One of the main strategic objectives of Arab Open University (AOU) is to double the current number of students in the next 5 years. In order to achieve this objective, a top priority is improving student retention and progress. This includes controlling dropout rates.
AOU operates in 8 countries. It uses a wide range of cloud based applications extensively to manage the student experience. So, using predictive analytics to identify the main factors that influence student progress, retention and dropout rates was a natural decision. Predictive data analytics is helping improve the quality of education and enabling decision-makers to tackle critical indicators. These indicators include student-specific, institutional factors or both, all of which can influence student progression, retention or dropout rates.
AOU conducted a study at their headquarters in Kuwait of the students in the Information Technology and Computing (ITC) program. There are 8,744 students in the program and they were all part of the study. AOU used IBM Watson Analytics throughout the study for data refinement, investigation and visualization. AOU explored trends, identified drivers, examined decision trees and identified key factors.
Student dropout rates
The first part of the study tracked the trend of student progress and dropout rates of ITC students from fall semester 2014 until summer semester 2016. Semester-wise, the dropout rate is inversely proportional to the student level. The study proved that the dropout rates have declined from one semester to another for all student levels. Enhanced quality assurance standards and academic advising are creating higher levels of student commitment to the program:
Still, lower student levels show higher dropout rates, but the dropout rate reached acceptable range by the end of the period under consideration.
Student progress rate
Another important aspect is student progress rate. Semester-wise, the progression rate is directly proportional to the student level. The behavior is almost constant from one semester to another as you can see here.
Therefore, the progression rate of lower level students need more investigation to know how to enhance the progression rates for such students.
Key academic performance indicators
AOU also investigated key academic performance indicators (KAPIs) to find out if or how they affect student retention, progress and dropout rates. The percentage of withdrawn students is a critical KAPI for any academic institution, and all universities want this KAPI to be as low as possible. The trend, evident here, is that the percentage of withdrawn students is decreasing. Again, thi sis the result of high quality assurance standards and proper academic advising procedures.
Maturity affects student retention
One of the most interesting relationships is that between the pass rate and percentage of withdrawn students. As you can see, Watson Analytics shows a general negative correlation the two. And, the percentage of withdrawn students is higher for low-level “core modules” but improves with higher level modules. This highlights an interesting trend. As students mature in the AOU system, fewer withdraw.
But what is driving student withdrawal?
Watson Analytics is not simply a smart data discovery service. It offers predictive analytics features. So, AOU asked it to identify the main drivers for student withdrawal. The results were interesting:
The main factors influencing student withdrawal percentages and, by extension, student retention were “Module Results Standard Deviation,” “Country,” “Module Level” and “Average Students per Class.” The standard deviation factor reflectsthe variety and diversity in student background and how prepared they are for the module. Although the AOU has unified admission criteria, the diversity in their high school educations can affect module results. Drilling down to the next factor, country, shows that conditions in specific countries, such as manpower, tutoring and student support, also affect withdrawal. Poor physical infrastructure and resources and facilities, along with political, cultural and certain socioeconomic conditions can take their toll.
The predictive analytics results from Watson Analytics also support the discovery about maturity. Students are more likely to withdraw in the lower levels of the modules because they have not yet adjusted to the rigor of study. And, it makes sense that the average number of students in a class can affect withdrawal. Students in a smaller class receive more attention and support than those in larger, more crowded classes.
Student Risk Factor
One of the most important KAPIs for identifying struggling students is the student risk factor. AOU defines this as a function of current academic status, historical behavior and progress rate. These students are at high risk of withdrawing from their study plans, which adversely affects student retention and progress. By identifying students at risk, IBM Watson Analytics can act as an early alert system. AOU administrators can use insights from Watson Analytics to determine how to support these students so they don’t withdraw. This visualization shows the risk factors at various student levels, emphasizing how high they are for lower level students.
Curiouser and curiouser
So, why the high risk numbers for lower level students? AOU needed answers fast to prevent a rise in dropout numbers and lower rates of products. A predictive model from Watson Analytics had some interesting insights.
Student level is a major factor in determining students at risk of dropping out because the open education system at AOU requires an adjustment in studying and learning. Then, there’s high school GPA. It’s another major warning sign. Students with average or poor high school performance find college studies more challenging. Country indicates that campuses in certain countries do not offer as much student support as others. The age factor shows that the more mature a student, the likelier he or she is to stay in college. Interesting, gender plays a role. Female students overall exhibit more seriousness and commitment to tutorial attendance and student work, especially in the Gulf countries. The previous dropout factor reflects the seriousness on the part of students to complete their studies, as does the degree-seeking status.
Clearly, there’s work to be done, but things are looking up
By applying predictive analytics, AOU now sees some trouble areas and trends to address. It’s possible that an orientation session that prepares first-year ITC students for an open education environment could help. And, adding more tutors and support in the problematic countries could reduce student withdrawal.
In addition, there’s good news. Overall, student retention and progress rates are improving. AOU made some changes after investigating student performance numbers with Watson Analytics last year. This appears to be paying off.
Get more of the back story
For more information about how Watson Analytics is helping AOU understand its students, read the case study. And, if your academic organization is seeking a way to use analytics to improve student retention, take a look at the Watson Analytics Academic Program.
About our guest blogger
Ashraf S. Hussein is a Professor of Scientific Computing, Ain Shams University, Cairo, Egypt. A senior ACM member, he is currently working as Vice President for Education and Information Technology and Dean of the Faculty of Computing and Information Technology, Arab Open University, Headquarters, Kuwait (on leave from Ain Shams University, Cairo, Egypt).