This paper serves as a road map for conducting case studies to build theory (almost from scratches). It starts from a general background of study methods and shift to a detailed introduction of steps to go through case studies. The introductions are complete (from getting started to reaching closure), holistic and critical (frequently compared with some other study methods), and supported by evidence (examples of some case studies). Further, the paper also raises the issues of strengths/weaknesses analysis, applicability, evaluation, etc.
[Steps to undertake case study]
It is important to have a research question in mind – so that all the following steps can be scoped down. Sometimes information about the constructs is also important. “Finally and most importantly, theory-building research is begun as close as possible to the ideal of no theory under consideration and no hypotheses to test.”
For building theory from case studies, the selecting process relies mostly on theoretical sampling (as opposed to pure statistical, random sampling). The basis is the research question and the perceived constructs.
Crafting instruments and protocols
This is about data collection methods. For case studies, one can use qualitative, quantitative, or both approaches. A typical method is obtaining insight from hard data (insight which is not salient to the researchers) and try to look deeper and understand such insight by analyzing the soft data.
Entering the field
In case study, it is legitimate or even encouraged to ‘mix’ data collection together with data analysis. Some key points to note: “… write down whatever impressions occur, […] to react rather tha to sift out what may seem important.”. Researchers can benefit from flexible data collections – being able to add or change according to any upcoming data or thoughts.
Analyzing within-case data
This is about looking at the data within an individual case (as opposed to cross-case), “Within-case analysis typically involves detailed case study write-ups for each site. These write-ups are often simply pure descriptions, but they are central to the generation of insight because they help researchers to cope early in the analysis process with the often enormous volume of data.”
Searching for cross-case patterns
There are at least three tactics toward cross-case patterns:
- Select categories or dimensions, and then to look for within-group similarities coupled with intergroup differences;
- Select pairs of cases and then to list the similarities and differences between each pair;
- Divide the data by data source (questionnaire, interview, observation, archival data, etc.)
There are several steps to follow in order to shape hypohteses
- Sharpening constructs (refining definition + building evidences)
- Verifying emergent relationships between constructs fit with the evidence in each case (confirming cases enhance confidence; disconfirming cases refine and extend the theory)
One should look for the contradiction and similarities between literature and his own work – addressing conflicting findings retains the confidence; relating similar results increase the “internal validity, wider generalizability, and higher conceptual level”.
Two issues decide whether one is reaching the closure:
- When to stop adding cases: when new cases no longer bring new insight;
- When to stop iteration between theory and data: when revisiting theory does not bring new insight.
[Strengths and weaknesses]
- Likely to generate novel theory;
- Empirically easy to test, measure and overthrow
- Likely to be empirically valid
- Intensive empirical evidence -> overly complex theory
- From cases to theory -> narrow and idiosyncratic theory
Evaluation of case study can be approached in three ways:
- Look at the resultant theory – does it conform to the criteria of good theory (parsimonious, testable and logically coherent)?
- Look at the process – does it conform to the standard of doing case study?
- Look at the novelty – is it new and framebreaking?
There are so many possibilities to go from individual cases to cross-cases integration.