Transcript Document
BIG DATA: OPPORTUNITIES AND CHALLENGES IN TODAY’S COMPETITIVE ENVIRONMENT DR. NANCY SZOFRAN, PROVOST COMMUNITY COLLEGES OF SPOKANE 1 Highpoint: 10.58 in 1973 Current: 5.57 in 2013 2 EMSI Executive Summary January 2011 The Economic Contribution of Washington Community and Technical Colleges 3 Findings: Economic Growth Analysis: • $822.4 million – Income to WA Economy Each Year • $746.6 million – Operations of 34 Community & Technical Colleges • $75.9 – Spending of International Students Economic Impact Analysis at a Glance Added Income College Operations Effect Student Spending Effect Total Spending Effect Student Productivity Effect Total Added Income in Washington ($ Millions) $822.40 $746,568,000 $ 75,869,000 $822,438,000 $10,225,902,000 $10,225.90 GRAND TOTAL Spending Effect Productivity Effect $11,048,339,000 4 2009-2010 $10.2 Billion in State Income Higher earnings of students and increased output of businesses 5 Investment vs Future Income $7.00 $6.00 $5.00 $4.00 $3.00 $5.90 $2.00 $1.00 $0.00 $1.00 For Every $1 Invested Cumulative in higher income 6 Washington benefits from: • Improved Health • Reduced Welfare • Reduced Unemployment • Reduced Crime • Savings to the public of $50.7 million per year 7 Taxpayer Return on Investment Comparative Rates of Return 8.00% 7.00% 6.00% 5.00% 4.00% 3.00% 2.00% 1.00% 0.00% 6.40% 7.00% 3.00% Discount Rate Washington's Stocks & Bonds Community and Technical Colleges 8 Washington Community and Technical Colleges are a Sound Investment Colleges enrich the lives of students and increase lifetime income. Taxpayers see increased revenues from an enlarged economy and a reduction in the demand for taxpayer supported social services. Colleges contribute to the vitality of state and local economics. 9 Total Job Postings in the Health Care Industry, Spokane Region January 2010 – June 2014 3500 3122 3000 2666 2500 2368 2000 1707 1500 1550 1000 806 500 0 2246 2108 1469 1377 1859 1897 1807 2027 1716 2185 2032 1845 Due to the economic growth and improved data-mining software, Burning Glass Labor/Insight ™ recognizes 62 percent more total job postings starting in Q3 2013. The data from Q3 2013 to Q2 2014 has been normalized to reflect this change. 0 10 Digital Footprints 11 Student Transition Information Project (STIP) “Empowering Community Colleges to Build the Nation’s Future” • 41 School Districts • 73 High Schools Enhance the data reporting that guides local and policy-level career and college readiness decision making 12 80 70 60 50 40 30 20 10 0 Active & Student Effort Collaborative Learning SCC 2011 No significant changes in benchmark aggregate scores since 2011 survey Academic Challenge Student-Faculty Interaction Support for Learners SCC 2014 Standardized Benchmark Score* Standardized Benchmark Score KEY FINDINGS REPORT – CHANGE FROM 2011 80 70 60 50 40 30 20 10 0 Active & Collaborative Learning Student Effort SFCC 2011 Academic Challenge Student-Faculty Interaction Support for Learners SFCC 2014 13 NEXT STEPS We will examine these results in more detail throughout the year Experiment with the use of CCSSE item responses as predictors of student success: Identify groups of students who may need additional help May help target the specific kinds of interventions required We will also examine results of the Community College Faculty Survey of Student Engagement (CCFSSE) Perception-matching between students and faculty 14 TODAY — CCFSSE: Online survey administered to the same faculty whose classes were selected for the CCSSE sample – 206 instructors districtwide 96 items that are matched to student items in CCSSE 85-90% are significantly different* We’ll examine items that show some of the greatest difference in perceptions between instructors and students District results, not college-specific 15 HOW STUDENTS SPEND THEIR TIME: Students said they are spending more time preparing for class than faculty believed. About how many hours in a typical 7-day week are spent preparing for class (studying, reading, writing, rehearsing, doing homework, or other activities related to your programs)? Percent Response 50.0 40.0 30.0 11 or more hrs/week Faculty: 31% Students: 42% 20.0 10.0 0.0 1 to 5 6 to 10 Faculty 11 to 20 21 to 30 More than 30 Students ¾ of students said they are not participating in extra-curricular activities at all! Faculty: 90% said 1 or more hour Students: 25% said 1 or more hour About how many hours in a typical 7-day week are spent participating in college-sponsored activities (organizations, campus publications, student government, sports, etc.)