HOW TO SUCCESSFULLY USE SOFTWARE PROJECT SIMULATION FOR EDUCATING SOFTWARE PROJECT MANAGERS

HOW TO SUCCESSFULLY USE SOFTWARE PROJECT SIMULATION FOR EDUCATING SOFTWARE PROJECT MANAGERS Patricia Mandl-Striegnitz 1 Abstract A crucial factor for the success or failure of software development projects is the qualification of project managers. But how can future project managers be prepared for their role? To successfully manage a software project, theoretical knowledge is not sufficient. Practical experience is necessary, too. This paper presents a new approach for teaching software project management based on the interactive simulation of software projects. This approach teaches theoretical knowledge on software engineering and project management and also offers students the chance to gather reality-like management experience. One important feature of this approach is to analyse the course and outcomes of the simulated projects to give students feedback on their management strengths and weaknesses. Index Terms Software project management, project simulation, project management education, software engineering INTRODUCTION Even in the late nineties software projects are still plagued by schedule and cost overruns, while product quality is often poor [2], [11]. Improvement activities especially focus on process and technology issues, but investigations show that the success of software projects mainly depends on the quality of project management [5]. Jones [3] states that successful projects are always characterized by effective project planning, progress and quality control, and most of all by qualified project managers. To improve the situation in software development, first of all we must improve the quality of software project management by better education of (future) project managers (called students in the remainder). But how can we achieve this? The traditional way of teaching project management from text books has been proved to be insufficient since it cannot provide practical experience. Performing enough software projects with the necessary complexity for training purposes fails due to the effort required. To solve these problems, this paper presents an alternative approach to teaching software project management. This work is part of our research project SESAM (Software Engineering Simulation by Animated Models). This project aims at teaching software project management by interactively simulating software projects [4]. The idea is that a student takes over the role of the project manager while everything else in the projects is simulated. As experience has shown, the simulation in itself is not sufficient for successfully teaching project management [6]. Hence, I have developed a concept to successfully integrate the simulation of software projects in project management education. It combines project simulation with a feedback session and traditional teaching from text books. Providing students with feedback on their performance in the simulated projects is especially important for meeting the learning objectives. I suppose that feedback sessions would have the same benefit in the context of students software projects. STATE OF THE PRACTICE Investigating the current practice in software project management not only reveals enormous deficiencies, but also shows that inadequate or inferior software project management is responsible for the serious problems of software development (software is late, over budget, and fails to meet the customer s requirements) [2]. The following statements reflect the results of two industrial case studies performed in the late nineties [5]. Most of the interviewed project managers only did a part of the management functions necessary for successfully managing a software project. Being promoted into managerial positions as a reward for their excellent work and technical skills as software developers without getting any training in project management, they had no idea of the set of new skills required people and management skills [12]. They neither had sufficient knowledge of their responsibilities and functions, nor did they know about the techniques available for managing their projects [10]. As a consequence, when faced with complex and intangible management issues, instead of trying to solve them, new project managers preferred to work on technical issues. Most of them dedicated a maximum of 35% of their overall working time to management functions. Consequently, most of the investigated projects were characterized by poor project planning (e.g. no risk management or adoption of cost estimation methods), poor 1 Patricia Mandl-Striegnitz, Dep. of Softw. Eng., University of Stuttgart, Breitwiesenstr. 20-22, 70565 Stuttgart, Germany, mandlpa@informatik.uni-stuttgart.de 1

