Why CASE Tools in Information Systems Development? - an empirical study concerning motives for investing in CASE tools Stefan Cronholm Dept. of Computer and Information Science Linköping University 581 83 Sweden e-mail:
[email protected]
Abstract For developing of computer-based information systems (IS) there nowadays exist different kind of tools. This study deals with tools for earlier phases (analysis, design), so called upper-CASE. CASE tools usually contain support for one or more ISD-methods. Working with CASE tools means that the ISD-work is both method- and computer-based. This paper is concerned about the motives for investments in CASE tools. Why do organizations invest in CASE tools? What objectives are the organizations aiming at? This study is also interested in if the stated objectives are reached. Are the motives for using CASE tools fulfilled?
1. Introduction For developing of computer-based information systems (IS) there nowadays exist different kind of tools. The last five years there has been an intensive development of CASE tools (Computer Aided Systems/Software Engineering). The aim of these tools is to assist the system developers. There are tools supporting different phases in information systems development (ISD). This study deals with tools for earlier phases (analysis, design), so called upper-CASE. When, in this paper, I use the term CASE tool I mean upper-CASE. CASE tools usually contain support for one or more ISD-methods. Working with CASE tools
1
means that the ISD-work is both method- and computer-based (Goldkuhl & Cronholm, 1993). The use of CASE tools means that different parts of the ISD-process will be automated (McClure, 1989). This automation leads to consequences for the IS-process. The use of methods will be more formalized than the earlier use of pen and paper or drawing-programs. These simpler aids allow a higher degree of freedom than CASE tools. This paper is concerned about the motives for investments in CASE tools. Why do organizations invest in CASE tools? What objectives are the organizations aiming at? This study is also interested in if the stated objectives are reached. Will the motives for using CASE tools hold? This paper is organized as follows. After the first chapter a description of the research method will be given. The next chapter presents results from empirical studies where identified motives are described. The following chapter discusses if the stated motives (goals) are reached. The last chapter consists of conclusion and discussions.
2. Research method In this study a qualitative method has been used. Data has been gathered through six case studies including about 20 interviews. One interview lasted between two and four hours. The interview period took place during 6 months in the winter 1994. The interviews ceased when my material became satisfied and when the interviews did not add something new to my data. The analysis method used is based on "grounded theory" (Glaser & Strauss, 1967; Strauss & Corbin, 1990). The reason for using qualitative approach is that I was interested in understanding the motives behind implementing CASE tools. I wanted to achieve a better understanding for why organizations use CASE tools. During the interviews I have used a tape recorder and after each interview a printout was sent to the responders for possible modifications and corrections. The analysis of the collected data followed "grounded theory" which consists of conceptualizing and categorizing data aiming to find similar and different patterns in the case studies. You try to relate identified categories to each other. During the analysis process you ask yourself questions like: What does this represent? What does this mean?
3. Related research There have been several other studies close to this research. Most of these research have been performed with quantitative methods (e.g. questionaires). One related study has been performed by Aaen et al (1992). This study concerns how well finish and danish companies and organizations are satisfied with CASE tools. Another related study is done by Wynekoop et al (1992). This study deals with the success of CASE implementation in relation with existing expectations of CASE tools. Other related research has been performed by Hayley & Lyman (1990), Kuvaja (1992), Orlikowsky (1988) and Lyytinen & Tahvanainen (1992).