? 80.0 Percent Response None 60.0 40.0 20.0 0.0 None 1 to 5 6 to 10 Faculty 11 to 20 21 to 30 More than 30 Students 16 BUILDING THE MODEL – OPERATING PHILOSOPHY Find and use leading predictors of change along with known enrollment data from current year. Winter Enrollment Forecast Summer 2012-13 Fall Winter Spring Summer 2013-14 Fall Winter Use half-year enrollment, plus other summer & fall data Spring Summer Run model, late January 2014-15 Fall Winter Spring Annual Enrollment 17 BUILDING THE MODEL – BEHAVIORAL INFLUENCES We examined dozens of potential economic variables. Variables that panned out: Job-related (Annual employment, Change in annual employment, Net change in jobs, Unemployment rate) Wage-related (Annual total wages, Change in wages, Average annual weekly wages) Tuition (State resident tuition, change in annual resident tuition) 18 BUILDING THE MODEL – VALIDATION CCS Enrollment Forecast for 2014-15 -- Winter Models 62000 Student-Quarters 60000 R = 0.94 58000 56000 54000 52000 50000 48000 ACTUAL FORECAST Model slightly over-estimates upward trend change, and underestimates downward trend change, but only by 2-3%. 19 ANCILLARY FINDINGS Race/Ethnicity and Financial Aid variables were overshadowed by other predictors. Ratio of females to males is predictive for certain groups – some variables serve as proxies for things that can’t be directly measured. Average credit load decreasing more part-time students higher per credit revenue. 20 “An area of statistical analysis that deals with extracting information using various technologies to uncover relationships and patterns within large volumes of data that can be used to predict behavior and events.” 21 Smart Companies: Holistic Approach to Big Data – Strategies That Enable Solutions Predictive Analytics uses data science to build highly predictive models of future outcomes. Predictions based on student characteristics and behaviors 22 How will predictive analytics help our students? Help define new student groups Capacity to predict behaviors from day zero What variables have greatest predictive power Create dashboard of student level data Evaluate existing student success interventions 23 WICHE Big Data Project Student Success This project has been able to specifically identify points of loss. 24 Actionable Models Quantified Intervention Effectiveness Results Closed Loop Field Tests (at-risk) Tutoring Student Services Email Text Message Alerts Institutional Benchmarks Collaborative Community of Experts 25 STUDENT SERVICES QUESTIONS Who are our students? What support services are most effective and in what sequence? What course sequencing is beneficial vs toxic? Early alert system: is the system actionable, meaningful? 26 PREDICT STUDENT BEHAVIORS Learning outcomes Recruitment Retention Aim is to make positive changes throughout the student life-cycle Increase operational efficiency Demonstrate accountability for accreditation Demonstrate positive efforts to legislature, et al. 27 Cannot measure: homesickness, missing girl/boy friend, emotionally unprepared for the freedom of living away from home. 28 LEARNER ANALYTICS Can assignments/ activities be a proxy for engagement? Successful behaviors in a class Course sequencing Rate of student progress Features of the learning environment that lead to better learning 29 LEARNER ANALYTICS, CONT. Impact of attendance Indicators of satisfaction and engagement Classroom – virtual or traditional Keeping the most personal aspects of teaching in place. 30 CHALLENGES Resources: time and people Data cleaning Data formatting and Data alignment Choosing what data to mine Involve stakeholders early and often Articulate clearly how data is collected and how it will be used 31 CHALLENGES, CONT. Technologies: interoperability Ability to translate data into action Resources for interventions Philosophically Intrusive approach vs Privacy Right to Fail 32 ARE YOU READY? What questions are you trying to answer? Will data mining help you answer the questions? Do you have a culture of evidence-driven decision making? 33 NEXT STEPS President and Provost are supportive? Capacity to collect and disseminate information? ROI should be quantifiable and clear. 34 The more data we have about more people, the more we can improve services to individual students. We can begin to offer more customized, personalized choices to help them meet their educational goals. 35