tag:blogger.com,1999:blog-2569772432953120875.post4605673873303165849..comments2024-02-22T19:45:29.872-05:00Comments on Sources And Methods: What Do Words Of Estimative Probability Mean? (Part 1 -- Introduction)Kristan J. Wheatonhttp://www.blogger.com/profile/02566135545863154089noreply@blogger.comBlogger7125tag:blogger.com,1999:blog-2569772432953120875.post-88320028839979659942008-02-28T04:53:00.000-05:002008-02-28T04:53:00.000-05:00The anonymous post is mineThe anonymous post is minePeter Butterfieldhttps://www.blogger.com/profile/07107508436506563615noreply@blogger.comtag:blogger.com,1999:blog-2569772432953120875.post-81063900896668593302008-02-28T04:50:00.000-05:002008-02-28T04:50:00.000-05:00I like the notion of estimative probability that u...I like the notion of estimative probability that utilizes the 'odds' values eg 75% + or - 10%. However I dont like the degree of exactitude they suggest - similar to my dislike of the use of the word 'predict(ion)'. We have developed a process that aligns with the legal standards - reasonable grounds to suspect; reasonable grounds to believe; certain knowledge. However we have been asked to attached numerical values to these concepts, ie reasonable suspicion is 25% sure etc. I objected but senior decision makers insisted on the grounds that it made it 'clearer'. So its my observation that decision makers seek certainty in the form of quantitative exactitude.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2569772432953120875.post-9031865047313455462008-02-27T22:20:00.000-05:002008-02-27T22:20:00.000-05:00Certainly a decision maker can play the poet and r...Certainly a decision maker can play the poet and run things by instinct or the seat of his/her pants or whatever his/her whim might be, but that doesn't mean that intelligence can abdicate its responsibility to develop better estimative processes. I'd hate to see numeric values interfering in the readability of qualitative analytic products, but it would be good for everyone to agree what each WEP means. <BR/><BR/>I agree with Kevin that the mid-range often covers too broad a spectrum. Just consider an estimate system that has values like: VERY LIKELY(90-99%), LIKELY(75-89%), and SOMEWHAT LIKELY (51-74%). SOMEWHAT LIKELY/UNLIKELY can easily have half the ballfield. There need to be more gradations, and estimates need to be more precise.Pathttps://www.blogger.com/profile/09082406276423678772noreply@blogger.comtag:blogger.com,1999:blog-2569772432953120875.post-87071898470624000892008-02-27T20:06:00.000-05:002008-02-27T20:06:00.000-05:00Ted,Many thanks! That is an extremely neat find.I...Ted,<BR/><BR/>Many thanks! That is an extremely neat find.<BR/><BR/>It is interesting the degree to which the Joint Pub disagrees with the way the NIC has defined probability and confidence over the last couple of years (at least in the public Key Judgments from the NIEs). The NIC distinctly separates the ideas of probability and confidence while the Joint Pub equates them! I wonder who is correct...Kristan J. Wheatonhttps://www.blogger.com/profile/02566135545863154089noreply@blogger.comtag:blogger.com,1999:blog-2569772432953120875.post-69077709850327988552008-02-27T20:03:00.000-05:002008-02-27T20:03:00.000-05:00Kevin,I agree. I have noticed the same thing. Th...Kevin,<BR/><BR/>I agree. I have noticed the same thing. The recent publicly available NIEs have strongly favored the word "probably" for the likely the same reason.Kristan J. Wheatonhttps://www.blogger.com/profile/02566135545863154089noreply@blogger.comtag:blogger.com,1999:blog-2569772432953120875.post-48664629170822703472008-02-27T15:12:00.000-05:002008-02-27T15:12:00.000-05:00For a quick reference, there's Appendix A "Intelli...For a quick reference, there's Appendix A "Intelligence Confidence Levels" in JP 2-0 Joint Intelligence, dated 22 June 2007. <BR/><BR/>The Joint Pub is readily available for download online and you can find the cited quick reference as figure A-1 on page 118 of the pdf file.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-2569772432953120875.post-11740376488966226752008-02-27T14:47:00.000-05:002008-02-27T14:47:00.000-05:00In my four years as an intelligence studies major,...In my four years as an intelligence studies major, I have noticed that students typically find comfort in using the WEP's of "likely" and "unlikely", myself included. I strongly agree that WEP's make for a better estimate, but have noticed that, at times, the words just above or below 50/50 can cover a broad spectrum, giving the analyst too much room to move within the 0-50, and 50-100% analysis.Anonymousnoreply@blogger.com