Analysts at the University of Adelaide and the Commonwealth Scientific and Industrial Research Organisation developed a set of classifiers to predict the performance level of a primary school student based on both individual and environmental variables.
Over the past two decades, Australia has witnessed a considerable decline in student achievement, as well as widening performance gaps across student populations. By detecting students who are more likely to face obstacles to scholastic success, teachers and administrators can intervene sooner rather than later. If an educator knows a certain student may be more prone to setbacks, it positions the educator to offer individualized assistance and prevent at-risk youth from falling through the cracks in the system. A combination of tailored teaching strategies and policy can be deployed accordingly to foster an environment where all students have the support they need to succeed.
When creating the predictive mechanisms, the researchers used data from the National Assessment Program – Literacy and Numeracy (NAPLAN), an annual exam administered to all Australian students in grades 3, 5, 7, and 9. Within this dataset, 3% of the sample performed in the “below standard” achievement range. For grades 5 and above, a student’s prior academic record was factored into the classification algorithm, but classification of students in grade 3 relied solely on family- and school-level predictors. To determine classifier performance, researchers examined the area under the receiver operating characteristic curve (AUC). The best performing classifiers for grades 5 and above attained an AUC of about 95%; those for grade 3 topped out with an AUC of about 80%.
With these formidable results, it is no stretch of the imagination to contend that “it is feasible for schools to screen a large number of students for their risk of obtaining below standard achievement a full two years before they are identified as achieving below standard on their next NAPLAN test.” Furthermore, as governments ramp up their data collection efforts, the predictive techniques employed by these researchers could be repurposed in other important policy spheres.
Find out more here.
I currently study Computational Linguistics as an M.S. candidate at Brandeis University. I received my Bachelor's degree from the University of Pittsburgh where I explored linguistics, computer science, and nonfiction writing. I'm interested in the crossroads where language and technology meet.