staffing (e.g. insufficient training of project staff in the application area), and a lack of project control (e.g. no collection and use of important data, no updates of the initial project plan). Project managers did not know about the importance of quality assurance activities. Proved software engineering techniques like reviews, testing, and documentation were either not at all performed or reduced or eliminated when the project fell behind in cost and schedule. In the late nineties, this situation is no exception, but rather the norm [8], [10]. As a consequence, inadequate project management caused 75% of the investigated projects to be completed late with deviations up to 150%. HOW TO SUCCESSFULLY EDUCATE SOFTWARE PROJECT MANAGERS? As stated above, deficiencies in project management are due to the poor qualification of project managers. To improve the situation, managers need to be adequately trained before being responsible for their first software project. Even if project managers would get formal education to learn about the functions and techniques of project management before managing their first project, this would be insufficient: To successfully manage a project, practical management experience is required, too. Therefore, to qualify future project managers, project management training must meet the following learning objectives: Students must learn important management functions as well as techniques (e.g. for project planning or control) for handling these tasks. Both, functions and techniques must be practically applied. Students must experience the positive (or negative) consequences of executing (or neglecting) important management functions. Students should face the complex issues they may have to solve in software projects. For example, they should experience major issues in project planning (e.g. estimating the size and complexities of the software project in order to give a realistic cost and schedule estimate [10]) or in project control (e.g. measuring the current project status). Additionally, they should experience problems like the loss of important team members or late delivery by subcontractors. Students must learn how to settle these difficulties to complete their project successfully. Students should experience the complex relationships in software projects. They must always take into account every possible influence on the project course to come to a sensible decision. Students must have the opportunity to try many different management approaches to learn about the consequences. This implies that they should be able to make mistakes without causing any damage. Students should solely concentrate on their role as project managers. But how can we achieve these objectives? Theoretical knowledge in software engineering and project management can be imparted in traditional teaching courses like seminars, lectures, and tutorials. But as students cannot learn programming without writing programs [7], to gain practical experience students must have the chance to manage software projects. During university education, students usually have no opportunity to take over the role of project managers. Furthermore, to acquire comprehensive experience, they would have to perform many different (large scale) projects. This is impossible due to the effort required. What we need is an alternative approach to software project management education. SESAM: BASIC IDEAS AND GOALS To give students the chance to manage projects, within the SESAM project at the department of Software Engineering of the University of Stuttgart an alternative approach for teaching software project management was developed. The basic idea of SESAM is to interactively simulate software development projects. The simulation is based on models of software projects. All objects taking part in the development process (e.g. developers and documents) as well as possible relations between them (e.g. a developer writes a document) are simulated. Both objects and relations are characterized by attributes (e.g. the developer s experience or motivation). There is only one exception: the project manager is real. This role is taken over by the student. To manage the simulated project, the student must interact with the simulation system. He has to execute management functions like hiring team members, assigning tasks to team members, or controlling progress of the project. The student must act depending on the system s reactions. If the student chooses the right management approach, he can make the simulated project a success. If not, it will be a failure. The SESAM simulator offers the chance to simulate typical effects and critical situations project managers face in software projects. Hence, students experience typical problems and learn how to avoid or solve them. A first simulation model was developed by Drappa [1]. The model concentrates on software developed under contract, mainly focusing on the effects of quality assurance. Since the simulation of a software project takes only a couple of hours, students can try many different management approaches and compare their projects results. 2

Experiments show that when using project simulation without additional training components the learning objectives cannot be met [6]. As a consequence, I have developed an educational concept to successfully integrate the simulation in project management education. This concept will be presented in the following section. A CONCEPT FOR USING PROJECT SIMULATION IN PROJECT MANAGEMENT EDUCATION To successfully prepare students for their future responsibility as project managers, I suggest a teaching concept that consists of six components. It combines project simulations with traditional teaching components and analysis components (called feedback sessions). In the following subsections these components are described in chronological order. In each subsection I give an overview of both the concepts and the