A Mobile-Based Data Collection System for Researchers
Chemwolo, Leslie Kimaiyo
MetadataShow full item record
Data collection by use of questionnaires is the most popular method of data collection. Paper-based data collection is however prone to challenges such as the fact that it is tedious, insecure and is susceptible to data inaccuracies. This study has explored existing mobile-based data collection tools with the aim of identifying gaps in the use of these tools for data collection. Analysis of the requirements for a mobile-based data collection tool has been performed. A prototype meeting the identified requirements and addressing gaps in existing tools has been designed, developed and tested. The tool was subjected to pilot use in a real data collection exercise in Kisumu and Eldoret towns of Kenya. A survey was done to understand if the prototype developed adequately meets the needs of data collection. The prototype allows researchers to create questionnaires using a web-based application and to publish them to respondents through two channels: web and mobile application. The target respondents can then respond to the surveys using either a web application or a mobile application. Researchers are able to collect data in offline mode when there is no network connectivity. The application also shows a map of the locations that data collectors visited. It makes use of end-to-end encryption to protect data. The dynamic nature of questionnaires has been catered for by allowing hiding of questions that are not applicable based on previous answers. It allows for encoding of questions and possible answers at questionnaire creation time and provides for downloading of collected responses in CSV format. The results of testing and the responses from the usability questionnaire demonstrate that the tool adequately meets the needs of researchers in data collection. Offline execution makes the application perform fast. Validation of responses both at the client side and the server side increases the accuracy by blocking out bad responses.