Ford Foundation

This article offers a detailed account from within the Ford Foundation of what it takes to subject a complex social change strategy to genuine evaluative scrutiny. Drawing on the work of the Foundation’s Office of Strategy and Learning (OSL) and its evaluation of the Natural Resources and Climate Change program, it describes what shifts in strategy and learning were necessary, what was required organizationally, and what changed as a result. The article delves into the evaluation process itself, including how OSL balanced the needs and perspectives of Ford leadership, program staff, and grantee partners. It shares how OSL centered equity in selecting an evaluator; elaborates on the relational work required to build buy-in across leadership and program teams; and shows how evaluation findings were used to refine the climate strategy, shaping decisions about what and whom to support and why. The case surfaces lessons relevant to philanthropies working on strategy and evaluation for climate justice and other complex social change efforts, particularly those navigating organizational change, initiating or strengthening evaluation functions, and grappling with concepts like systems change and equity in practice.

Research-practice partnerships (RPPs) can be an effective strategy for bridging disconnects among research, policy, and practice, as they present unique opportunities for rigorous research that produces clear, actionable findings that can directly inform the work of governmental agencies on high-priority topics. This paper describes two interconnected randomized controlled trials conducted by a decade-long RPP between researchers at New York University (NYU) and New York City Public Schools (NYCPS) policy leaders in the Division of Early Childhood Education (DECE), starting at the launch of New York City's universal pre-K system. This RPP navigated opportunities and challenges for conducting rigorous research to evaluate professional learning (PL) for pre-K teachers assigned to "Explore," a math-focused PL approach (Study 1) or "Create," an arts-focused PL approach (Study 2). In each study, we examined impacts on classroom practice and quality (using observation data collected by NYCPS) and children's school readiness (using direct assessment data collected by NYU), relative to content-agnostic PL serving as a business-as-usual control. Results indicated that Explore did not have significant impacts (relative to control), likely because of low levels of implementation. In contrast, Create had several significant and substantively meaningful positive impacts (relative to control) on teacher practice, quality, and children's learning. Overall, our findings suggest that PL implemented by a large district at scale can have positive impacts when the PL model is robust, when PL has clear targets or focus, and when participation in PL is sufficient. We discuss how findings generated by our RPP have informed DECE's work, as well as challenges encountered conducting the research and utilizing findings, particularly during a mayoral transition. The study was pre-registered with Open Science Framework in July 2023.

This study investigates sentiment polarity biases, specifically, differences in how accurately AI models classify positive versus negative reviews across languages and model architectures. Large language models show a negative bias in French and are more accurate on negative reviews, while encoder models exhibit positive bias in Japanese, missing negative reviews that use indirect criticism. These language-specific polarity biases have implications in both social and business domains deploying multilingual sentiment analysis systems.

Background:Underrepresented in medicine (UIM) interns have unique lived experiences that affect their paths to medicine, and more information is needed for medical residency and fellowship programs to better support them.Objective:We describe self-reported differences between UIM and White physician interns in key demographic areas, including household income growing up, physician mentorship, and adverse childhood experiences (ACEs).Methods:Between 2019 and 2021, we administered a diversity survey to incoming medical interns at the University of Minnesota-Twin Cities. Response rates across the 3 years were 51.2% (167 of 326), 93.9% (310 of 330), and 98.9% (354 of 358), respectively. We conducted analyses to compare UIM and White groups across demographic variables of interest.Results:A total of 831 of 1014 interns (81.9%) completed the survey. Relative to White interns, UIM interns had lower household incomes growing up, lower rates of mentorship, and higher rates of experiencing 4 or more ACEs. The odds of experiencing the cumulative burden of having a childhood household income of $29,999 or less, no physician mentor, and 4 or more ACEs was approximately 10 times higher among UIM (6.41%) than White (0.66%) interns (OR=10.38, 95% CI 1.97-54.55).Conclusions:Childhood household income, prior mentorship experiences, and number of ACEs differed between UIM and White interns.