Despite the plethora of ISD methods available, organizations are seldom satisfied with existing methods. Surveys as well as case studies reveal that organizations tend to develop their own local “variants” of methods, or adapt methods available according to their situation-specific needs (Pyburn 1983, Russo et al. 1995, Hardy et al. 1995, Wijers and van Dort 1990, Aalto 1993, Aaen et al. 1992)[4]. This means that methods from outside an organization do not meet the requirements for its ISD efforts, i.e. they are not considered applicable. As a result, the only choices are to abandon the method, try another one, continue the use of the method, or develop methods locally.

By local method development we mean organizations’ attempts to develop their own method or methods. This means that the local method includes aspects which are not included in any other single method. In our case of modeling related methods these extensions can include whole modeling techniques, new modeling concepts, or new constraints. Local method development is often carried out by combining and modifying existing methods. Surveys investigating the use of methods have shown that 38% (Hardy et al. 1995) or 36% (CASE Research Corporation, cited in Yourdon 1992) of the organizations used methods developed in-house. These, however, can also be adapted to organization, project, or individual needs. Thus, the difference between local method development and adaptation is noteworthy. By method adaptation we mean attempts to modify available methods, including local variants, for situational needs. On the adaptation side, a survey by Hardy et al. (1995) showed that 88% of the organizations studied had adapted methods in-house. In another study the percentage of method adaptation is similarly 88% (Russo and Klomparens 1993, Russo et al. 1995).

Studies of CASE tool usage (Wijers and van Dort 1990, Aaen et al. 1992) have obtained similar results. They show an obvious need for local method support and indicate that although companies have introduced CASE tools with particular methods, they face difficulties in CASE use due to the limited possibility to adopt and develop situation specific methods (Wijers and van Dort 1990). Different development situations, cultures, skill levels, and types of IS require different ISD approaches. Accordingly, standard-like methods are less common than expected, and less popular than their local variants. This is interesting since tool markets have focused on fixed and method-dependent tools. This may explain the relatively low acceptance of CASE tools which lack method modification and adaptation possibilities (cf. Aaen et al. 1992). Therefore, a need for more flexible and customizable tools has been emphasized (cf. Forte and Norman 1992, Seppänen et al. 1996).

We can find in the literature some case studies of local method development including extensions of current methods (cf. Aalto 1993, Nissen et al. 1996). These describe how methods have been modified and explain reasons for their evolution. There is, however, surprisingly little empirical knowledge available on local method development and method use (Wynekoop and Russo 1993). In fact, most of the reported work has concentrated on developing new standard methods. Though many organizations develop methods in-house or adapt them, we know little as to why and how this is done, or whether local or adapted methods work better. Some studies on method development indicate that although many companies are “rolling their own”; the selection of methods, their development and introduction seems to be done in an ad-hoc manner by choosing tools and methods on a trial-and-error procedure (Smolander et al. 1990). Although local method development is common there is a lack of proven principles. These principles include how to construct and adapt methods for particular needs, how to check the applicability of the method, and how to organize method development efforts.

To understand local method development efforts in more detail we distinguish five steps that every organization faces while developing methods in-house. The identification of these steps is based on analyzing and synthesizing the literature (e.g. Smolander et al. 1990, Tagg 1990, Tolvanen and Lyytinen 1993, Brinkkemper 1996). These steps are illustrated in Figure 1-1 and described in more detail in Section 3.2. It must be noted that the figure is an assertion rather than a generally proven process model of local method development.

FIGURE 1-1 Steps of local method development.

1) Selection of methods. First, every organization or ISD project must make a decision which methods to follow and use. Even an organization that does not use methods at all has made some decision, either explicitly or implicitly. Similarly, cases of local method development (cf. Jaaksi 1997) are often based on a selection of a well-known “text-book” method which is introduced first in its standard form, and later - after gaining experience about its use - modified to better meet local needs.

2) Method construction is a task in which selected methods are composed or new ones are created to meet specific objectives of ISD (Kumar and Welke 1992, Heym and Österle 1993). This task includes building, improving and modifying a method by specifying its components and their relationships.

3) Tool selection and adaptation. Tool adaptation can be defined as a task in which a given method is represented and implemented for an ISD tool in such a way that the tool can support tasks as prescribed by the method (Tolvanen and Lyytinen 1993). If such a tool modification is not possible, organizations still face the question “which method-specific tools should be selected?” A more detailed discussion about different selection strategies for CASE can be found from Bubenko (1988).

4) Introduction of methods deals with various tasks for initiating method use, such as teaching, carrying out possible pilot projects etc.

5) Method use refers to an actual ISD effort in which methods are utilized together with supporting tools.

The transitions described show a succession of steps through the selection of methods to their use, but it is also possible to omit some steps. For example, local method development can be carried out without selecting or using any ISD tool, or without proper method introduction. Similarly the steps can be overlapping (Nuseibah et al. 1996). Other transitions are also possible or can even be more common. For example, an organization does not necessarily use, and thus need to select, a computer-aided tool, or an organization can start from the middle step by choosing a CASE tool, and indirectly the accompanying methods.

The specification of transitions suggests a “loop” to method refinements which is of major interest in our study: At each step of method development new experience can lead to method modifications. This part is illustrated by arrows on the left and right sides. An organization or an ISD project not only produces ISs, but also gains and creates knowledge about the ISD. Typically, at least part of this knowledge can be incorporated into ISD methods.

The figure identifies several transitions for method refinements. Such refinements can occur either before, during or after the use of a method. In the former case, method refinements occur during method construction, tool selection or adaptation, or method introduction. For example, the capabilities of the tool for method adaptation can lead to new method modifications. In the latter cases, situations that have taken place during method use are analyzed, generating new insights on how to use methods. This experience-based method refinement can be characterized using Schön’s term, reflection-in-action (Schön 1983). In this loop the situation “talks back” and the practitioner reframes the situation. Depending on how experiences are externalized (Nonaka 1994), refinements can take place during or after method use. Refinements which occur during method use - “on-the-fly” - deal mostly with an individual’s interpretations, and give new meaning to a constructed method. It must be noticed that this type of refinement often occurs without any documentation, and thus takes place in the dimension of tacit knowledge (Nonaka 1994). If these experiences are made explicit, and thus available for other method users, they can be related to the methods constructed and used to modify earlier methodical understanding. In this thesis we mostly examine method refinements, because we believe that the applicability of methods can only be improved when experience is made explicit for future ME efforts.

According to this a posteriori view of ME, an important factor in local method development is the capability of an organization or project to learn of its method use, create knowledge about the applicability of a method, and utilize this knowledge for refinements. This is true not only of method development but also of organizational knowledge creation in general (Nonaka 1994). In this study our interest is not in how efficiently an organization develops ISs, but in how it creates explicit knowledge of the ISD method. Our focus will be on defining, refining, validating and discarding ISD knowledge. Methods are seen as one part of organizational knowledge, which evolves and needs to be collected, analyzed, maintained and purged. We believe that how and to what extent this is done in different situations is of great importance to the success of local method development and to the usefulness of methods.

[4] The empirical studies on method use and on local method development are discussed in more detail in Section 2.4.