Daniel provides strategic direction and content for AI’s electronic publication Higher Ed Impact, including market research and interviews with leading subject matter experts on critical issues. Since the publication’s launch in 2009, Daniel has written or edited more than 500 articles on strategic issues ranging from student recruitment and retention to development and capital planning. If you have a question or a comment about this article, feel free to contact Daniel at firstname.lastname@example.org.
Early alert programs have been emerging on college campuses for the last 10 years to varying degrees of success. Too often after the initial startup, many early alert programs fail to fully meet their designed purpose of identifying and reaching out to academically at-risk students -- in part because these programs are often focused on reactive rather than proactive identification and outreach, relying heavily on faculty to provide a "single stream" approach to flagging at-risk students.
This week, we reached out to two of the architects of Arkansas State University's forward-thinking approach to early alert: Jill Simons, executive director of Arkansas State's University College, and Darla Fletcher, director of technology services and support. Fletcher and Simons provide three key pieces of advice for making an early alert program truly effective:
- Use predictive modeling to proactively identify and reach out to your most at-risk students.
- Empower students to self-identify and self-report when they may be at risk.
- Reach out to your most successful, not just your least successful students, celebrating milestones in their progress toward their degree.
Let's take a closer look.
Use Predictive Modeling
Mining your data is key. Simons speaks to the importance of moving to a "pre-emptive" approach to early alert, identifying students who may be at risk even before the term begins and having a plan for outreach from the start. Real-time, operational data and feedback from faculty early in the term remains crucial, but if you can identify which admitted students may be in most need of help, you can provide support services in a more targeted way from the outset.
However, this may entail more than checking standardized test scores or noting which students are first-generation or out-of-state or are working adults. Even past academic performance, taken by itself, is not always a reliable indicator of success or risk for an incoming student. Look at other indicators as well:
- Mine your admissions data to find out which courses students took in high school (or at your feeder community college) -- how many credits in English, for example, and what grades did they earn in those courses?
- Have students complete an assessment of non-cognitive skills in order to note their resilience to stress and their academic engagement or the level of diligence they bring to their studies
Look to your historical data over the past few years. Reviewing demographics, academic performance, and the other data available, can you identify patterns of shared characteristics among both students who failed to persist and students who were most successful? This can help inform a more targeted approach to identification and early outreach to at-risk students.
Darla Fletcher suggests automated sending of messages before the term begins to those students who have been flagged -- based on predictive data -- as potentially at risk academically. In this way, a staff member charged with early alert outreach can devote their limited time in a more targeted way to scheduling advising sessions and problem-solving with students.
Fletcher suggests that the automated first round of communication include:
- Information on the support offices available on campus. Citing her dissertation research over the past couple of years, Fletcher notes that many students are simply unaware that the campus has a career counseling office or mental health services.
- Encouragement to speak with faculty. "Do everything you can to promote open lines of communication between students and faculty," Fletcher suggests.
Read our recent monthly edition, "Success Leaves Clues: Predictive Modeling in Higher Education," for a look at what you can achieve even with some early and simple data mining.
Empower Students to Self-Identify
By adding predictive modeling to the mix, you now have two information streams for notifying your office when a student may be at risk. One stream is reactive (faculty identifying students who haven't attended class regularly or have performed poorly on an initial assignment), and one stream is proactive (mining your data to find out which students are likeliest to be at risk, even before the term begins).
Here's a third stream. Empower your own students to self-identify when they may be at risk academically.
To achieve this, Arkansas State University is looking into implementing a home-grown degree audit calculator that allows students to self-audit their progress toward academic goals. Inputs into the calculator include how long the student has been enrolled, whether they have selected a major, how many credits they have accumulated, etc. Populating the calculator with this information, the student can find out whether they are on track toward their degree.
"We want students to be empowered to gauge their own progress," Fletcher explains. "We would like to build student confidence and make sure they know whether they are on track and what choices may need to be made. This is a chance for them to check themselves at various intervals during their college career."
If a student's self-audit identifies them as "at risk," the tool offers the student the opportunity to self-report (for example, by clicking a "I would like some help" button). This will send a message to the early alert office, and, based on the student's demographic data, it will also forward the message to the appropriate support service on campus. For example, if the student is an international student, the international student services office would receive a message identifying the student and noting what they are behind in, based on the data collected by the calculator.
This approach empowers students to investigate and take ownership of their own progress.
Reach Out to the Successful Students
Just as you want to invest in helping students realize when they may be at risk so that they can persist and seek assistance, you will also want to invest in helping students realize when they are achieving milestones toward their degree, so that they can build confidence and momentum.
This is critical because we know from the past decade of research on student retention that GPA is not the key determining factor in persistence. Aside from financial barriers, a frequent reason that students fail to persist is lack of momentum toward achieving a degree or even lack of perceived momentum.
"Reach out to those students who are doing well," Simons advises. "Let them know they are on track. Students may not realize they are making significant steps toward graduation." Focus especially on milestones that correspond to common barriers to persistence. For example, when students pass a course that has a high DFW rate, send them an email celebrating the fact. Let them know that only 55% of students pass that course, and recognize that they just did. Send them the message that this is a significant milestone toward graduation and one that they can be proud of.
Workshop Your Early Alert System with the Experts
This December, a diverse panel of experts on early alert intervention will convene in Phoenix, AZ for three days to help participating institutions outline an early alert program or improve an existing one. You will leave with an action plan to identify at-risk students on your campus—based on retention research, your campus profile, and your own data. A "work and learn" event, Designing Early Alert Systems for At-Risk Students is the only conference designed exclusively for teams tasked with early alert intervention.