2
4. Description of case studies Each case study or research unit consist of an ISD-project . The interviews have been performed with different participants and roles in the ISD-process. The following roles have been identified and interviewed in the ISD-projects: • project manager • method expert • tool expert • system developers • customer • end-user One role can be founded by several persons. For instance method knowledge has been founded by several persons. One person can also possess different roles. E. g. persons can have both method and tool knowledge. The methods and tools used in the studied organizations is presented below.: Case
Model/method
Tool
A
LOGIC (Cap Gemini, 1992)
Qualicycle
B
Svea (Kommundata, 1992)
SVAR
C
LOGIC (Cap Gemini, 1992)
ADW
D
Svea (Kommundata, 1992)
SVAR
E
REFLEX (Programator AB, 1990a; Programator REX AB, 1990b)
F
Direct (Axelsson & Ortman, 1990)
SYNON
Figure 1 Used methods and CASE tools
The investigation was performed in Sweden and thus there were mainly Swedish methods used. All the ISD-projects were developing administrative systems such as systems for insurance, purchase, school administration and old age care. The method maturity are in all organizations relatively high since the methods has been used in several earlier projects. In one ISD-project an object-orient variant of the traditional method was used for the first time. The tool maturity differed among the organizations. Four organizations have used the CASE tool in several earlier projects. One organization has only partly and to a minor extent used the tool. One organization used their tool for the first time. In one of the case studies there were are lower-CASE (SYNON) used. Lower-CASE supports the later phases in the IS-process. The reason for chosing this organization as a case study is that I was interested in why the organization not have invested in a upper-CASE. This organization has combined a ordinary drawing program with a lower-CASE.
3
5. Motives for using CASE tools The extension of analysis why the organizations have wanted to implement a CASE tool have varied. In several cases there are well formulated and specified requirements of attributes for the tools. These organizations have thoroughly analyzed the benefits for using CASE tools. They have identified some working procedures where CASE tools can prove beneficial. In some cases the motvies have not been analyzed to such a great extent. The foremost motive for using CASE tools is that the organizations wanted to reach competitiveness. Reaching competitiveness is the core category which is interpreted from collected data. The organizations is in a market situation and they are continuously searching for new possibilities to improve the ISD-process. Competitiveness can be reached by: • faster IS-development • improved product quality Reaching faster IS-development means continuous work. Customer requirements increase and the organizations feel a pressure to produce new systems faster. The organization also wanted to produce more in the same time. Supporting the category of faster development there are statements claiming the importance of rationalising the ISD-process through making working procedures easier and managing document handling more effective. In effect the organizations will be able to: • update and modify diagrams easier • achieve a tool to assist method institutionalizing (achieve a higher degree of method use) • achieve a faster dialogue with customers/end-users To be able to update and modify graphs easier is a pervading motive in all the case studies. The updating of graphs has earlier been a manual process which was time consuming and difficult. In some cases even so difficult that no documentation was produced at all. The organization also wanted a tool for getting a method institutionalizing. This means that the organizations wanted to achieve a more standardized way of performing ISD. They wanted a better structure of the ISD-process through a wider use of methods. The implemented methods in the tools will secure that methods rules will be followed. This will hopefully lead to an easier IS-maintenance. The motive to achieve a faster dialogue with customers is partly method based. It is the method and not the tool which will guide the way of performing ISD. This motive is valid because the organization wanted to reach a higher use of methods through using tools The CASE tools will contribute to a faster feedback of discussions from different meetings (e g seminars). In these seminars the system developers and the customers/end-users together defining the problem domain with help of different models; (e g dataflow diagrams (Yourdon, 1989), datamodel diagrams (Chen, 1977). After the seminars these models are documented in the CASE tools by the system developers/tool experts. There is a request for a fast feedback to the customers for verifying the correctness in these documents. Improved product quality is also a strong argument for using CASE tools. The term product means the information system which is to be delivered. The reason for this argument is that there are 4
