Presentation on theme: "Mission, Margin, or Both? How Data Analysis and Modeling Can Help You Find the Balance Mission, Margin, or Both? How Data Analysis and Modeling Can Help."— Presentation transcript:
Mission, Margin, or Both? How Data Analysis and Modeling Can Help You Find the Balance Mission, Margin, or Both? How Data Analysis and Modeling Can Help You Find the Balance Middle States Regional Forum February 2008
Goals of Enrollment Management Understand factors that influence enrollment behaviors Positively influence enrollment decisions, then retain and graduate those who enroll Optimize size & composition of student body, consistent with mission Control expenditures
Trends Leading to Strategic Use of Financial Aid Similar to those which prompted enrollment management models: Intense competition Competing pressures Enrolling desired numbers, quality, mix Achieving revenue goals Maintaining commitment to mission & values All-consuming focus on revenue
Plenty of Pressures - Little Control The need for a data-supported approach to managing enrollment and resources has never been more important. Sophisticated models and tools are needed to fully understand the trade-offs and impacts of various strategies.
Data Analysis Progression Table analysis Aggregate data analysis Segmented data analysis subgroup or individual data Regression analysis individual data & modeling
Typical Enrollment Yield Table Males enrolled at a rate of 30% and had an average GPA of 3.0. Females enrolled at a rate of 20% and had an average GPA of 3.4. MaleFemale Yield Rate 30%20% Mean GPA 3.03.4
Limitations of Table Analysis Is the difference in enrollment rates by gender because men are more inclined to come to the institution or because people with lower GPAs are more likely to enroll? If all admits had the same GPA, how much would gender affect the likelihood of enrollment? If gender were a constant, how much difference would GPA make? A table alone does not provide the answer.
Modeling & Regression Analysis Probability models make projections about the future based on evidence from the past. Regression analysis controls for the influence of multiple variables – it holds other things constant while testing the impact of various factors
What is a Predictive Model? An equation that explains variation in something – such as the probability that a student will enroll. A financial aid model helps you understand the relationship of students characteristics to their ability and willingness to pay.
Example Equation Enrollment = f (SAT, GPA, state, ethnicity, gender, expected major, need, grant) Translation: The probability of enrolling is a function of the following combination of factors (SAT, GPA, etc.).
Benefits of Modeling Pinpoints opportunities to use aid resources thoughtfully and effectively to achieve enrollment and fiscal goals and support institutional values. Makes clear the consequences and choices that result from new awarding policies.
Where weve been Where we are How we got here How to deal with success How research has supported the effort
A Classic Turnaround Story New president arrived spring 1999 – Inspired by Benjamin Rush to transform Dickinson Strategic plan Operational strategies Key performance indicators Budget – in the black & endowment spending controlled Enrollment – doubling of demand for Dickinson Improved academic quality
Transformation in Four Parts Power of leadership narrative Strategic planning Key to transformation Three levels Execution Results
Success by the Numbers improved Quality of student – improved attained Financial stability – attained established & defines our brand Mission, Vision, Goals, objectives – established & defines our brand
What have you done for me lately? How to deal with success – tough to maintain momentum and not burn out New group of aspirant schools = new goals Understanding new financial situation Preparing for leadership change (President, senior executives)
Role of Institutional Research Define the problem or area needing attention Establish timeline & objectives Importance of intermediate objectives Input from statistically savvy & assistance to statistically challenged
Role of Institutional Research Remove black box results Confidence in model required so results will be believed Put technical analysis into common terms Leverage expertise of subject matter experts, (financial aid, admission, enrollment exec) Facilitate decision-making
Role of Institutional Research Keeper of Data in conjunction with others Ensure a common voice Frozen data sets Keep tabs on the various definitions Funnel all data requests (internal & external) through IR Peer group comparisons & benchmarking Difficult to ensure apples are compared to apples LOTS of different groups; admissions overlap, Presidents list, Deans list, other aspirant lists.
