Predictive Analytics and Evidence Building

Kelsey Gray, Senior Evidence Analyst, attended the Predictive Analytics for Human Services and Education meeting hosted by the National Academies of Sciences, Engineering, and Medicine and the Annie E. Casey Foundation.  Here she shares her thoughts.
It was a day of intense discussion and engaging conversation about the role of predictive analytics in evidence building.  Three main themes emerged from the meeting that highlight the importance of:

  1. Decision makers: Algorithms can be powerful, but people must be involved in the decision-making. It is critical for practitioners and policymakers to remain engaged in determining practice and policy.  Ultimately, predictive analytics should be used as a tool, among many, that supports practitioners and policymakers in their decision-making.
  2. Change Management: Organizations must adapt to make decisions based on predictive analytics. For predictive analytics to have an impact, we must build organizational cultures that strengthen data-driven decision-making.  Staff at all levels of an organization must see value in and have the capacity to use and interpret predictions to enhance decision-making.
  3. Ethical Considerations: Decision makers must consider the ethical implications of leveraging algorithms to improve practice or policy. It is important to evaluate the model components and the model outputs. Decision makers should consider when it is negligent not to act based on a prediction versus when it is overstepping to intervene based on a prediction. 

As Project Evident helps nonprofits and their funders pilot, build and use strategic evidence plans and predictive tools, these thoughts guide our work:

  • Nonprofit leaders are the conductors. Their insights are critical to appropriate use and “gut checking” of the algorithms’ outputs.
  • Plans and tools  won’t stick unless accompanied by talent and change management support—what we call “foundations for implementation.”
  • Building evidence is an iterative process.  Well funded evidence plans support alignment and transparency with staff, funders and researchers as the organization iterates over time.