Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


osu1199129269.pdf (830.26 KB)

ETD Abstract Container

Abstract Header

Meta-uncertainty and resilience with applications in intelligence analysis

Schenk, Jason Robert
http://rave.ohiolink.edu/etdc/view?acc_num=osu1199129269

Abstract Details

2008, Doctor of Philosophy, Ohio State University, Industrial and Systems Engineering.
Uncertainty plays a major and inevitable role in human decision-making. Meta-uncertainty about the uncertainty can also be important but it is generally less studied. Such meta-uncertainty has arisen in medical contexts as researchers and practitioners strive to improve conceptualizations of efficacy and mortality related data. Similar but less studied issues arise in the study of human conflicts, in related intelligence analysis, and in responding to business crises. For any given year, the chance of a new conflict arising between a pair of nation states or “dyad” is generally small even if those nations are “politically relevant” to each other. Predicting “no conflict” is almost always correct. Yet, the probabilities of conflict and their meta-uncertainty can be of great interest to military and civilian planners. This dissertation reviews and synthesizes methods available for both conflict probability prediction and meta-uncertainty estimation. It also proposes Bayesian mixture modeling approaches for these purposes and clarifies their potential advantages in relation to actual human conflict data. Intelligence analysis involves gathering and synthesizing a multitude of different data sources into a coherent explanation of events using adductive reasoning. The outputs often involve predicted probabilities of events, which are commonly used in real time briefings and after action reviews (AARs). Given a variety of time, data quality constraints, it can be important to convey the “rigor” or meta-uncertainty associated with probability prediction. For the context of intelligence analysis, this dissertation provides a visual and systematic framework for convey and document meta-uncertainty for intelligence analysis. This framework is based on the proposed “consequence likelihood” diagrams and can be referred to as “hypothesis scrubbing.” Resilience engineering offers new ways to conceptualize responsiveness and reserve capacity. This dissertation reviews and synthesizes many quantitative measures of system resilience. It also explores the application of a recently proposed “master” stress-strain model to evaluate response alternative to crises at a major call center. A main conclusion is that resilience engineering can be viewed as a response to high levels of meta-uncertainty. Also, the synthesis has illuminated a potentially important concept called the “graceful degradation angle” which rates the system’s ability for self-diagnosis.
Theodore Allen (Advisor)
106 p.
Latent Dirichlet Allocation; prediction interval; stress-strain; consequence-likelihood

Recommended Citations

Citations

  • Schenk, J. R. (2008). Meta-uncertainty and resilience with applications in intelligence analysis [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1199129269

    APA Style (7th edition)

  • Schenk, Jason. Meta-uncertainty and resilience with applications in intelligence analysis. 2008. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1199129269.

    MLA Style (8th edition)

  • Schenk, Jason. "Meta-uncertainty and resilience with applications in intelligence analysis." Doctoral dissertation, Ohio State University, 2008. http://rave.ohiolink.edu/etdc/view?acc_num=osu1199129269

    Chicago Manual of Style (17th edition)

osu1199129269
616
© 2007, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.