Grouping Personnel for Performance Evaluation via Integer Programming

Each year, the United States Air Force Academy in Colorado Springs, CO graduates and commissions roughly one thousand young men and women as Air Force officers. In the spring of their 2nd Class (junior) year, each cadet's record is reviewed by the USAFA General Board, which evaluates all cadets...

Full description

Saved in:
Bibliographic Details
Published in:Military operations research (Alexandria, Va.) Va.), 2007-01, Vol.12 (1), p.5-17
Main Authors: Armacost, Andrew P., Lowe, James K.
Format: Article
Language:English
Subjects:
Air forces
Demography
Human resources
Integer programming
Military operations
Military personnel
Musical intervals
Objective functions
Operations research
Standard deviation
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Each year, the United States Air Force Academy in Colorado Springs, CO graduates and commissions roughly one thousand young men and women as Air Force officers. In the spring of their 2nd Class (junior) year, each cadet's record is reviewed by the USAFA General Board, which evaluates all cadets with respect to their academic, military, and athletic performance and their leadership potential. The ranking that results from this board is used in a variety of decisions, including matching cadets to career fields and assigning cadets to bases. Because the board consists of two rating panels with each reviewing only half the cadet records, it is imperative the two sub-groups of cadets look identical with respect to known performance measures and key demographic attributes. In this paper, we develop an integer programming model that allows us to fairly split the cadets and offer comparisons to a randomized approach. We also address key implementation issues we faced when adopting this model within the institutional Cadet Administrative Management Information System (CAMIS). More generally, this work applies to many personnel situations where it is necessary to define multiple groups that are equivalent with respect to user-defined attributes.
ISSN:1082-5983
2163-2758
DOI:10.5711/morj.12.1.5