Report calls for improving US intelligence analysis using new methods, research from behavioral, social sciences

The report, which was requested by the Office of the Director of National Intelligence, urges the intelligence community to routinely evaluate the performance of its analytical methods. One important step in that direction is to attach whenever possible numeric probabilities with uncertainty estimates for the events that analysts assess and forecast. Without explicit quantifiers, analysts cannot communicate their conclusions clearly or evaluate the accuracy of their analyses over time. Policymakers need to know how confident analysts are and how well they understand the limits to their knowledge, the report emphasizes. It recommends many specific steps that DNI can implement as part of analysts' everyday work.

"The social and behavioral sciences have long studied topics central to analysts' work, such as how people evaluate evidence and collaborate on difficult tasks," said Baruch Fischhoff, chair of the committee that wrote the report and professor of social and decision sciences and of engineering and public policy at Carnegie Mellon University, Pittsburgh. "That research has had some impact on that work. Our report shows how the community can take full advantage of that research -- and of its dedicated analysts -- by adopting an evidence-based approach to its own analytical methods. We envision a community engaged in continual learning, both absorbing scientific research into the analytical process and evaluating its own performance."

Improving Analysis and Prediction

By evaluating its analytical methods, the intelligence community can learn which ones work best in which situations and have the agility that a changing world requires, the report says. In order to implement this approach, it is critical that the DNI create incentives that focus on learning rather than determining culpability. For example, the report proposes an "Analytical Olympics," in which analysts and methods compete to provide the most accurate predictions and the most appropriate confidence levels.

Criteria for Hiring and Training

To ensure that the intelligence community has the strongest possible work force, the DNI should use evidence-based methods in its recruitment, hiring, and training, the report says. Given the importance of subject-matter expertise to intelligence analysis, it may be tempting for the intelligence community to overemphasize such expertise when hiring. But research shows the importance of having staff with strong general intellectual skills for organizations that face diverse, changing challenges.

In this light, the report recommends hiring practices that focus on stable individual qualities, such as cognitive ability, complemented by training and feedback to develop job-specific skills. The report strongly recommends that intelligence analysts' training include basic familiarity with the full range of analytical methods with strong scientific foundations, in areas such as probability theory, operations research, and analytical history.

Better Collaboration and Communication

The report applauds the intelligence community's recent emphasis on collaboration as essential to integrating information scattered among individuals and units, pointing specifically to collaborative tools such as A-Space and the Intellipedia. However, these tools, too, should be evaluated. While such tools can help analysts to create self-organizing groups when new problems arise, using them can also be time-consuming, and they may supply analysts with information from unfamiliar sources and of uncertain quality, the report notes. Empirically evaluating them will help the intelligence community see whether the benefits outweigh the risks.

The research suggests other approaches to enhance collaboration, such as providing incentives for sharing information and carefully implemented rotations across units. Paradoxically, increasing specialization within the intelligence community may enhance collaboration, the report says. Publicly signaling that analysts' expertise is limited can make it easier for analysts to turn to others for help and not intrude on one another's domains while also advertising the kinds of expertise that are available.

In communicating with the users of its products -- such as military and political decision makers -- the DNI should use standard protocols that elicit users' specific information needs and provide answers in terms that clearly show what analysts believe, why they believe it, and how confident they are. These protocols, too, should be evaluated and refined using scientific methods.

The report also recommends steps to make behavioral and social science more accessible to the intelligence community. It suggests increasing direct communication between members of the academic and intelligence communities so that they become familiar with one another's perspectives and build personal relationships for future communication.

The committee that authored the report will hold a public briefing to discuss it from 10 a.m. to noon on March 28 at the 20 F Street Conference Center, 20 F Street, N.W., Washington, D.C.

The report was released along with a collection of individually authored papers titled Intelligence Analysis: Behavioral and Social Scientific Foundations. Each chapter introduces readers to a fundamental behavioral or social science approach as it applies to the kinds of complex, uncertain problems facing intelligence analysis. The topics covered include analytic methods, group dynamics, individual decision making, intergroup relations, evaluation, and communication.

The study was sponsored by Office of the Director of National Intelligence. The National Academy of Sciences, National Academy of Engineering, Institute of Medicine, and National Research Council make up the National Academies. They are private, nonprofit institutions that provide science, technology, and health policy advice under a congressional charter. The Research Council is the principal operating agency of the National Academy of Sciences and the National Academy of Engineering.