Confidence Intervals and Validation of a Forecaster of Quality Navy Enlistments,
Abstract
A substantial amount of effort has been expended in the last few years to attempt to be able to improve the forecasting of the number of quality recruits that will enter the various services over some given period; the 'quality' label refers to those supply limited recruits with High School Diplomas and/or those scoring in the Upper Mental categories on the AFEES exams. The explanatory variablles used in the forecasting models are the levels of key resources such as recruiters and advertising levels of different types, and key demographics such as the unemployment rate, the number of male high school seniors, etc. As numerous and diverse as these efforts have been, very few have been subjected to rigorous types of validation and none, to the knowledge of the authors, has yielded rigorous confidence intervals. This deficiency was one of the major criticisms of the discussants at the ONR Personnel Supply Models Workshop in late January, 1981. The need for statistical confidence intervals is of course to put into proper perspective the single point estimates generated, and to quantify the uncertainty or risk remaining. Decision makers are then in a position to apply their own risk preferences and to factor in the non-quantifiable considerations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1982
- Accession Number
- ADA118265
Entities
People
- John M. Mccann
- Richard C. Morey
Organizations
- Duke University