Building Data Confidence and Tools to Improve Decision Making in Kyela

In a PEPFAR Dreams District, one local government leader is transforming his team through better use of data.
Tanzania

Background

In Tanzania, local government authorities (LGAs) play a critical role in delivering ‘frontline’ public services to citizens.
LGAs are however not just a conduit. They shape how, where, and when services such as healthcare, public infrastructure, and development support are offered. LGAs are supposed to base their decisions on data, but—outside of major metropoles—often struggle to find credible local data that is granular enough to be actionable.

Kyela LGA is one of Data Zetu’s focal districts. Kyela’s new District Executive Director (DED) is Musa Mgata. One of his key jobs is to ensure that Kyela LGA is prioritizing the right services, in the right ways. He is also responsible for feeding credible data back to the national government, to improve their resource and development planning.

So, what kind of data is most relevant to Musa?

Kyela is a predominantly rural municipality, near the Malawi border, with an agricultural economy that struggles to maintain direct access to national markets. Flooding is a major issue, regularly destroying transport infrastructure and crops, and causing water borne diseases. Resource constraints mean that citizens are often frustrated about the quality and access to public health and education services. Compounding this, Kyela is also on a major cross-border trucking route, fueling sexual and reproductive health (SRH) and other social challenges.

Musa Mgata, District Executive Director of Kyela District Council, shares his experience and challenges working with data at a round-table and training session in May 2017.

Problem

Watch a video interview with Musa as he describes the data problem and opportunity: https://youtu.be/z4952OG4t7M

• Musa’s biggest challenge, since assuming the role of DED of Kyela LGA not long ago, is “the quality and reliability of data” that his team collects and reports to him. Through interviews with Musa, several root causes for this challenge emerged, including:
• Data collection happens on paper-based surveys, making it hard and expensive for data to be collected efficiently, at scale, or at the level of resolution required to present sufficient information to make informed investment decisions;
• The sharing of data is difficult, as most datasets are stored offline, on papers that are difficult to keep track of. When data is stored digitally, it’s often done so on individual laptops;
  There is a lack among Musa’s team of basic skills on data analysis, cleaning, visualization, and survey design.

These challenges affect different stakeholders. Most immediately, it’s Musa’s team who encounters difficulty turning their collected data into actionable insights—whether that means synthesizing road development updates to report to supervisors, or taking stock on the prevalence of certain diseases across the district’s towns and villages.
And of course, citizens living in Kyela are unable to benefit from well-prioritized, evidence-based decisions by their leaders. This can lead to failing infrastructure, misaligned investments that do not match actual need, and lower confidence or political efficacy in the communities.

Solution

Together with Musa, the Data Zetu team identified several specific activities to help address these problems. It’s important to note that these are only a subset of planned activities to support Musa in addressing his identified data challenges. For instance, in the near future the Data Zetu technology team, led by Code for Tanzania, intends to work with Kyela community groups to pilot a series of medium-term ‘citizen data’ interventions that tackle flooding and water quality challenges. These can help address some of the data gap and quality challenges Musa experiences as the DED.

For now, the activities Data Zetu have implemented to address these challenges are:

• Training LGA officials on data use practices, including basic data cleaning and analysis;
• Training LGA officials on survey methodology, including questionnaire design;
• Developing a centralized data management tool for Musa and LGA administrators;
• Developing a low-cost digital data collection tool for Musa’s team and enumerators;
• Training LGA staff on how to use those tools.

The technology solutions themselves are deliberately designed for interoperability and compatibility with national and wider transnational processes and resources. Code for Tanzania has therefore deployed a customized version of the Open Data Kit (ODK) mobile survey tool, which will feed data into an instance of the (open source) CKAN data management software for Kyela, on a ‘cloud-based’ solution to minimize hosting costs, to guarantee service ‘uptime’ and to enable virtual support from Data Zetu and the wider open data ecosystem.