more requests from customers and therefore the compentence of the system developers competence must increase. An increase in the competence of the system developers can among other things be achieved from increased method use. This motive is especially valid for inexperienced developers. The motive for a wider use of methods is valid both for achieving a standardized way of performing ISD and for achieving an improved product quality. Identified opinions for getting an improved product quality are that the organizations wanted to: • be able to work in a more flexible way • achieve an improved dialogue with customers/end-users • achieve an improved consistence in the documentation • be able to produce more presentable graphs • increase presumptions for problem solving To be able to work more flexibly means that the organizations wanted to vary ways of performing ISD according to the nature of the problem domain. The organizations wanted to be able to choose appropriate method parts in the CASE tool according to a specific situation. An improved dialogue with customers/end-users want the organizations to reach through different system models which will facilitate communication between system developers and customers/end-users. The dialogue will also be improved by achieving an improved consistence in the IS-documentation. The CASE tools automatically support consistence controls of ISDdocumentation. For instance, it is possible to get an automatic control if all existing objects in a graph are described in the repository or if the description performed is set according to the method rules. The CASE tools will with this functionality provide a more error-free documentation and minimize ambiguity. This will lead to an improved comprehension during discussion with customers/end-users. Another identified motive for using CASE tools is that the organizations wanted to be able to produce more presentable graphs. They wanted to have better graph quality than earlier when the graphs were handmade. More presentable graphs will improve the readability of the documentation. Another important motive is that the organizations wanted to reduce working time on routine procedures. This time reduction can instead be used for problem solving and thereby a higher quality of the product can be achieved. Apart from the motives presented above I have also identified a motive as "it is important to follow the technical development". The organizations wanted to utilize new techniques in order to study and identify how these tools work. Using a CASE tool also ensures that you perform ISD in a modern way. This motive are of more symbolic character and will show that you follow the technical development as a system developer. The identified motives are motives for using a CASE tool but these motives also constitute goals for the organization. Figure 2 - figure 4 presents relations between identified motives (goals). These relations are shown in goal graphs from the IS-method CA/SIMM (Goldkuhl & Röstlinger, 1988). The presented motives (goals) are identified motives from performed interviews. The relations between identified motives (goals) have been carried out by myself. The graphs present a hierarchy where the foremost motives (goals) can be found in figure 2.
5
Goal diagram Refcode: Foremost
Concern: Identified foremost goals
goals
Reach competitiveness
Faster IS-development
Improve product quality
IS-development
Product quality
Figure 2 Goal graph: Foremost motives (goals)
6
Goal diagram Concern: Identified related goals Improved product quality
Refcode: Product quality
Foremost goals
Improve product quality
Improve consistence in the documentation
Improve customer dialogues
Increase competence by system developers
Be able to produce more presentable graphs
Improve problem solving
Be able to work more flexible
IS-development Increase method use
Figure 3 Goal graph: Improve product quality
7
Goal diagram Concern: Identified related goals faster IS-development
Refcode: IS-development
Foremost goals
Faster IS-development
Rationalize the ISD process
Simplify working
procedures
Simplify modifying of graphs
Achieve a faster dialogue with customers
Increase standardization
Increase method use
Introduce tools for method institutionalizing
Figure 4 Goal graph: Faster IS-development
8
Productquality
6. Fulfilling of motives for using CASE tools 6.1 Assessment A documented assessment could be a base for a judgement if expressed motives are fulfilled. Only in one case a documented assessment has been performed. In this case the tool was first tested in a pilot project and experiences from this project were evaluated. The result of this assessment led to a decision of not using the actual tool in the future. Note that the meaning of the decision was not to use CASE tools at all. The decision was a rejection of the specific tool. For the present there is an ongoing research on new tools which will better fit the actual needs and environment. In all the other cases the empirical data is in principle based on experiences from using the CASE tool. These experiences are founded on subjective assumptions. In one of these cases there has however been a minor assessment and these organization tried to quantify the benefit through measure needed calendar time for carring out a requirements analysis. The result from the measurement was that the organization does not have to spend equal working hours when a CASE tool is used. The organization estimated that 15 % fewer working hours per project are needed when a CASE tool is used. Another observation was that the quality of the documentation was higher. The following reasons have been stated in the other cases for not performing a documented assessment: • it is hard to measure the benefit from the tool investment • lack of time The organizations considers that it is hard to measure the tool investment in numbers and that the benefits must be based on experiences from using the tool. In some cases the organizations think there is a lack of time for performing assessments. These organizations argue that there are always new projects waiting for attention. To pay attention to new projects has been given priority since this leads to new incomes. The organizations which have not performed a documented assessment have motivated this by its being more important to pay atention to ongoing activities.
