Square Peg in a Round Hole:Developing Student Retention Models in Programs Designed for Adult Learners
Rachael DenisonDirector – Enrollment Research, Strategy, & Data
Management
Matthew HendricksonAssociate Director – Strategic Enrollment Research
October 24, 2013
Agenda
• Intro• NU & CPS• Context• Retention definition• Reports• Analysis of Indicators• Findings: W/I’s & Interventions• Next Steps
About Northeastern – a top tier private research university
Northeastern University• 7 colleges, 2 schools• 16,385 full-time UG students• 4,202 full-time GR students• Signature co-op program• Boston, Charlotte, Seattle, Online
College of Professional Studies (CPS)• Certificate Doctoral degrees• Faculty of scholar-practitioners and industry
professionals; practitioner-based degree programs• 67 degrees offered online• 11,000 undergrad, grad, English language and
international pathway program students
Need• Mission-driven• Predict & stabilize
enrollments• Financial impact• Focused recruitment
strategies• Focused advising &
student support services
Challenges• Standard measures
don’t fit• Stop outs vs. drop outs• Work, family impacts• Application predictors
limited (no standard test scores)
Our square peg/round hole predicament
SM1
SM2
FL1
FL2
WN1
WN2
SP1
SP2
Fiscal Year
FL1
FL2
SP1
SP2
Fiscal Year
Retention Defined
Return
Graduate
SuccessFY
Cohort
Combined UG Report (Masked Data)Initial Year 2nd Year 3rd YearCohort Start
SizeReturn % Grad % Success % Return % Grad % Success%
FY05 150 50% 5% 55% 30% 5% 35%FY06 200 55% 5% 60% 35% 10% 45%FY07 300 60% 5% 65% 40% 15% 55%FY08 500 65% 5% 70% 40% 20% 60%FY09 550 70% 5% 75% 45% 20% 65%FY10 850 65% 10% 75%FY11 800
Combined UG Report (Masked Data)Initial Year 2nd Year 3rd YearCohort Start
SizeReturn % Grad % Success % Return % Grad % Success%
FY05 150 50% 5% 55% 30% 5% 35%FY06 200 55% 5% 60% 35% 10% 45%FY07 300 60% 5% 65% 40% 15% 55%FY08 500 65% 5% 70% 40% 20% 60%FY09 550 70% 5% 75% 45% 20% 65%FY10 850 65% 10% 75%FY11 800
Combined UG Report (Masked Data)Initial Year 2nd Year 3rd YearCohort Start
SizeReturn % Grad % Success % Return % Grad % Success%
FY05 150 50% 5% 55% 30% 5% 35%FY06 200 55% 5% 60% 35% 10% 45%FY07 300 60% 5% 65% 40% 15% 55%FY08 500 65% 5% 70% 40% 20% 60%FY09 550 70% 5% 75% 45% 20% 65%FY10 850 65% 10% 75%FY11 800
Predictive Analytics
• Goal = Predict Success
• Sample– Fiscal Years: 2009, 2010, 2011– Graduate Students
• Analytical Methods– Decision Trees– Neural Networks– Regression
Data Types• Application:
– Employment history– Military service
• Enrollment:– First Term:
• GPA• Withdrawal or
Incomplete
– Special programs
• Demographic:– Age– Ethnicity– Location– Gender
• Financial Aid:– Application– Granted– Type– Amount
Data Types• Application:
– Employment history– Military service
• Enrollment:– First Term:
• GPA• Withdrawal or
Incomplete
– Special programs
• Demographic:– Age– Ethnicity– Location– Gender
• Financial Aid:– Application– Granted– Type– Amount
Initial Findings
• Surprised of weak predictive value– Data inconsistencies– Small samples– Financial aid– Demographic information
• Good outcome – Withdrawal or Incomplete– Small sample, but large impact– Early alert– Student outreach
Next Steps
Further build out of predictive analytics• Overlay admissions,
LMS, early alert, call center data
Monitor other measures of retention• Within terms• Term 1 to term 2
Implement informed retention strategies• Measure impact of
strategies