Tanzania’s Open Data Initiative (TODI) and dLAB’s data portal also run on CKAN, and data can therefore easily be shared between the three portals. In addition to the standard off-the-shelf features, the Data ZEeu technology team led by Code for Tanzania has built in new functionality into Kyela’s platform to allow users (both LGA / national government officials, and citizens) to request and suggest datasets for inclusion. The platform comes with a powerful API (application programme interface) portal, which allows upstream and downstream government users, and anyone else that the LGA specifies, to get automatic, machine-readable access to the underlying data to power their own apps or services. This could include the social entrepreneurs using job or tender alert bulletin data to run public services, or water or weather data for civic alerts or agricultural news services, etc.

The platform comes with a powerful API (application programme interface) portal, which allows upstream and downstream government users, and anyone else that the LGA specifies, to get automatic, machine-readable access to the underlying data to power their own apps or services

Process

Data Zetu has adopted a user-centric design approach to working with the LGA. This means that its tools team (led by Code for Tanzania) has used an iterative development process, consisting of:

• User interviews and on-site systems analysis in Kyela LGA, obtaining detailed specifications from Musa and his team, as well as from a physical inspection of existing data systems, management systems, etc, along with a skills assessment of the LGA staff;
• Developing a minimum viable prototype (MVP) survey tool on ODK and data management platform on CKAN for the LGA, for robust testing and feedback by Musa and his team;
• Iterative improvements to the MVP, along with additional feedback from Data Zetu’s partners at the dLAB and stakeholders in national government.

This—like all prototyping approaches—was not a straightforward or linear process. For instance, Musa initially insisted on custom-built technologies that his internal team could easily maintain and expand with their current skillsets. It quickly, however, became apparent that such bespoke tools would be incompatible with the existing subnational and national data ecosystem which uses CKAN as its minimum benchmark. Musa’s concerns about hosting costs (for the more complex CKAN) and technology skills were both addressed via hosted (cloud-based) solutions, removing the burden of operating costs and maintenance from Kyela. Adoption of CKAN also encourages interoperability with TODI and dLAB, which will help Musa enforce data standards to reduce the “data chaos” and lack of data sharing that Kyela identified as key problems.

Parallel to the tools’ development phase, Musa and his colleagues at the Kyela LGA have participated in trainings facilitated by Data Zetu and the dLAB. This includes a week-long workshop focused on basic data skills in May, and on-site mentoring in Kyela by Data Fellows in October. These activities are referenced in a separate Data Use Story developed jointly with dLab.

Outcomes and Impact

Although the digital tools that Data Zetu developed in cooperation with Musa have yet to be deployed, there are early indicators suggesting that the data skills trainings are playing a role in fostering better data use in the Kyela LGA. This is important, because technologies to improve data management are only impactful if their users perceive value in using that data to inform their decisions.

In that respect, we are seeing progress. In trainings delivered to stakeholders including representatives of the Kyela LGA, over 70% of participants of trainings report positive change in their ability to engage with data. Perhaps more importantly, 60% reported a positive change in perceived value of data, suggesting that the role that data can play in local decision-making is viewed as more important and useful to those local decision-makers. As the digital tools for data collection and management roll out, making data more accurate and accessible to data operatives within the Kyela LGA, we anticipate that these numbers will increase.

The job is not nearly done, however. More trainings are planned, and more technologies can be developed to support the LGA’s objectives. Musa has for instance proposed other potential solutions. He told us: “As a District Director, I need a dashboard of data for all the wards [in my jurisdiction], to save time and help me make informed decisions or proposals on time. It will also help us on accountability.” This may pave the way for a roadmap of even more sustained engagement with Musa’s team over time.

“As a District Director, I need a dashboard of data for all the wards [in my jurisdiction], to save time and help me make informed decisions or proposals on time. It will also help us on accountability.”

Key Collaborators

Data Zetu aims to empower communities to make better, more evidence-based decisions to improve their lives. Through partnerships and collaborations with local communities, Data Zetu works with stakeholders to build skills and develop digital and offline tools that make information accessible to everyone.