6.2 Fulfilling of motives Following categories which concern fulfilling of motives has been generated: • tool functionality • ISD-development • standardized way of performing ISD • product quality • flexible way of performing ISD • knowledge about CASE tools In two of the cases the organizations thought that the motives for using CASE tools has not been fulfilled. Lack of tool functionality is the foremost reason. In one of these organizations this reason has been the determining factor for not continuing using the CASE tool. In the other organization this 9
factor has also been decisive together with the experience that the used tool does not fit the actual environment. In both these cases there have been higher expectations of performance and functionality than the tools have been able to offer. Missing or lack of functionality consists of: • low covering degree • insufficient functions for word processing • unstable tool • hard to generate report formats according to requirements Low covering degree means that the tools facilities only covers a minor part of the method. There is an insufficient method support. The tool is not properly adapted to the used IS-methods and there is a lack of support for more method parts. For instance there is a lack of support for prototyping and the later coding process. Insufficient functions for word processing is a pervading problem in all the studied cases. Development of functions for word processing has not been given priority from tool constructors. Especially in one case the tool is unstable. Many functions did not act as they were designed to act and there were also unexplainable interruptions when using the tool. Another lack of functionality deals with the possibility to generate report formats. The tool users have not been able to generate report formats of their own. The only way to produce reports is to use predefined report formats. With one tool this should be a possibility but the user did not succeed . It was to difficult and too time consuming to create ones own report formats. In the other cases, there was no functionality supporting designing of own report formats. This lack is a problem. There is a need to design and group information in varying ways. In the other cases the organizations accept the tool functionality. In these cases there were also negative opinions about the tool, but these opinions were not so determining as in the two cases mentioned above. In these cases the organizations are satisfied with the tool in principle and think that the tool offers a good support. In these cases the organization thinks that the motives are fulfilled. The organizations think that the tool among other things improves possibilities to achieve a higher consistency in the documentation through offering different kinds of automatic controls. The positive organizations also think that the tool supports analysis work and customer dialogues. In the cases where the organizations are more positive they think that they have achieved a faster ISD-development thanks to the CASE tool. They think that the tool gives a good support and it is possible to carry out ISD cheaper with a CASE tool. This is first and foremost thanks to a faster documentation process. Further more the organizations think that they have reached increased possibilities of introducing a standardized way of performing ISD and thus getting closer to method institutionalizing. The tools has contributed to carrying out ISD in a more formalized way since the tool contributed to a higher use of methods. This fact is also mentioned by Goldkuhl (1991). Goldkuhl means that introducing models and methods among other things is aiming to standardize ISD. Furthermore the more positive organizations think that the tool provides a higher product quality. Product quality depends on both methods and tools. In the two cases where the organizations
10
were unsatisfied with the tool they think that it is the method and the IS-competence which gives products high quality. The used tools do not rule in the way that they are forcing a way of performing ISD. If you want you can choose different feasible parts of the methods. E. g. you can only perform data modelling or some other method part if you want to. In this way the tools allow a flexible way of performing ISD. On the other hand there are no possibilities of adapting the tools according to other methods with other notations. In this sense the motive flexible way of performing ISD has not been achieved. In these cases where the motives have not been fulfilled the organizations use the CASE tool as a motive for achieving increased knowledge about CASE tools. The use of a tool has provided a valuable knowledge for establish request and demands on future tools. These organizations think that it is important to test the tools in order to get a better understanding about the tools functionality and usability. Another observation is that requirements of CASE tools has increased during tool use. The requirements have also been expressed more exactly and refined from the original list of requirements. This is a sign that the organizations have developed new knowledge which made it possible to improve the requirements. This new and extended knowledge has not yet been realized because changes in the tools is to expensive to carry out.
