Given the newness of the research area and of the technology being investigated, it was considered appropriate to conduct an exploratory case study [53
], undertaking hypothesis generation (concerning the feasibility of community-level spatial data access through a web-based open-source GIS) rather than hypothesis testing. The collection of "rich" data from multiple perspectives is prioritised over rigorous system testing. From this perspective, the prototype system is as a tool that promotes discussion and facilitates exploration rather than an artefact in itself.
The technical feasibility of developing the system proposed using open source software was investigated by developing a prototype. The functionality of the prototype is described below. The prototype was subjected to a performance test and was also used to conduct usability tests.
Functionality of the prototype
The prototype system can be used for the following purposes:
• To access spatial information from a central database by local communities
• To upload spatial information collected by communities
• To access spatial information collected by communities from other government levels
Local community access to spatial information
The GIS prototype uses choropleth maps to present indicators of access to basic facilities in local communities. These data are also presented in a table.
Collection of spatial information by communities
The GIS can be used by communities to collect their own data. Through a user friendly interface, end-users can enter attribute data for selected areas. These data are then stored in the central database. This allows the users to analyze their situation, develop a clearer understanding of local needs and gain bargaining power in budget discussions through the display functionality of a GIS. However, one of the key aspects that will require attention in future research is the factor of maintaining data at community level.
Presentation of spatial information collected by communities to other government levels
Other government levels can gain access to the data entered by end-users at local community level. This can help to verify data at community level and helps to fill data gaps. Communities often have a different experience of the problems they are facing and collecting information that is relevant to the community shows higher decision making levels the priorities of a community.
To be able to determine if the system would be able to cope with a reasonable number of users, the system was subjected to an automated load test. During this load test, an increasing number of requests were produced to simulate an increasing number of users. This allowed conclusions to be drawn with regard to the performance of the system. The average response time of the system can be calculated to measure performance.
The performance of the system was assessed using the Apache JMeter software. JMeter is a desktop application which can be used to load test functional behaviour and measure performance of amongst others web applications (both static and dynamic) and databases [54
]. The specifications of the hardware used during the performance tests can be found in Table . Both the application server and the database were running on the same PC, eliminating network latency effects. Characteristics of the spatial data used during the performance tests can be found in Table .
Specifications of PC used during performance tests
Data used during performance tests
A usability evaluation was used to investigate the second research question. A usability evaluation involves testing the ability of a system to allow users to complete their tasks effectively, efficiently and enjoyably [55
The usability evaluation consisted of the following parts:
• Two tasks the users had to perform. The tasks were consistent with the typical use of the prototype. Task one and task two involved identifying an area at the lowest available scale and changing the attribute data of that area. The only difference between these two tasks was the area the participants had to identify. Areas were assigned randomly to participants. The task analysis allows for the identification of problems users may have when using the system. The time the participants needed for each task was measured using a stop watch. There were never more than two people completing the usability tests at once, allowing the facilitator to record accurate times
• Directly after the usability tasks were completed, the participants completed a System Usability Scale (SUS) test in which the participants rate the system to measure the overall usefulness, ease of use and appropriateness of the system. The SUS test is designed to gain a universal and subjective assessment of a system's usability [56
The results from such tests are generally used to provide comparative values indicating the progress of the system meeting its users needs, and are used to monitor such progress during the development cycle of a system. Formal statistical analysis is therefore not applicable to the results gained from the usability tests, and were not used in the analysis of the results.
A convenience sample of users used for tests at the University of Cape Town (UCT). In order to gain a more complete picture of the usability of the prototype, usability testing was also carried out with potential users of the system in a case study of a South African municipality.
The municipality chosen for this case-study was the Overstrand municipality which is located in the south eastern extreme of the Western Cape province of South Africa. The Overstrand municipality has a population of about 80 000 people, but this is growing rapidly, with a population growth rate of 50% from 1996 to 2001, which is expected to continue in the future.
The following potential user groups participated in the testing of the prototype:
• Municipal officials including those who deal with planning of services and service provision, those who deal with community involvement in decision making and those responsible for the ward committees.
• The clinics, represented by clinic personnel, those responsible for the management of clinics and sisters at the clinics
• Community leaders, represented by community organisation members and members of ward councils within the study area
The municipal structure of the study area is representative of other small or medium sized municipalities. The participants were selected based on being representative of those who would use a system as the one described in such a municipality. Together, these groups represent all users working with spatial information in the study area.
All of the local municipalities within the Overstrand District were contacted telephonically (additionally by letter or email if requested), and meetings were arranged with area managers. Only one of the local municipalities within the Overstrand District refused to be involved in the testing. Area managers participated from all the municipalities, community officers and information managers also participated from the district municipality level. The sample of municipal officials represents the potential functional users and managers of the system described. A list of all the clinics in the district was obtained from the municipality, and after obtaining permission from the district clinic, each local clinic was contacted and a staff member was asked to be available for testing at a predetermined time. The majority of the clinics in the area responded positively and participated in the study. Due to the large number of community organisations, not all were contacted, and only those that were thought to have interest in the system and testing were contacted, few of which chose to participate in the study.
The user profile of the participants is summarized in Table . Participants received no training prior to the tests. They were introduced to the system, and given an introduction to the purpose and goals of it.
Summary of user profile of participants in usability analysis
To gain more insight into the appropriateness of the system in terms of existing structures (the third research question) interviews with potential users in the South African municipality described above were conducted. An semi-structured interview is an interview without a predefined format. The interviewer however did prepare some key questions. It is a widely used method of finding out what users want [57
]. The municipal officials, clinic personnel and community representatives who participated in the usability test, participated in the interviews as well.