全球智库信息集成服务系统

Holding a Homeschool or Distance Learning Diploma Instead of a Regular High School Diploma Had Negative Effects on Attrition Rates

• Overall, estimated attrition rates for recruits with homeschool diplomas are slightly, but statistically significantly, higher than attrition rates for recruits with regular high school diplomas. The effect is larger for those with Armed Forces Qualification Test scores lower than 50.
• Holding a distance learning school diploma affected attrition more negatively than did holding a homeschool diploma, regardless of whether the recruit scored above a 50 on the AFQT.

The U.S. military services have traditionally used a tiering system, including education credentials such as high school diplomas, in combination with Armed Forces Qualification Test (AFQT) scores to help gauge the likelihood of a recruit persevering through his or her first term of service. But what about less traditional credentials, such as diplomas earned through homeschooling and distance learning? The Office of the Under Secretary of Defense (Personnel and Readiness) asked RAND to examine whether its current education-credential tiering policy is still useful in predicting first-term attrition. The authors examined attrition rates at 12, 24, and 36 months of service for all enlistees from 2000 through 2011. Using statistical regression techniques, they compared attrition rates for those with distance learning or homeschool credentials to those of high school diploma holders, after controlling for other observable population differences. Overall, the analyses support current tiering policy classifying homeschool diplomas as Tier 1 if a recruit's AFQT score is 50 or higher (i.e., they are treated the same as high school diploma holders) or Tier 2 if a recruit's AFQT score is lower than 50. The results also support classifying distance learning credentials as Tier 2 regardless of AFQT score.

• Chapter One

  Introduction

• Chapter Two

  Attrition Analyses

• Chapter Three

  Study Findings

• Chapter Four

  Conclusions and Recommendations

• Appendix A

  Descriptive Statistics for Key Variables

• Appendix B

  Effectiveness of the Propensity Weights

• Appendix C

  Models Used in Doubly Robust Regression Analysis

• Appendix D

  Simple Logistic Regression Model