RESEARCH METHODOLOGY
3.3 Selecting cases
3.3.1 Sampling and replication considerations
Theory building from case studies relies on theoretical sampling, i.e. that cases are chosen
hypotheses testing research it is necessary to limit the „population‟ of cases as it helps to
control extraneous variation and defines limits of generalisation of the findings. Yin
(1994) distinguishes between literal replication and theoretical replication, and argues
that, under different circumstances, it is acceptable to use a single case study that is
analogous to a single experiment, particularly useful when the item under investigation is
extreme, unique, critical, revelatory or so infrequent that it would be silly not to capture it.
In contrast, while repeating more studies of the same type will add external validity via
literal replication, multiple case studies will not necessarily develop the theory much
further. Contrasting studies that demonstrate the opposite to be true will be much better
suited to developing the theory. Pettigrew (1990) encourages the use of cases where
„polar types‟ or extremes are observable especially as there is usually a limit to the
number of cases it is feasible to observe, and suggests that you chose cases where the
process you wish to observe is „transparently observable‟.
3.3.2 Single or multiple case studies designs
In this research the use of a multiple case study approach enables the emerging results to
be verified against a variety of circumstances, thus improving the opportunities for
generalising beyond the immediate case studies. Figure 3.1 shows the model developed
Single-case Designs
Extreme, unique, critical, revelatory
Multiple-case Designs
Replication not sampling logic
Holistic (Single unit of analysis) Global nature of an organisation or a program Embedded (Multiple units of analysis)
Project within a programme, role or location within an organisation
•A single business process where commodity adoption has taken place, but with multiple units of analysis e.g. people, technology, know-how, and plant & machinery.
•A single business process where commodity adoption has taken place, the process itself being the unit of analysis.
•A selection of business processes (the unit of analysis) where commodity adoption has taken place (literal replication). Could be followed by: •A selection of business processes that have not adopted a commodity approach, (theoretical replication). Polar types.
•A single business process (say call centres) with a mixture of both commodity adoption and non commodity adoption (theoretical replication).
•A selection of business processes (the unit of analysis) where commodity adoption has taken place (literal replication), and with multiple units of analysis e.g. people, technology, know-how and plant & machinery. And, or
•A selection of business processes (the unit of analysis) where commodity adoption has not taken place (theoretical replication), and with multiple units of analysis e.g. people, technology, know-how and plant & machinery.
Figure 3.1: Single and multiple case study designs (Based on Yin 1994)
This research has adopted an embedded design that incorporates the components of a
work business process as subunits to enhance the insights into the cases. Care must be taken to ensure that the focus of the study doesn‟t shift away from the prime unit of
analysis (which is the design and decision making activity of internal corporate ventures which have „call centres‟ as a key business process within that venture), becoming a study
of the components that make up business processes instead.
3.3.3 The use of longitudinal case studies
Case studies exhibit great versatility in their applications from providing description, to
testing or generating theory. Table 3.2 shows three examples of the use of case studies for
STUDY DESCRIPTION OF CASE RESEARCH PROBLEM DATA SOURCES INVESTIGATORS OUTPUT Kidder 1982 (Provide description) Data General Describing the intense effort behind the design of a computer at Data General. Direct observation of engineers of Data General design and build a new 32-bit minicomputer in just one year.
Single investigator. Insight into the world of corporate, high- technology America Pinfield 1986 (Theory testing) Canadian government bureaucracy An opportunity to evaluate the two perspectives (Structured and Anarchic) and to develop a more general theory of organisational decision making Direct observation Interviews (during the processes) Interviews (After closure of the process) Single investigator as participant and observer A synthesis of two decision process models (Structured and Anarchic) Harris & Sutton 1986 (Generate theory) 8 Diverse organisations Parting ceremonies during organisational death Interviews Archives
Research team Conceptual
framework about the functions of parting ceremonies for displaced members.
Table 3.2 Examples of Case Study types (Based on Eisenhardt 1989)
3.3.4 Advantages and disadvantages of using case studies
Typically, case studies are well suited to situations where the contextual conditions of a
situation are considered important. For example processes, “a sequence of individual and
collective events, actions, and activities unfolding over time in context” as defined by
Pettigrew (1997), are embedded within context, and therefore require to be studied in that
way. Pettigrew goes on to describe this processual analysis as to “Catch reality in flight”.
Yin (1994) asserts that case studies are:
“Best suited to research where „how‟ and „why‟ questions are being asked about a contemporary set of events, over which the investigator has little or no influence”
Metaphors such as “Drowning in data” (Anderson et al 1995), or „data asphyxiation‟
(Pettigrew 1990) are commonly used to describe the problems faced by a researcher with
this approach. Pettigrew goes on to caution against the inevitable consequences and
considers that “cycles of induction and deduction”, or “going native” (Dyer and Wilkins
1991) is a more realistic approach to take. Similarly, this retrospective analysis relies heavily on people‟s memory, and participants frequently move on to new positions within
the organisation.
3.3.5 Why these particular cases?
All four case studies were chosen from the same organisation and contain the call centres
business process as the focus of the research. They were chosen for their contrasting
approaches to the business problem and diversity of the final outcomes thus encouraging
theoretical replication. The ventures were approached as independent activities and were
launched during a thirteen-month period between March 1997 and April 1998. The four
cases are:
Customer Service – a redesigned venture which developed and implemented an
in-house call centre capability,
Loyalty card – a new venture with an outsourced call centre which was later
brought in-house,
Mail order – another new venture with a completely outsourced call centre, and
Insurance – a new joint venture with the call centre operated by the partner.
Limiting the investigation to a single business process (call centres) eliminates many of
the extraneous variables that might otherwise be present if looking across a number of
disparate business processes, and restricting the case studies to one organisation limits the
concentrate on the „decision-making‟ characteristics rather than the organisational or
business process ones.
Eisenhardt (1989) describes how it is acceptable to add cases later on in the research
process as appropriate, and cites the study of strategic decision-making (Eisenhardt and
Bourgeois 1988). This approach extended the study and allowed for the inclusion of
additional data from two extra cases where the composition of the teams under
investigation had changed. This additional data could then be dynamically compared with
the results from those teams where the team composition had not changed.