In the subjects mathematics, computer science, science and technology, up to 40 percent of the students stop already in the introductory phase. A research team from the Method Center at the Faculty of Economics and Social Sciences at the University of Tübingen has now developed a statistical procedure that allows you to predict an average of eight weeks in advance whether they will stop their studies, the University of Tübingen states in a press release.
The team also made an overall methodological advance with the newly developed algorithm. During the prediction, the algorithm is able to take into account the differences between the individual students that already exist at the start of the study, such as the general cognitive performance. It can separate these from the well-being of individual students over time. This makes it possible to predict the probability of student drop-out, also for students who are in principle suitable.
Such a separation of different levels of influence is also interesting for numerous issues in other areas. The team published an article on the development of the method in the journal Psychometrics.
At the start of studies, there are various factors that affect the chance of dropping out for students in science / technical subjects. “For example, it is clear that high school math performance and the individual student’s overall cognitive performance are different. Lower performance initially leads more often to dropouts in the entry phase,” says Professor Augustin Kelava of the Center for Methods. how to identify those who are dropping out fast among similarly skilled beginners. “
For the study, 122 students at the University of Tübingen were surveyed in their first semester of mathematics in a large preliminary study of their prior knowledge of mathematics, their interests, their school careers and their financial background, and collected personality variables, including emotional stability.
“The results of the entrance exam gave us a picture of the stable characteristics of each individual student,” says Kelava. This was followed by five minutes of surveys three times a week, 50 times in total over 131 days of the semester, where students reported how they were feeling at the time and whether they felt like they were going to class.
“We also knew who was admitted at the end of the semester so we could check the predictions and we knew the grade of the final exam. In addition, the acceptance of our study was very high, ”he says. The research team did not intervene specifically in individual study histories, “which, however, could be a future individually targeted application,” the scientist said.
The predictions were calculated using the newly developed statistical method, an algorithm that uses data collected in real time; a so-called Forward Filtering Backward Sampling (FFBS) algorithm. “The levels of influence are complex. They overlap, a host of variables play a role in the decision to continue or drop out.”
As a result, the research team was able to predict the intention to drop out an average of eight weeks in advance, ie at a time when students are still coming to class. “Often students are away after Christmas after the start of winter,” says Kelava. “We have succeeded in separating the two levels of influence of the students’ stable characteristics on the one hand and the changes in their mood over time on the other hand by predicting ulterior motives. From their own information, how they feel and how they handle their studies, we can determine when they develop a latent tendency to drop out, which is not yet immediately noticeable at that time. ”
In practice, the statistical method offers an instrument for targeting individual students who are in principle qualified for the subject, but who show a tendency to drop out, for example with a guidance offer. In general, the method is also suitable for certain research questions in other areas, such as separating stable impact variables from situational development in stock prices in economics or engineering applications.