6.3 Line of action in future ISD-projects In all the studies except one the organization will continue to use the CASE tool. In the case where the organization rejected the tool the principal reasons for this are: • the tool is to expensive • the tools functionality is not acceptable In one case the tool is only used to a minor extent. The foremost reason for this is that the tool does not fit the actual environment. In this case, as well, the organization is not satisfied with the tool functionality. In both these cases the organizations are searching for new tools that better fulfil the existing needs and requirements. In the other cases there are more positive than negative aspects of using the tools. The organizations think that the tools contribute to an improvement of the ISD-process. Compared to manual work the use of the tool has provided positive consequences. However, the organizations are not fully satisfied and they are searching for new ways of improving the environment for performing ISD. In some cases there are investigations of it is possible to introduce more object-oriented ways of performing ISD in the future (e. g. Coad & Yourdon, 1991). The trend is that ISD is moving in the direction of object-orientation. Object-oriented methods are for the present a popular research subject (Bubenko & Wangler, 1992). Of course the tools must change according to the methods in order to be used in a proper way.
11
7. Conclusions and discussions The foremost motive for CASE tool implementation is that the organization wants to achieve competitiveness in the form of faster ISD-development and improved product quality. These findings agree with motives founded in studies performed by Hayley & Lyman (1990), Aaen m fl (1992), Kuvaja (1992) and Orlikowsky (1988). A motive that I have not found in other studies is that CASE tools are implemented for achieving a higher use of method institutionalizion. I think this motive is interesting and I interpret it as organizations searching for means that will popularize method use in order to reach a more standardized way of performing ISD. The fulfilment of the motives varies among the studied cases. In most of the cases the organizations think that the motives are fulfilled. These organizations accept the tool functionality and think that the tool contributes to an improved ISD-process. These organizations are not fully satisfied with the tool but they accept it. They think that the tools can be more improved for fitting in the ISDprocess. In two cases the organizations do not think that the motives are fulfilled. There are many reasons for this. One reason is that one of these two organization has introduced a new ISD-method at the same time as the tool was introduced. The new method was an object-oriented ISD-method. This means that the organization has two barriers to over come at the same time. The first barrier is to learn and understand a new method (a new way of performing ISD) and the second barrier is to learn how the new tool works. Another reason is that the tool was recently introduced at the market and was not acting according to the expectations. The importance of tools fulfilling expectations is discussed by Wynekoop et al (1992). Wynekoop et al means that a higher acceptance of CASE tools is reached when the expectations are fulfilled. Even Lyytinen & Tahvanainen (1992) consider that one reason for failure is that the CASE tools do not live up to the users expectations. Introducing a CASE tool is a long-term investment and it is therefore important to try and assess the tool use. In the studied cases the organization have given priority to a more short-term policy. Only one organization has performed a documented assessment. The reason for the lack of assessments is that the organizations chose to put their resources on carry out new projects. This is a short-term policy. Nilsson (1991) has in his studies founded that "careful and comprehensive assessments simplifies the following adaptation. Further more Nilsson (1991) thinks that "a quick choice with a lack of assessment leads to impossible adaptations". By adaptation is meant to minimize the differences that appears between the tools functionality and the ISD-process. According to Nilsson (1991) there is seldom a perfect match between the tool and the ISD-process. An assessment should lead to a documented confirmation of whether the motives are fulfilled or not. A confirmation of the success of introducing CASE- tools implies advantages in a long-term perspective. According to Bubenko & Wangler (1992) and Wynekoop & Conger (1991) assessments of the use of CASE tools are seldom performed. An obvious risk with lack of assessment is that CASE users do not know what they want and that CASE designers do not know what the users want to have.
