Event Details:

When organizational leaders and their funders want to study if a social program is making an attributable and significant difference, they turn to program evaluation. The method of choice is to conduct a controlled comparison group experiment. This evaluation method is very costly and requires gathering new data over a long period of time, waiting months to years to draw conclusions. There is another way. In today’s big data world, more and more social programs are using database technologies to track, monitor and manage cases. This database proliferation is resulting in the rapid growth of participant, program delivery and outcome data. Alongside these data advances have arisen the development, refinement and advancement of cutting-edge analytics, including machine learning algorithms, that can produce remarkably accurate predictive and prescriptive insights. In fields like healthcare and economics, machine learning algorithms are now being trained to evaluate causality, including figuring out what works for whom. We are now applying these causal modeling tools and techniques in the social sector. This session will present an overview of how machine learning can be trained to evaluate causality of social programs, including sharing case studies where machine-guided causal modeling has been applied to program data from child welfare, workforce development, mental health and juvenile justice agencies.

Speaker Info:

Peter York is Principal at BCT Partners. He has over 20 years of experience as a consultant and researcher in the evaluation and nonprofit fields, as well as a national spokesperson for social impact and impact measurement issues. He has spent the last seven years developing analytic techniques that leverage machine learning algorithms and big data to create predictive, prescriptive and rigorous evaluation (causal) models and tools for social change agents in many fields, including child welfare, juvenile justice, workforce development, adult justice and child mental health. York recently co-authored a peer-reviewed article in the journal Children and Youth Services Review, entitled, "Predictive and prescriptive analytics, machine learning and child welfare risk assessment: The Broward County experience." He also co-authored a book chapter, “The Application of Predictive Analytics and Machine Learning to Risk Assessment in Juvenile Justice: The Florida Experience.” All of these projects train machine learning to control for selection and other biases by identifying block matched comparison groups, finding counterfactuals within these groups, and determining precisely what works for each segment of the population. In addition to his data science work, he has designed and led numerous research and evaluation studies with private philanthropies, corporations, nonprofit organizations and government agencies. He has authored book chapters, academic and professional articles, and a book on the topic of evaluation – “Funder's Guide to Evaluation: Leveraging Evaluation to Improve Nonprofit Effectiveness”. He is a popular speaker on evaluation, capacity building and data science/analytics, presenting regularly at national and regional conferences throughout the U.S. He was the principal designer of nationally-recognized assessment tools and automated, data-driven evaluation tools like the Core Capacity Assessment Tool (CCAT), the Service Enterprise Diagnostic (SED), the impact Capacity Assessment Tool (iCAT), and the Youth Development impact Learning System (YDiLS). Mr. York has served as an advisor, member or board member of social impact groups such as the Alliance for Effective Social Investing, Reimagining Service/Presidio Institute, Data Analyst for Social Good, the Alliance for Nonprofit Management, and the Social Innovation Fund. York has recently become a Leap Ambassador, a private community of nonprofit thought leaders who are committed to the adoption of high performance in the nonprofit sector.