educational contents. I will especially focus on the feedback session. Analysing the course and outcomes of the simulated projects to give students feedback on their strengths and weaknesses during project execution is a prerequisite to achieving a learning success, no matter simulation or students software projects are used. Introductory Session In the introductory session the use of the SESAM simulator is explained. The management actions offered by the model used in the particular training course are outlined. First Simulation Session: Planning and Performing the Simulated Software Project First, students get information about the software project they have to manage during the simulation. This information includes important project characteristics (e.g. estimated project size) and the customer s objectives (like available time and budget as well as quality specifications). Second, students have to plan their project according to these objectives, taking account of general conditions which apply to the project. Then they have to manage the simulated project according to their project plan (which they may have to change several times in the course of the simulation). Neither project planning nor decision making are supported by a tutor or any help functions. To meet the educational objectives, this aspect is especially important. Like in real software projects, only the project manager is responsible for project planning, staffing, directing, and controlling. Only if students manage their project on their own, they will perceive the project results as a consequence of their management decisions. This goal implies several requirements to the underlying simulation models. First, to allow students to take over the role of the manager of the simulated project and most important to take this role seriously, the management functions offered by the simulation model must correspond to the actions a project manager would conduct in real software projects. This also implies that the project manager is responsible for determining the project status and if necessary for initiating correcting actions. Therefore, the simulation model provides feedback on the progress of the project and the quality of documents only if the student performs actions for monitoring the software project (e.g. formal reviews). The less effort the project manager spends on controlling the project, the less precise is the information provided by the simulation model. Second, to allow for learning about the complex relationships in software projects, models must cover all project phases (from analysis to system delivery), because defects introduced early in the development process will become obvious towards the end of the project. First Feedback Session: Analysis of the Projects Course and Outcomes Due to the complexity of software projects students need feedback. Therefore, after the first simulation is finished, a tutor analyses the course and the results of each project with respect to particular aspects. Examples of aspects are: Quality Assurance: Have quality assurance activities like reviews or software tests been performed in the project? Did the project manager take care of the consistency between products? Controlling: Did the project manager regularly monitor the project status? Have all milestone documents been formally reviewed by developers (and customer) before continuing with the next project phase? Staffing: Did the project manager take the qualification of the staff into account when deciding whom to hire for the project or whom to assign to a particular task? To prepare the feedback session, the tutor uses the quantitative characterization of both the process and the products available at the end of the simulation as well as a log of the project course. The feedback session is then organized as follows: In a joint session the tutor presents the strengths and weaknesses of project management with respect to the evaluated aspects. To demonstrate the impact of different management approaches on the quality of documents, on time, and on money spent on the project, the results and procedures of selected projects are presented by contrasting them with each other. Different results are explained by the different strategies of the project managers during the simulation. By 3

comparison one can demonstrate positive and negative examples of management decisions in particular situations. That way, students recognize the interaction of many different influences as well as correlations between their management decisions and the project outcomes. To achieve a learning success it is important to reduce the complexity of correlations in the software projects. Therefore, the tutor creates views of the available project data. Each view corresponds to specific aspects which in turn support certain learning objectives. For example, one view focuses on how the effectiveness of quality assurance activities is influenced, while another view concentrates on the influences on consistency between documents. In the feedback session students should participate in explaining the reasons for differing project outcomes. Providing feedback is a precondition to achieve the educational objectives. Only then mistakes can be identified and avoided in the next project [9]. In addition to the joint feedback session students also get individual