12
I think that most of the CASE constructors do not have proper knowledge of what ISD is about. There must be a better dialogue between designers and users. One way to improve the communication between constructors and users is to collaborate about assessments. Constructors and CASE users together could carry out assessments aiming to establish a more well-grounded tool development and a better understanding of what users request.
References Aaen I, Siltanen A, Sörensen C, Tahvanainen V-P (1992), A tale of two countries: CASE experiences and expectations, in proceedings of the IFIP WG 8.2 Working conference on The Impact of Computer Supported Technologies on Information Systems Development, Kendall K E, Lyytinen K, DeGross (eds), Minneapolis, Minnesota, USA, 1992 Axelsson L, Ortman L (1990), In swedish: Direct-modellen - en utvecklingshandbok, en vidareutveckling av SVEA, Studentlitteratur, Lund Bubenko J, Wangler B (1992), Research Directions, in Conceptual Specification Development, in Conceptual Modeling, Databases and CASE, Loucopoulos P, Zicari R (eds), Wiley Cap Gemini (1992), In swedish: Metodhandbok, Stockholm Chen PP (1976), The Entity-Relationship modell - toward a unified view of data, ACM TODS, Vol 1, no 1 Coad P, Yourdon E (1991), Object-Oriented Analysis, Prentice Hall Glaser B, Strauss A (1967), The discovery of grounded theory, Aldine, New York Goldkuhl G (1991), In swedish: Stöd och struktur i systemutvecklingsprocessen, Institutionen för datavetenskap, Linköpings universitet Goldkuhl G, Cronholm S (1993), Customizable CASE environments: A framework for design and evaluation, presented at COPE' IT in Copenhagen, NordDATA, 1993 Goldkuhl G, Röstlinger A (1988), In swedish:Förändringsanalys - arbetsmetodik och förhållningssätt för goda förändringssätt, Studentlitteratur, Lund Hayley K J, Lyman H T (1990), The Realities of CASE, Journal of Information Systems Management, Summer 1990 Kommundata (1992), In swedish: Metodhandbok v 0.5 Kuvaja Pasi (1992), Effects of CASE implementation on the productivity and maturity level of software development - a case study in a large industrial company, Department of Information Processing Science, University of Oulu, Finland Lyytinen K, Tahvanainen V-P (1992), Towards the Next Generation of Computer Aided Software Engineering (CASE), in Next Generation CASE Tools, Lyytinen K & Tahvanainen V-P (eds), IOS Press, (1992) McClure C (1989), CASE is Software Automation, Prentice Hall, Englewood Cliffs, New Jersey Nilsson A (1991), In swedish:Anskaffning av standardsystem, Ekonomiska forksningsinstitutet (EFI), Handelshögskolan i Stockholm Orlikowski J Wanda (1988), CASE Tools and the IS Workplace - Some Findings from Empirical Research, in Proceeding of the 1988 ACM SIGCPR Conference, College Park, Maryland, 1988 Programator AB (1990a), In swedish: Handbok Reflex Systemutveckling v1.2, Stockholm Programator AB(1990b), In swedish: Handbok Reflex Verksamhetsanalys v1.2 Wynekoop J L, Conger S A (1991), A review of computer aided software engineering research methods, in proceedings of the IFIP TC8/WG 8.2 Working Conference on the Information Systems Research Arena of the 90's Challenges, Perceptions, and Alternative Approaches, Nissen H-E, Klein H K, Hirschheim R
13
(eds), Copenhagen, Denmark Wynekoop J L, Senn J A., Conger S A (1992), The implementation of CASE tools: an innovation diffusion approach, in proceedings of the IFIP WG 8.2 Working conference on The Impact of Computer Supported Technologies on Information Systems Development, Kendall K E, Lyytinen K DeGross (eds), Minneapolis, Minnesota, USA, 1992 Yourdon E (1989), Modern structured analysis, Prentice Hall, Englewood Cliffs, New Jersey
14