U.S. Army Reenlistment and Extension by Occupation: A Reduced-Form Trinomial Probit Approach

Abstract

This report presents estimates to the Lakhani-Gilroy model of the extension and reenlistment choices of first-term Army enlistees, using the reduced-form trinomial probit approach developed by Terza (1985). The principal advantage of the probit framework over the multinomial logit estimator used by Lakani and Gilroy (1984) is that the former does not impose the theoretically restrictive assumption of the independence from irrelevant alternatives (IIA). The IIA assumption is especially troublesome in the context of the present study, since it is likely that the extension and reenlistment alternatives are perceived to be closer substitutes than either extension and separation or reenlistment and separation. Consequently, the multinomial logit model would almost surely predict too high a joint probability of extending or reenlisting. The probit estimates were compared with those obtained with the multinomial logit estimator and a variant of the latter, which poses the constraints on the reduced-form coefficients that are implied by the structure of the underlying utility equations. Unfortunately, the model performed poorly across all three estimation techniques, so it was difficult to evaluate them unambiguously. Consequently, further development of the theoretical model and an investigation into the computational feasibility of a constrained probit estimator would seem to be warranted. Keywords: Mathematical models; Mathematical predictions; Reenlistment.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1986
Accession Number
ADA182139

Entities

People

  • Joseph V. Terza
  • Ronald S. Warren

Organizations

  • University of Georgia

Tags

Communities of Interest

  • Biomedical
  • Energy and Power Technologies
  • Human Systems
  • Weapons Technologies

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Computer Programs
  • Databases
  • Economics
  • Employment
  • Enlisted Personnel
  • Equations
  • Estimators
  • Management Personnel
  • Manpower
  • Military Personnel
  • Military Research
  • Personnel Management
  • Probability
  • Recruiting
  • Social Sciences

Readers

  • Mathematical Modeling and Probability Theory.
  • Naval Personnel Management
  • Regression Analysis.