In the near term, it seems to me that the Intelligence Community is going to be faced with (at least) three big challenges: Inexperienced analysts, tight time schedules and moving targets. "Accelerated Analysis", a process that emerged from our coursework and research projects here at Mercyhurst, seems to hold the promise to address all of those issues.
This promise largely stems from the work of one of our recent graduates, Mike Lyden, whose thesis, "The Efficacy of Accelerated Analysis in Strategic Level Intelligence Estimates," is now available for full text download here. I was Mike's primary reader and I can say, without hesitation, that his research is both compelling and provocative. The fact that Mike knows how to write also makes it a darn interesting read.
Mike puts his results succinctly in his abstract:
- "This thesis presents the findings comparing the accuracy of strategic-level estimative judgments made under conditions of accelerated analysis by undergraduate analysts at the Mercyhurst College Institute for Intelligence Studies to estimates of similar scope found in declassified National Intelligence Estimates produced by the United States National Intelligence Council. These historical research studies found that not only are the student estimates of greater nuance than their National Intelligence Estimates counterparts, but they were also of statistically equal accuracy."
Mike lays out the need for alternative methods and processes in the very first part of the thesis. Experienced professionals will find little new here but students and academics just entering the field will find a good summary. The literature review gets more interesting when Mike begins to talk about the Accelerated Analysis process in the sections beginning with "A Model Found" (p . 23). He describes, in the remaining sections of his literature review, all of the disparate threads that seem to get tied together by the Accelerated Analysis process. All of this, in turn, points to a set of clearly worded hypotheses and an interesting method for testing them.
I'll only steal a little of Mike's thunder on the detailed results. One of the ancillary findings that emerged from his study was the relative accuracy of statements in National Intelligence Estimates that use words of certainty (like "will") versus those that use words of estimative probability (such as "likely"). The chart with the results is reproduced here but the counterintuitive bottomline is that the NIC was statistically significantly more likely to be correct when it used words of estimative probability than when it used words of certainty.
I don't think Mike would conclude that he has "proven" that Accelerated Analysis works. I do think his results suggest that it is a new process that is worthy of further study, however. Certainly our own, albeit anecdotal, results, imply that he is correct.
What Do Words Of Estimative Probability Mean?