Intended audience: DataKind Volunteers
In the Evaluate Stage, it is essential to revisit the project with an ethical lens. At DataKind, we are always looking for possible unintended data science project issues, challenges, or consequences. This practice includes ensuring the model or data product is not passed to the wrong people, that it is used for good, and that there are no potential biases or disparate impacts against certain demographic groups baked into any models, operational tools, or project results. In order to evaluate the project with an ethical lens after it’s been completed, consider the following:
- Data Security. Are people’s privacy appropriately protected in the way the model, project result, and/or data tool is being used and deployed?
- Project Risk and Ethical Assessment. Is there any change in how the model, tool, or results are now or could in the future cause harm? Is there still a plan in place for mitigating any possible harm and taking the model, tool, or decisions made from the project results offline if harm is found? Are the project results still sustainable, as they were designed to be, or does more need to be done to ensure they can be used?
- Consulting Additional Key Stakeholders. Does the way that the partner organization is using the model or product incorporate accountability to the communities impacted by it, as the project was designed? Is there anyone to consult in the evaluation process to ensure the product, results, or model are being used ethically?
- Data Inclusion. Are the communities impacted appropriately represented in the datasets being used and any dataset outputs of the project?
- Evaluating Bias. Are any demographic groups being systematically excluded or negatively impacted by the way they are classified in the model? Is the team regularly evaluating for bias as part of their model implementation practices? What actions are taken if bias is identified?
- Technical Quality Checks. Are there any additional quality checks that should be conducted for this project, such as whether it is being implemented fairly and equitably as intended?
This list is not meant to be exhaustive; rather, it is a starting point to help you think about how to evaluate the project at this stage with an ethical lens. The questions that come up in this activity will likely cause you to open up discussions with the volunteer team and partner organization staff, and this is extremely valuable and healthy. Since our utmost concern is to do good with data science, it is essential that this step in the project process is done well and thoroughly. If you have any questions or concerns or need any help, please do not hesitate to reach out to your DataKind support team! This is exactly what we are here for.
Contributer(s): Caitlin Augustin, Benjamin Kinsella, Emily Yelverton
Contact us
If you would like to learn more about us, partner with us, or get in touch, email us at community@datakind.org
Subscribe