Strategic Indicators and Goals Admissions Strategic Indicators and Goals ~ Admissions 2001200320052007 2010 Goal Total Applicants38204633478458446000 Acceptance Rate64%52%43%42%40% Yield25%26%27%25% Total Freshmen611624648621600 Top 10% Class Rank47%50%52%48%60%
Strategic Indicators and Goals ~ First-Year Financial Aid 20012003200520072010Goal % Aided (w/ Dickinson Grant) 61%55%57%48%53% % Aided Need-Based52%47%48%43%45% % Aided Non-Need- Based 11%9%10%5%8% Average Grant$14,160$16,668$19, 203$21,163 Average Grant as % of Comp Fee 44%47%48% Tuition Discount Rate34%32%34%28%35%
Strategic Indicators and Goals ~ Student Body 2001200320052007 2010 Goal Total Matriculant Enrollment 21592235230123492250 % Male42%44% 45% % International1%2%5%6%7% % Minority6%8%13%14%18% % Out-of-State59%65%72%75%
Modeling Starts with Key Questions What questions are you trying to answer? What data do you have available to answer these questions? How do you get started?
What are your Target Groups and Objectives? Number, quality and mix of students: Underrepresented populations Academically gifted, special talents In-state/out-of-state Specific majors On campus vs. off campus Containing expenditures
Potential Data Elements Admission fields: application type and status, quality measures Financial aid fields: grant, need, awards Demographic fields: ethnicity, gender, state of residence Enrollment indicator
Influences on Enrollment Decisions Academics Reputation Location STUDENT CHARACTERISTICS PRICE Net price is the single, easiest factor for the institution to control
Ability and Willingness to Pay Ability to pay based on need Willingness to pay … Perceived value of enrolling …Commitment to institution – students will pay more to attend their first choice …Even if you are their first choice, the price must be perceived as affordable
Policy Implications Evaluating yields of students with different combinations of academic ability and need can help in understanding the relationship between willingness and ability to pay. Institutions can respond to this information through tuition discounting, but should they? Implications for institutional missions and values, fairness, and families perceptions of aid practices.
Competing Priorities VALUES If more funds are directed at middle- and high-income students, low- income students may end up with more unmet need or self-help. REVENUE To focus solely on equity could quickly deplete resources and therefore be fiscally inefficient. BALANCE Focusing on the efficient use of funds may increase the opportunities for the use of aid funds in the pursuit of equity.
Competing Priorities If there arent enough students bringing tuition dollars to campus, there will be inadequate resources for delivering a quality education and inadequate resources for need-based financial aid. Understanding the mutually reinforcing aspects of equity & efficiency lead to the suggestion that equity goals and strategic planning must enter into financial allocation decisions. Sandy Baum
Data Considerations Garbage in, garbage out – not all offices treat data with the same respect. Everyone must be committed to data quality. Data is incomplete or inaccurate No aid offers for non-enrolling students Missing quality indicators (GPA, test scores) or demographic information on a significant segment of the admitted pool External factors not included in the model (what is happening with your primary competitors… their recruiting strategies, their awarding policies?)
Managing Expectations Modeling is not a crystal ball. Predictive modeling is based on the past. The future may look different.
Its Only Data Data and information from modeling + Lessons learned by experienced practitioners + Intuition + Institutional context and values = Well-founded & informed policy decisions
Strategic Use of Financial Aid Resulting aid policies should: Be congruent with mission & values Optimize enrollment of desired mix and number of students Encourage retention & degree completion Blend principles of meeting need with awareness of market realities Seek to balance ideal with practical
Conclusions We will not be able to maintain or increase equity and access in higher education if we do not control costs and use funds efficiently. Without careful analysis of data, any change in policy or practice is equally likely to succeed or fail. Risk can be minimized and mission can be accomplished through data analysis & modeling.
Presenters Mike Johnson Director, Institutional Research Dickinson College email@example.com (717) 245-1019 Anne Sturtevant Director, Financial Aid Solutions College Board firstname.lastname@example.org (571) 262-5991