Abstract: Data mining is typically associated with business and marketing. For example, Amazon uses people's past purchases to suggest books they might be interested in buying. Similarly, academic analytics can be used to identify and predict students who might be at risk, by analyzing demographic and performance data of former students. However, there is no clear consensus on how to intervene with current students in a way they will accept and not associate with academic "profiling." Why should students think they are exceptions to our rules? This panel presentation will share how three institutions are approaching this problem and provide an overview of related issues.
Using Analytics to Intervene with Underperforming College Students
1. USING ANALYTICS TO INTERVENE WITH UNDERPERFORMING COLLEGE STUDENTS Kimberly Arnold (Purdue University) John Fritz (University of Maryland, Baltimore County) Eric Kunnen (Grand Rapids Community College) January 20, 2010
37. GRCC - STARFISH EXAMPLE INSTRUCTOR MANUALLY RAISES FLAG The instructor can select one or more students from the student list and manually raise a flag on the student. When raising a flag, the instructor writes a description of the flagged issue. This information is forwarded to someone who can help the student, as determined by the flag rule setup by the administrator.
38. GRCC - STARFISH EXAMPLE INSTRUCTOR RESPONDS TO A “FLAG SURVEY” EMAIL Administrators can email survey requests to instructors. Clicking on the request takes the instructor to a flag survey where they are prompted to flag their students if they are experiencing any specified problems.
39. GRCC - STARFISH EXAMPLE AUTOMATIC FLAGS BASED ON BLACKBOARD GRADEBOOK/COURSE ACCESS Administrators can set up flags to be raised that are auto-generated. Flags can be raised by the system by grades and average scores and specific gradebook columns in Blackboard. Flags can also be raised based on students’ access to their courses in Blackboard. Additional customization is available through API’s.
SIS data= admissions data such as HS GPA, HS rank, highest science taken in HS, SAT CMS data= Time spent in content files, time in discussion boards, time spent doing practice quizzes etc. and other technologies such as level of engagement from Hotseat, online tutor session, blogs, wikis etc. Other data= help seeking behavior (office hours, help centers), work study status, on campus/off campus ------ By combining all the data and building various predicative models, we can customize the algorithms. Institution level (PU students are different from UMBC, UMBC students are different that GRCC etc.) College level (ENGR students differ from LA students so a. data points are different b. different data elements are more predictive for certain college/programs) Course level (each course has different characteristic—some for majors, some survey level, inst differences with TA etc. different level of “success” as defined by instructor
Customizable—each instructor creates their own intervention based on the multiple data points Real-time this can be updated instantly Specific—faculty AND students know exactly where they should be focusing on (ie you are not utilizing the practice quizzes as much as your peers and you are not going to online tutoring as recommended, please try—we want you to be successful) actionalble—tell the students EXACTLT what to do
Academic analytics requires patience! We have been doing research for 7 years. When you are talking about wrangling all these data sources—some dynamic, some static—it takes time! And we have to be mindful of privacy issues as well.
1) The Big Idea = Evaluating and measuring the use of Blackboard on your campus is becoming more and more important. With budget cuts and challenges… along with the need to continue to transform education by leveraging technology and the tools therein. 2) Focus of Presentation = Will be to review Project ASTRO and it’s capabilities. 3) Themes = Tracking/Reporting/Discovery/Sharing/Acting How many faculty and students are using Bb? What tools are used in our system? How can we improve and promote new tools? Who do we need to communicate with for the next upgrade? Who are our innovative faculty? How does faculty use compare to student use? Which departments are using Bb and to what degree? How can we use these data to encourage deeper use and create awareness of Bb? What is the impact of Bb in teaching and learning? Goal - To build a community based advanced reporting tool and Blackboard Building Block that will empower institutions to become more accountable and able to use data-driven decision making to enhance, optimize, and advance Blackboard Academic Suite usage in teaching and learning. Awarded in 2007 Code Name = Project ASTRO
Identify = Detect It costs more to recruit students that it does to retain them. LEVERAGE CMS data