feedback. Since meeting with each student to personally discuss his performance as project manager would be too time consuming, each student gets an individual printout covering important information about the course of his project and the project outcomes. The project data is edited in descriptive, self-explanatory tables and diagrams. Like in the feedback session, each diagram or table covers a particular view. After the feedback session students must have some time to reflect on their own performance as project managers and on the experiences gained so far. Reflection enables students to realize their strengths and weaknesses and hence leads to a deeper understanding of the functions of project management. Seminar Session After students had the chance to gain reality-like management experience they are more susceptive to information on management functions and techniques. Therefore, it is highly recommended to discuss selected subjects of project management, for example software metrics or risk management. Second Simulation Session: Repeating the Project After some time for reflection students repeat their project. For the second simulation, project characteristics as well as requirements and general conditions remain unchanged. Repeating the same project is especially important. First, students get the chance to implement the findings on project management they conducted from the first simulation. Second, by comparing the project outcomes of both simulation runs, improvements in managing their project immediately become obvious to the students. This feeling of achievement adds to the teaching success and strengthens the modified behaviour. Second Feedback Session: Discussion of Lessons Learned The second feedback session especially focuses on the lessons learned. For all aspects discussed in the first feedback session improvements as well as impairments are presented. The tutor should answer the following questions: In which aspects did students improve in the second project? How did the improvements affect the project outcomes? Are there any remaining problems? EXPERIENCES IN USING THE CONCEPT IN EDUCATION In the winter semester 2000/2001 I implemented a management training course based on this concept. The course was performed as part of a tutorial on software engineering. The undergraduate students participating in the course were in their third semester. So far, they had performed a programming course and attended a lecture and tutorial on software engineering and project management. Altogether 40 students participated in the course. I decided to perform the management course quite early in the curriculum to motivate learning about software engineering and project management. We know that practical experience is a prerequisite that students believe in the importance of software engineering and project management for project success. The students had to simulate a 200-function-point software project within nine months spending a maximum of 450.000 DM. The customer requires at least 95% of the overall functionality to be implemented in the code and to be described in the manuals. The code must not contain more than twelve errors per thousand lines of code, the manuals not more than 0,5 errors per page. Figure 1 gives an example of how information is presented in the feedback session. It shows the sequence of the project tasks performed in one of the projects. This visualization of the project course obviously reveals errors of project management. For example, the project manager of this project decided to enter the next project phase before reviewing the milestone documents. Hence, errors are spread over all intermediate project results. In contrast to this, another project course was presented. Here, all documents had been reviewed and corrected before continuing with the next project phase. Comparison of project outcomes then 4

Spec SpecReview Corr_Spec Design DesReview ModuleSpec ModSpecReview Corr_ModSpec Manuals ManReview Corr_Manuals Coding CodeReview Corr_CodeRev ModuleTest IntegrTest SystemTest Corr_STest AcceptTest Spec Design ModuleSpec Manuals SpecReview Corr_Spec Corr_Manuals Coding DesReview ModSpecReview Corr_ModSpec ManReview CodeReview Corr_CodeRev ModuleTest IntegrTest SystemTest AcceptTest Corr_STest 1999/08/16 1999/09/13 1999/10/11 1999/11/08 1999/12/06 2000/01/03 2000/01/31 2000/02/28 2000/03/27 1999/08/02 1999/08/30 1999/09/27 1999/10/25 1999/11/22 1999/12/20 2000/01/17 2000/02/14 2000/03/13 FIGURE 1 OVERVIEW OF THE COURSE OF THE SIMULATED PROJECT demonstrates the effects of proper quality assurance. Figure 2 focuses on the project manager s capability to hire and occupy team members efficiently. In this case, the project manager started his project with five employees in the specification phase. Team members remained to the project until the end of the project. Since project size was relatively small, some team members were idle most of the time. Especially Richard and Diana have not been to any task since December and January, but still remained in the project team. They only cost money without contributing to the project results. Obviously, the project manager lost track of his employees. Comparing the performance of the project managers in the first and second simulation clearly shows improvements of most of the students in the aspects captured by the simulation