Chapter 6 Case Study Design
6.2 Case Selection
The next step in designing the case study is to select cases:
“When using a multi-case design, a further question you will encounter has to do with the number of cases deemed necessary or sufficient for your study.
However, because a sampling logic should not be used, the typical criteria regarding sample size also are irrelevant. Instead, you should think of this decision as a reflection of the number of case replications – both literal and theoretical – that you would like to have in your study.” (Yin, 1994, p. 50).
As no rival theories are defined per se, external validity is the main driver for the number of theoretical replications:
“When you are uncertain whether external conditions will produce different case study results, you may want to articulate these relevant conditions more explicitly at the outset of your study and identify a larger number of cases to be included.” (Yin, 1994, p. 50).
External conditions must therefore be explicated to ensure appropriate case selection.
6.2.1 External Conditions (for SCM and SCRM)
Conditions relevant to the practices of SCM and SCRM may include:
1. the size of the company might have an impact on the resources available for staff functions,
2. the number of direct competitors and the history of competition might influence the need to lock in critical suppliers (and customers),
13 See Chapter 3.3.2 for a discussion of supply (risk) management.
3. recent accidents might impact resource allocation, increasing the focus on risk management,
4. the complexity of internal processes might increase the awareness of complexity in the environment and thereby impact the risk management practice, and
5. the nature of inputs (simple versus complex, or standard versus unique) might influence the supplier base and thereby the SCM practice, and furthermore the perception of supply risks.
Albeit this list is undoubtedly incomplete it introduces too many variables. The first examples are justified by an assumption of larger companies having more resources available (e.g. Chopra & Sodhi, 2004) and might have a longer time horizon on investments in processes and technologies. whereas the fourth and fifth are based on an assumption of higher levels of complexity require more risk management (e.g. Harland, Brenchley, &
Walker, 2003; Ottesen & Gronhaug, 2002e; Perrow, 1984). From a SCM perspective the fourth argument might be relevant due to the need for precision of internal processes when dealing with unique/complex inputs (here it is assumed complexity in input leads to segmentation of supply market – basically the specialization argument, see Gripsrud, Jahre, &
Persson (2006)). The last example points towards ‘internal SCM’, assuming internal integration support complexity in processes (e.g. Storey et al., 2006; Simatupang & Sridharan, 2005). The second and third examples do not combine SCM with SCRM and may thereby be removed from the list. External conditions (subjectively chosen) for this study are thereby:
External Condition 1: Size of Company.
External Condition 2: Complexity of Processes.
External Condition 3: Complexity of Inputs.
Recruiting SCM Companies
Furthermore it is deemed critically important to include “Best in Class Supply Chain Management” companies in the convenience sampling as to ensure variation across the case companies. A corresponding variation across SCRM practices would have been preferable, but none such mechanism could be identified. Therefore only a fourth (and not a fifth) condition is added to the list above:
External Condition 4: SCM Award Winning Company.
Due to the use of analytical generalization (in contrast to statistical generalization, see above) each external condition is evaluated separately. For each external condition the convenience sample should contain at least 2-3 cases to allow for variation within each group. Weighing
this argument against the level of existing knowledge and the effort needed for each case (e.g.
Eisenhardt, 1989b; Dyer & Wilkins, 1991) a convenience sample size of ten case companies is considered sufficient.
6.2.2 Case Selection Method
The convenience sampling must be performed in a manner ensuring above mentioned criteria are met. As only two (‘Size’ and ‘SCM Award’) of the four external conditions are readily available “from outside the company” the classification process is split in two, see Figure 6-2 below.
Figure 6-2: (An Iterative) Case Selection Method
Classification I:
•Size?
•SCM Award?
Classification II:
•Complexity(I)?
•Complexity(P)?
Outside / Distance Population
(Danish Manufacturing
Companies)
Participating Companies Inside / Interaction
Goal: Even split Large/Small, and SCM
Award/No SCM Award
Goal: Even split Complex/Simple Input
and Complex/Simple Processes Goal: Ten participating companies
As shown in the figure above the case selection method is thereby iterative as the second evaluation needs to be performed before another company can be included. Companies are therefore included one at a time.
Case Selection Process
In order to meet the fourth external condition half the companies were picked from the lists of award winning companies, winning either The Danish Logistics Award or The Post Denmark Logistics Award. The former is depicted in Table 6-5 below, the latter in Table 6-6 (participating companies are highlighted).
Table 6-5: The Danish Logistics Award14 Year Company
2005 Vest-Wood A/S
2004 Netto A/S
2003 Oticon A/S
2002 LK A/S
2001 Grundfos A/S
2000 Arla Foods A/S
1999 Carlsberg A/S
1998 Olicom A/S
1997 Coloplast A/S 1996 Bang & Olufsen A/S
1995 Voss-Atlas A/S
1994 Herning Galvanisering A/S
Table 6-6: The Post Denmark Supply Chain Award15
Year Large Company Small or Medium-sized Company
2005 Fritz Hansen A/S Bolia.com
LINAK A/S (manufacturing) 2004
Sanistål A/S (trading company) Skagenfood A/S Gate Gourmet Scandinavia (manufacturing)
2003 Coop Denmark A/S (trading) Dekra Automotive Services A/S (trading)
For each of the companies the remaining three external conditions are determined according to the following guidelines.
