains, and with structured interviews, researchers can gather much reliable data pertinent to specific research objectives. Even though case studies may not be generalisable in a scientific sense, they can be used to add to experience of a domain, and improve our understanding of the context under investigation (Stake 2000).