My research was conducted using AR. Within AR the research situation determine which concrete research methods to be used. This can be both qualitative and quantitative methods. AR do not have any absolute criteria to determine if the interpretations made are valid. AR research will always be open to new interpretations. To give some idea of how valid my research is I will use the validity criteria mentioned in section 3.1.1.
To reach appropriate conclusions in AR it is important for the research to go through many cycle. At the end of each cycle intermediate conclusions are made. Based on this intermediate conclusions better research questions can be asked. Through my research I have constantly refined my understanding of FLOSS, HISP, Ethiopia and developing countries through direct and vicarious experiences. I have read a lot of literature, I have participated in FLOSS development in Ethiopia and in the DHIS 2 project. In my research I have gone through many small cycles. As Figure 3.2 goes I have only gone through one full circle for each of my two case studies. This thesis is the first real result of the Specifying Learning phase.
Throughout my research I have made use of triangulation. I have made use of both qualitative and quantitative methods to verify my conclusions. I have also used many information sources like my own experiences, the experiences of my coworkers and stakeholders, and various literature. All in all I have been quite rigorous in my information handling. I developed a good relationship with the other members of the Tigray team, the relationship developed with the workers in the Tigray health bureau and in the two district was limited. During the one week training at the health bureau I got a quite good relationship with the workers. They had no problems with asking questions and coming with feature request. In the DHIS 2 case I unfortunate developed a too limited relationship with the others in the DHIS 2 virtual community.
Within AR it is considered important to subjugate research data, methods and interpretations to peer review. In the process of writing this master thesis I have only done this to a limited extent. I made one interview with Knut Staring, which is one of the managers of the DHIS 2 project, to obtain data about DHIS 2 and to verify my assumptions of the DHIS 2 project. In Tigray I naturally talked with the other team members about what we should do, and about what we experienced in our dealings with the health bureau. Hopefully other people will read this thesis and be able to improve my interpretations.
The outcome of the Tigray case was not exactly what we hoped for. DHIS 1.3 never came into regular use, not even in the pilot districts. We succeeded, however, in bringing much needed attention to the data gathering practices in Tigray. Tigray have expressed interest in DHIS 2, and perhaps this system will come into regular use in Tigray. This still remains to be seen. It was not a goal in it self for Tigray to use DHIS 1.3, the goal was to improve the information handling practises, and we assisted the Tigray health bureau in this. It was their choice to not start using DHIS 1.3 in the pilot districts.
When I started on the DHIS 2 case I was hoping to build an ICD code plug-in for DHIS 2. From the experiences I got in Ethiopia I saw the need for such a plug-in to DHIS 2. Unfortunately I never came around to actually build this plug-in. The DHIS 2 project was at an early stage of development and it was not clear how DHIS 2 should facilitate extensibility. I worked on a plug-in framework for DHIS 2, and I managed to make a throwaway prototype of a framework. I concluded that this framework would create more complexity than it would give value in extensibility.
I have most definitely moved towards a better understanding of FLOSS, HISP, Ethiopia and developing countries. I have also moved towards a better understanding of IS theory, and I have improved as a programmer. The knowledge and experiences I have obtained through this research will be of great assistance in a future work situation. The DHIS 2 case have taught me important lessons about software development and I have learned to use interesting technologies. Using AR to do practitioner study on my own work situation as I did in the DHIS 2 case will also make me more self-reflective in a future work situation. My knowledge about FLOSS will make it easier for me to participate in FLOSS projects, and my experiences from Ethiopia have made it easier for me to work with people from other cultures. All in all I have grown as a human being, and I feel prepared to work with programming in various social contexts.
It is more difficult to assess how much the workers in the Tigray health bureau have moved to a better understanding of their organisation and information handling practices. Damitew and Gebreyesus (2005) can tell more about this than I. I do not think I have contributed much to make the other DHIS 2 developers understand FLOSS and the DHIS 2 project better, at least not while I conducted my research. I do not think anybody but myself have moved significantly to a better understanding of FLOSS, neither in the Tigray case or in the DHIS 2 case, through the research process. However, those who choose to read this thesis can learn a lot about FLOSS, HISP, IS theory, Ethiopia and developing countries.
In the Tigray case the research was most definitely done in collaboration with the stakeholders. We offered our services to the Tigray health bureau and they accepted. In the Tigray case we acted as facilitators, facilitating discussions about the data gathering practises in Tigray. We also acted as software developers, configuring DHIS 1.3 for Tigray. User feedback was sought through the training sessions, and by questioning the different departments about the reports DHIS produced or should produce. The DHIS 2 case was a practitioner study and I should have sought different views on how to make DHIS 2 extensible from the other DHIS 2 developers. Both our research in Tigray and my research in the DHIS 2 project was relevant to the local context.