Criteria for Size
Albeit revenue is the standard measure for size16 the SCM context calls for a more structural argument for size. The following variables are used to determine size:
¾ Number of Local HQ’s (Hubs)
¾ Number of Subsidiaries / Sales Offices
¾ Number of Distributors / Representatives
Table 6-7 below contains the result of the classification for size.17
14 Source: www.logistikkonferencen.dk.
15 Source: www.supplychainprisen.dk.
16 The definition of SME used within EU uses number of employees and either turnover or balance sheet total.
For more info, please see Szabo (1998).
17 The reason for the two groups (large and small) not being equal in size (six large and four small companies) is due to the need to “balance” according to the other two variables, and due to delayed confirmation of participation from another two case companies (who ended up declining the invitation to participate).
Table 6-7: Classification for Size18 No of local No of subs / No of distributors /
Company HQ / hub ? sales offices ? representatives ? Other ? Size
Bang & Olufsen 2 30 20 Prod., Show Rooms Large
Brüel & Kjær 8 48 77 Service Centers Large
Coloplast 3 26 49 Production Large
Dyrup 0 8 15 Production Large
Fritz Hansen 0 7 400 Show Rooms Small
Linak 1 26 7 Production Large
Novozymes 0 18 0 Production Large
Oticon 1 15 72 Production Small
RIEGENS (1) 2 0 Production Small
SDC Group 0 5 0 Production Small
Criteria for Complexity
As mentioned previously the inclusion of companies were based on a preliminary evaluation of complexity in input and processes. As both of these criteria need a more thorough analysis access to the company is a pre-requisite for classification. Upon obtaining access to the case company evaluating inputs and processes for complexity should be possible. But since no generally acceptable constructs for measuring complexity in neither input nor process was identified, the classification is left to the subjective choice of the researcher.
6.2.3 Learning Points
One of the first learning point was the (in hindsight) obvious fact that complexity in process might vary greatly across supply chains. And since a company might (and certainly many do) participate/have multiple supply chains, the model needs to distinguish between supply chains. Pondering the criteria for determining whether there are three or four supply chains (or business models) in a company reveal that the case studies challenge the classification. As described in Gardner & Cooper (2003) the exercise of mapping a number of supply chains might result in numerous iterations ultimately leading to either despair or a simpler model than desired. A second learning point was that complexity in process might not be that easily described (‘Simple’ or ‘Complex’). In several companies the process performed by the technology is deeply complex and relies on extreme precision. In these cases, complexity is embedded in the manufacturing process, whereas the process as understood in a SCM context (see Chapter 8.1.2 for a thorough discussion of the term process) is simplified as a direct result.
Classifying the Case Companies
For each supply chain the complexity of input and process were evaluated, see Table 6-8 below.
18 Analysing the companies by turnover and number of employees would place Brüel & Kjær in the group of smaller companies. The rest of the companies would not change classification.
Table 6-8: External Conditions per Supply Chain
Company Size Supply Chain(s) Input Process SCM Award
Bang & Olufsen Large All Complex Simple Yes
‘Project Sales’ &
‘Customer Project’ Complex Complex Brüel & Kjær Large
All Other Complex Simple
No
Coloplast Large All Simple Simple Yes
Dyrup Large All19 Simple Simple No
Fritz Hansen Small All20 Simple Simple Yes
LINAK Large All Simple Simple Yes
Novozymes Large All21 Simple Complex (embedded) No
Oticon Small All Complex Complex (embedded) Yes
‘Project’ Simple Complex
RIEGENS Small All Other Simple Simple No
SDC DANDISC Small All Simple Complex (embedded) No
In order to obtain one set of external conditions per case company the above listed values are manipulated in the following ways:
1. The ‘Process’ construct is split into two constructs ‘Process’ and ‘Manufacturing Process’, the latter describing the complexity of the manufacturing process. As a result, both constructs can hold the values ‘Simple’ and ‘Complex’.
2. For each company having more than one supply chain, the most “advanced” value is kept. For Brüel & Kjær, for instance, the company ends up with the values ‘Large’,
‘Complex’, ‘Complex’, and ‘Simple’ (for the external conditions ‘Size’, ‘Input’,
‘Process’, and ‘Manufacturing Process’, respectively).
Table 6-9 below holds the resulting set of external conditions per company.
Table 6-9: External Conditions per Company
Company Size Input Process Mfg.
Process
SCM Award
Bang & Olufsen Large Complex Simple Simple Yes
Brüel & Kjær Large Complex Complex Simple No
Coloplast Large Simple Simple Simple Yes
Dyrup Large Simple Simple Simple No
Fritz Hansen Small Simple Simple Simple Yes
LINAK Large Simple Simple Simple Yes
Novozymes Large Simple Complex Complex No
Oticon Small Complex Complex Complex Yes
RIEGENS Small Simple Complex Simple No
SDC DANDISC Small Simple Complex Complex No
Participation
Even if most invited companies accepted the invitation almost all had concerns. First of all the workload for the companies is non-trivial. Insisting on interviewing key personnel was a
19 Actually another supply chain exists but since it’s quite marginal it’s left out.
20 The production of very expensive (primarily wooden) products is best described as crafts-based production.
The other two supply chains differ in distribution solution alone.
21 The two supply chains are identical in respect to external conditions.
cause for discussion in several of the companies. A second (and perhaps the most critical) concern was the subject of the research project: some companies were reluctant (to say the least) to share information on their risk management practices22. Once recruited to the study, though, each company appeared to be quite committed.