180 likes | 308 Views
An Empirical Study on the Utility of Formal Routines to Transfer Knowledge and Experience. Reidar Conradi, NTNU Tore Dybå, SINTEF Tele og data conradi@idi.ntnu.no, phone +47 73.593444 Norsk Informatikk-konferanse 2001 (NIK’01) Tromsø, 26-28 Nov. 2001. Contents. Background
E N D
An Empirical Study on the Utility of Formal Routinesto Transfer Knowledge and Experience Reidar Conradi, NTNU Tore Dybå, SINTEF Tele og data conradi@idi.ntnu.no, phone +47 73.593444 Norsk Informatikk-konferanse 2001 (NIK’01) Tromsø, 26-28 Nov. 2001 NIK’01, Tromsø, 26-28 Nov. 2001
Contents • Background • Research context and questions • Knowledge of routines • Use and updating of routines • Routines as a medium for transfer of knowledge and experience • Discussion and implications • Limitations and recommendations for future research • Conclusions and postscript NIK’01, Tromsø, 26-28 Nov. 2001
Background • Formal routines are emphasized in both quality assurance (QA) and software process improvement (SPI). • At the same time; there is often a considerable skepticism among developers to learn from and adhere to formal routines. • Therefore, this study was initiated to explore how developers and manager perceive formal routines as a medium to express and disseminate knowledge and experience. NIK’01, Tromsø, 26-28 Nov. 2001
Research Context SPIQ - Software Process Improvement for better Quality: • National project, funded in part by the Research Council of Norway, 1997-99 • Three research institutions and 12 IT companies, mostly SMEs. • More than 20 SPI pilot projects were run in these companies. Method: • Interviews with 23 persons (13 developers, 5 QA managers, and 5 SW manager) in 5 companies participating in SPIQ (convenience sample). • Qualitative and quantitative data collected by two last year M.Sc. students, supervised by the authors. • Data analyzed by the students in cooperation with the authors. NIK’01, Tromsø, 26-28 Nov. 2001
Research Questions1 Q1: What is the knowledge of the routines being used? Q2: How are these routines being used? Q3: How are they updated? Q4: How effective are they as a medium for transfer of knowledge and experience? 1with sub-questions NIK’01, Tromsø, 26-28 Nov. 2001
About the same Q1.1: Knowledge of Routines NIK’01, Tromsø, 26-28 Nov. 2001
Most developers (9/13) had a LOW degree of involvement!! Most managers (7/10) had a HIGH degree of involvement!! Q1.2: Degree of Involvement During Introduction of Routines NIK’01, Tromsø, 26-28 Nov. 2001
Q2: Use of Routines • In general, about 50% of the companies’ routines were in use • The more experienced developers used the routines to a lesser extent than the more inexperienced developers do. • It was a common agreement that: • “There is no point in having routines that are not considered useful” • The status of the routines among the software developers was highly divergent: • “The routines are generally good and useful, but some developers are frustrated regarding their use.” • “The system is bureaucratic – it was better before, when we had more freedom to decide for ourselves what should best be done.” • “The routines are easy to use.” • “Routines are uninteresting and revision meetings are boring.” NIK’01, Tromsø, 26-28 Nov. 2001
Q3: Updating of Routines • None of the companies had scheduled revisions as part of the process for updating their routines. • Most answers to this issue were rather vague. • Some respondents explained that revisions were informally triggered. • Other respondents did not know how to propose and implement changes to existing routines. • Respondents from all of the companies said that the employees could participate in the revision activities if they wanted to. NIK’01, Tromsø, 26-28 Nov. 2001
Again, strongly skewed, as for Q1.2. Fisher’s exact test = 13.02 (p = 0.002, two-tailed) Q4.1: Routines as a Medium for Transfer of Knowledge and Experience NIK’01, Tromsø, 26-28 Nov. 2001
Q4.2: Alternative Mediafor Knowledge Transfer Similar ranking, but different perception of contents. NIK’01, Tromsø, 26-28 Nov. 2001
Degree of Involvement vs. Assessment of Formal Routines Again, strongly skewed, as for Q1.2 and Q4.1. Fisher’s exact test = 14.71 (p < 0.0005, two-tailed) NIK’01, Tromsø, 26-28 Nov. 2001
Discussion of Results • Occupational culture • Managers and developers don’t seem to understand each other • This leads to failure in organizational learning. • Participation • All of the companies had a low degree of employee involvement in the development of their routines. • Consequently, the routines were seen as coercive and not enabling. • Situated learning • Significant learning should not be divorced from its specific context. • Developers viewed stories and tacit social activities (e.g. newsgroups and discussion groups) as more flexible, adaptable and relevant than formal routines. NIK’01, Tromsø, 26-28 Nov. 2001
Knowledge “Conversion”1 1Nonaka and Takeuchi (1995). NIK’01, Tromsø, 26-28 Nov. 2001
Implications • Studies of the effects of idealized formalizations should focus on their relevance for actual routines and their implementation. • We must learn how to establish better cross-cultural dialogues between the different occupational cultures. • Managers should balance discipline and creativity, in order to supplement formal routines with collaborative, social processes. • Both managers and developers should maintain an open dialogue with feedback regarding the utility of formal routines. NIK’01, Tromsø, 26-28 Nov. 2001
Limitations and Recommendationsfor Future Research Validity: • The small sample and lack of randomness in the choice of respondents may be a threat to external validity. • A major threat to internal validity is that we have not assessed the reliability of our measures (e.g. degree of involvement) Future studies: • Should examine the enabling features of formal routines in much more detail; both cross-sectional and longitudinal studies. • Should include a multiple respondent approach to cover all major occupational cultures. • Should do supplementary, ethnographic studies on how developers really work and how their work relate to formal routines NIK’01, Tromsø, 26-28 Nov. 2001
Conclusions • Developers in general are positive to experience transfer and guidelines. • Formal routines alone is not an efficient way to transfer knowledge and experience. • Large differences between managers and developers regarding the perception of formal routines to transfer knowledge and experience. • In spite of a small sample, we think that the results are representative for a large class of software companies. • A more cooperative and open work atmosphere, with strong developer participation is needed. • The major and most difficult work remains non-technical, that is, to build a learning software organization. NIK’01, Tromsø, 26-28 Nov. 2001
Postscript • Quality assurance: under-taught in CS curricula. • Timothy C. Lethbridge:“The Relevance of Software Education: A Survey and Some Recommendations”, Annals of Software Engineering, No. 6, 1998 identifies three under-taught topics: • Testing and Quality Assurance • Configuration Management • Process Standards • Tor Stålhane, IDI: further interviews last summer in 15 Norwegian software companies on attitudes to formal QA systems: • Same pattern: disrespect for formal routines. • Quote:”Oh, quality assurance – that is what those people in that other department is doing”. • So you will here more from us! NIK’01, Tromsø, 26-28 Nov. 2001