Search

In the early 1980s, the U.S. Army opened the National Training Center (NTC) as its main U.S. training venue for battalions and brigades. The NTC's primary purpose was training, but because the simulated combat was fairly realistic, many hoped it would provide a son of laboratory for identifying problems and exploring ways to fix them. This article describes a project that studied tactical reconnaissance at the NTC. The project's substantive purpose was to look for problems in the organization, doctrine, equipment, and training for tactical reconnaissance, and to suggest remedies. Its methodological purpose was to learn how to measure reconnaissance activities and how to use those measurements for analytical purposes, neither of which had been done previously. This article gives background on the NTC and on tactical reconnaissance, and then traces the steps of the reconnaissance project. The two main methodological issues are how to gather measurements in venues like the NTC and what kinds of questions such venues can be used to answer. As it turns out, high-tech measurements are generally of little use low-tech measurements are useful for diagnosis (description) and hypothesis generation, predictions and causal inferences being fraught with hazard. Modest as these uses are, the NTC data are nonetheless highly valuable.

For predicting accident frequencies, a succession of log-linear models for Poisson data, some of which include nested random effects, is introduced. By applying maximum likelihood and empirical Bayes estimation techniques to these models, one can incorporate the actuarial notions of risk classification, model-based smoothing, credibility theory, and experience rating under a unified statistical approach to loss prediction. The performance of these methods is evaluated by using accident data from California. [ABSTRACT FROM AUTHOR]

Outlines the schedule of activities in the joint statistical meetings of statistical societies in Los Angeles, California. Topics addressed in the meetings; Speakers in the event; Time and venue of each session.

Evaluates the 1981 version of BMDP computer program in California. Modules of BMDP; Assessment on the features of BMDP models; Lists of model developer of BMDP.

Reports the development of a data analysis program 'UNIX/STAT 5.2,' in San Diego, California. Development of directions maintenance to improve program quality; Compatibility of the programs for Microsoft-Disk Operating Systems and UNIX; Inclusion of the program in manipulating data and statistical summary and analysis.