09.12.2012 Views

I__. - International Military Testing Association

I__. - International Military Testing Association

I__. - International Military Testing Association

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

The Use of Artificial Neural Networks<br />

in <strong>Military</strong> Manpower Modeling<br />

Jack R. Dempsey, D.A. Harris, and Brian K. Waters<br />

Human Resources Research Organization<br />

A new ides is delicate. It can be killed by a sneer or a yawn: it can be stabbed to death by a<br />

quip, and worried to death by a frown on the right man’s brow.<br />

-Charlie Brower--<br />

The military has been a trailblazer in the realm of manpower modeling and personnel measurement. According<br />

to an old saying, “Necessity is the mother of invention.” Well, due to the formidable recruiting and selection tasks<br />

facing the Services, pioneering efforts have been made and continue to push the military to or past the state of the<br />

art. There arc again innovative techniques which the Services are (or should be) considering to aid military selection<br />

strategies.<br />

<strong>Military</strong> selection policies are a topic of high level interest and scrutiny. Each of the Services sets standards<br />

for selection on the basis of citizenship, age, moral character, physical fitness. aptitude, and education credential.<br />

The latter two entry criteria are the ‘most visible screening mechanisms and the ones which the Department of<br />

Defense (DOD) uses to define and report recruit quality levels to Congress and other interested parties. Aptitude, as<br />

measured by composite scores from the Armed Services Vocational Aptitude Battery (ASVAB), is used to predict<br />

military technical school performance. Education credentials are used for adaptability screening. That is, they assess<br />

the likelihood of attrition, or positively, that a recruit will complete an obligated term of service. Both aptitude and<br />

education crcdcntial standards have been called into question of late by Congressional watchdogs. Actually, the<br />

flurry of interest in aptitude standards dates back to 1980 when Congress learned that between 1976 and 1980 the<br />

ASVAB norms were incorrect. This resulted in accepting hundreds of thousands of recruits who did not meet the<br />

intended minimum aptitude standards. Furthermore, Congress learned, much to its dismay, that enlistment standards<br />

were validated against training performance not actual job performance. Congress continues to inquire: What is the<br />

relationship between aptitude and job performance? And, how much quality is needed to ensure adequate job<br />

performance? A Herculean, on-going, multi-year job performance measurement (JPM) project has provided answers<br />

to the fist question while the answer to the second is in progress.<br />

More recently, education standards have come under attack by Congress and educational lobbying groups.<br />

Currently the plethora of credentials are categorized into one of three tiers based upon attrition rates. Each tier has<br />

differential aptitude standards and recruiting preferences. While education credential is the single best predictor of<br />

attrition, objections to this policy revolve around the fact that many individual members of the non-preferred tiers<br />

arc successful in service and are therefore wrongfully denied enlistment on the basis of group membership.<br />

The dual problems of linking quality requirements to job performance and implementing more cquiuble<br />

adaptability screening methods requires innovation. Classical statistical techniques may not provide the answer.<br />

These military selection questions require more sophisticated and less familiar modeling techniques. Just how do<br />

techniques such as neural networks complement the more common modeling procedures? The performance prediction<br />

and attrition screening applications described below provide at least a little food for thought and may suggest that<br />

a more in-depth look is required.<br />

Linking Standards to Job Performance<br />

This project’s purpose is to bring the Joint-Service Job Performance Measurement/Enlistment Standards (JPM)<br />

Project to fruition. This will be accomplished through four lines of endeavor. First, the military’s recruit selection<br />

measures (e.g., ASVAB) must be related to job performance in virtually all occupations. Second, a methodology<br />

must be developed so that empirical data can inform the setting of enlistment standards. That is, me expecled job<br />

performance of recruits, over their first term of enlistment, should match total job performance requirements. Third,<br />

improved. trade-off model(s) must be developed so that force quality requirements--based on empirically grounded<br />

job performance requirements--are considered along with related costs in the determination of enlistment standards.<br />

Finally, the Services’ personnel allocation systems must be made responsive to empirical information about the<br />

pcrformancc requirements of particular jobs.<br />

The data used for the Linkage Project consisted of 8,464 individual scrvicc mcmbcrs in 24 different occupations<br />

who had been administered hands-on performance tests as a part of the IPM Project. Each record contained ASVAB<br />

25<br />

.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!