Intended audience: DataKind Volunteers
DataKind’s best work is done when we are able to include both subject matter experts and local stakeholders in the Design Stage - each group can provide invaluable context and insights to inform project design.
Subject Matter Experts
Use connections across the entire DataKind community and the partner organization to find subject matter experts (SMEs) that might be useful to talk to, run ideas by, or bring into scoping conversations. Consider looking for SMEs in both the project sector and the proposed type of data science solution space.
Broadly, there are four types of SMEs:
- The end users themselves are often the principal domain experts who understand their problem context the best. Include them in the design process when possible, or someone closely connected to them and very aware of their needs and preferences if direct inclusion is not possible.
- Professionals intimately involved in the sector field on the ground are also a useful resource. These field sector experts are essential in deciding what problem to focus on and what is truly needed.
- Data experts in the sectorare great resources for understanding what might be possible with different publicly available datasets and common challenges in the field. Social science academics in the field might also be valuable for this expertise. These SMEs are useful for understanding what has been attempted before to make changes in the past.
- Finally, technical data science SMEsare useful to bring in when conducting specific data approaches and AI applications. Whether it’s scraping text data, NLP, data visualization, optimization, machine vision, mapping traditional machine learning, or anything else, it is best to bring someone into the design process who is an expert in the specific data science field you are considering. Also consider additional computational approaches that may be standard in domain-specific applications, as related to bias mitigation and accuracy.
See if you can connect with at least one SME from each of the four above categories, even if it’s just for a very brief chat! To find SMEs, we recommend reaching out to the DataKind community on slack with more information to ask anyone if they have any relevant contacts and asking the Project Champion at the partner organization.
Impacted Communities
Consulting with the impacted community is essential in building a project that is sustainable and valuable, as the results of the project can go unused without community buy-in. One piece of this is acknowledging our own limited domain expertise, and that we cannot successfully build a solution without equal partnership with domain experts on the ground. Additionally, this accountability provides insight into the local context and culture, allowing us to identify and get past some of our implicit biases and biased perspectives. Note that members of the impacted community are not necessarily the same as the end users mentioned above - for example, for a tool for child welfare case workers, the end users are the case workers while the community impacted is the children.
If the partner organization is open to holding design sessions with the local stakeholders, include them in a way that is mindful of interpersonal dynamics and respectful of their time. Consider participatory design practices that design researchers are starting to use in marginalized communities. Use client journey mapping and human-centered design to understand, from the communities’ perspective, how they are engaging with a partner organization and/or project; obstacles and challenges they face; as well as social, economic, and political pressures that may need to be accounted for in the project design. This can help account for the potential biases and other challenges that might be reflected in the data or included in the analysis without awareness, especially as we evaluate for data inclusion and exclusion risks.
Design with the communities impacted by your work that have historically been disadvantaged by technology and follow their lead as you develop new technologies with them. At DataKind, we listen closely to and hold the highest respect for the value of underrepresented points of view.
Resources for further reading:
Contributer(s): Afua Bruce, Mitali Ayyangar, Benjamin Kinsella, Cassy L Cox, Nathan Banion, Rachel Wells, Caitlin Augustin, Emily Yelverton, Daniel Nissani, Phil Azar, Mallory Sheff
Contact us
If you would like to learn more about us, partner with us, or get in touch, visit our website or email community@datakind.org.
Subscribe