model and discussed in the feedback session. Nevertheless, even in the second simulation run the project managers still had problems to control their project and to occupy team members according to their qualification. Even though students reported after the first simulation about the importance of regularly monitoring their project and keeping track of project progress, they still had problems to implement these functions of management when managing the project again. At the end, students were asked about their view on the management training course. Their statements can be summed up as follows: All students are positive about having learned about the functions of project management and the difficulties of successfully managing a software project. They consider the feedback session to be the crucial point of the teaching concept. Most state that without getting feedback they would have had no idea of how to improve in the next project. In any case students welcomed the chance to repeat the project to try different management strategies. Immediately seeing the positive effects of applying what they have learned is very important. APPLYING THE IDEA OF FEEDBACK TO STUDENTS SOFTWARE PROJECTS Obviously, providing the students with feedback on their performance was essential for achieving the educational goals. I suppose that the same is true in students software projects. Students will only learn from their mistakes and understand how they can improve if they get feedback about their project and have the chance to reflect on their performance. Thus, they should get the same kind of feedback after performing students software projects. But to be able to present the project data the way I do after simulation students must collect important data about their project. Of course it is impossible to get the data as accurate and as detailed as from the simulated project. For example, no one can exactly tell the number of defects left in the 5

Axel Bernd Christine Diana Richard 1999/08/16 1999/09/13 1999/10/11 1999/11/08 1999/12/06 2000/01/03 2000/01/31 2000/02/28 2000/03/27 2000/04/24 1999/08/02 1999/08/30 1999/09/27 1999/10/25 1999/11/22 1999/12/20 2000/01/17 2000/02/14 2000/03/13 2000/04/10 documents. Nevertheless, students should at least document the course of their projects (which tasks have been performed, begin and end of each task, involved team members etc.) and the effort spent on different tasks. Tracking the defects found during reviews and testing is also recommended. After finishing the project a tutor analyses this data to discuss the strengths and weaknesses. In case there are several teams that performed the same project, here, too, different development approaches can be presented by contrast to compare the effects. Then, students should reflect on their project and get the chance to try again. OUTLOOK Since my concept for teaching software project management using simulation has proved to be successful, I will conduct further courses. For example, I want to develop training courses for different target groups like experienced software developers in industry. Furthermore, I will apply the concept of tutor feedback and reflection in student projects and report on my experiences. Further information on SESAM and on how to teach a course in software project management using the SESAM simulator is available from the following web page: http://www.informatik.uni-stuttgart.de/ifi/se/research/sesam/ index_e.html. REFERENCES FIGURE 2 PERSONNEL ASSIGNED TO THE PROJECT VERSUS PRODUCTIVE PERSONNEL [2] Gibbs, W.: Software s Chronic Crisis. Scientific American, 9 (1994), pp. 86-95. [3] Jones, C.: Software Systems Failure and Success. International Thomson Computer Press, Boston, 1996. [4] Ludewig, J.: SESAM: Grundidee und Überblick. in: Ludewig, J. (ed.): SESAM Software-Engineering-Simulation durch animierte Modelle. Bericht Nr. 5/94, Universität Stuttgart, 1994. [5] Mandl-Striegnitz, P.; H. Lichter: A Case Study on Project Management in Industry Experiences and Conclusions. in: Coombesi, H. et al. (eds.). Proceedings of the European Software Measurement Conference FESMA 98. Antwerp, Belgium, 1998, pp. 305-313. [6] Notter, A.: Eine Untersuchung zur Wirksamkeit der Projektmanagement-Ausbildung am Simulator. Diplomarbeit Nr. 1724, Institut für Informatik, Universität Stuttgart, 1999. [7] Parnas, D.: Education for Computing Professionals. IEEE Computer, Vol. 23, 1 (1990), pp. 17-22. [8] Paulk, M.; B. Curtis; M. Crissis and C. Weber: The Capability Maturity Model for Software. in: Dorfman, M.; R. Thayer (eds.): Software Engineering. IEEE Computer Society Press, Los Alamitos, California, 1997, pp. 427-438. [9] Schneider, K.: Komm, wir spielen Projektleiter! in: Software Engineering im Unterricht der Hochschulen/SEUH München 1994. Teubner, Stuttgart, Germany, 1994, pp. 118-128. [10] Thayer, R.: Software Engineering Project Management. in: Thayer, R.; E. Yourdon: Software Engineering Project Management, IEEE Computer Society Press, Los Alamitos, California,1997, pp. 72-104. [11] The Standish Group: CHAOS Report. Available online at www.standishgroup.com/chaos.html, 1995. [12] Thomsett, R.: The Care and Feeding of Project Managers. American Programmer, Vol. 11, 2 (1998), pp. 5-11. [1] Drappa, A.: Quantitative Modellierung von Softwareprojekten. Doctoral Dissertation, Shaker-Verlag, Aachen, 2000. 6