An Analysis of Educational Credentials and FirstTerm Attrition

1

• An Analysis of Educational Credentials and
  First-Term Attrition
• Christopher Bownds
• Stephen Mehay
• Naval Postgraduate School, Monterey, CA
• Presented to International Military Testing
  Association, Bruxelles, October 2004.

2
Background

• Prior research has linked education and military
  performance, especially attrition
• DoD developed Three-Tier Classification System in
  1980s
• TIER I Primarily Traditional High School
• Graduates
• TIER II Alternative Credential Holders
• TIER III Non-High School Graduates
• The Three-Tier scheme is the basis of the Navys
  current Recruit Quality Matrix

3
Background

• However, since the development of the Three-Tier
  classification system in the 1980s, there has
  been an explosion in the number of educational
  certificates, credentials, diplomas, and
  institutions offering them.

4
Research Questions

• What is the effect of alternative education
  credentials on recruit performance (attrition)?
• Is the Navys Current Recruit Quality Matrix
  still an effective screening tool?
• Are there differences in the predictive ability
  of the different education credentials?
• Would a revised matrix improve the screening and
  selection of applicants?

5
Recruit Quality Matrix
Prior to FY04
6
Methodology

• Predict first-term attrition
• 1 year
• 2 years
• 3 years
• Predict attrition at Boot Camp (RTC)
• Estimate multivariate (Logit) predicting models
• Estimate attrition probabilities based on AFQT
  scores, education credentials, and other
  individual characteristics

7
Methodology (2)

• Analyze Boot Camp Attrition
• Data provided by Navy Recruiting Command (CNRC)
• Navy enlisted cohorts for FY 1989-2003
• 140,000 Observations
• Analyze First-Term (4-year) Attrition
• DMDC data
• Navy enlisted cohorts for FY 1989-1997
• 400,000 Observations

8
Methodology (3)

• Steps
• Validate existing matrix
• Test for differences in predictive accuracy of
  education credentials within tiers/columns
• Develop new matrix to Improve predictive ability

9
Results from First-Term Attrition

• Current matrix still valid Attrition
  probability18-pts higher for Tier II-III
  recruits than for Tier I

Difference in Attrition Probability (Relative to
Tier I)
10
Results (2)

• Nonetheless, significant variation within
  tiers/columns

Difference in Attrition Probability (Relative to
High School Grad)
11
Summary cont.

• Potential gains from new Matrix
• Improve Accuracy of Recruit Screening
• Reduce Type I Errors (False Positives)
• Reduce Type II Errors (False Negatives)

12
Predicted Probability ofFirst-Term Completion
Educational Credential
AFQT
Not a Traditional High School Graduate
13
Matrix Improvements
Educational Credential
Area 1 High Aptitude GED Recipients Currently
B-Cell but, Completion Rates Similar to Cu-Cell
(gt50)
AFQT
14
Matrix Improvements
Educational Credential
AFQT
Area 2 Middle Aptitude Traditional HS
Graduates Targeted because Tier I, AFQT gt
50 but, Completion Rate lt 60
15
Matrix Improvements
Educational Credential
AFQT
Area 3 Tier I Alternative Credential Holders A
and Cu-Cell Targeted because AFQT gt 31 but,
Completion Rate lt 50
16
Matrix Improvements
Educational Credential
AFQT
Area 4 Medium Aptitude B-Cell Recruits Can
Enlist because AFQT gt 50 but, Completion Rate lt
40
17
Matrix Improvements
Educational Credential
AFQT
Area 5 Low Aptitude Traditional HS
Graduates Cannot Enlist because AFQT lt 31 but,
Completion Rate gt50
18
Conclusions

• Benefits of New Matrix
• Increase predictive accuracy of applicant
  screening process
• Reduce costs associated with first-term attrition
• In some cases, identify low-cost alternatives to
  traditional high school diploma graduates