Ensure Online Test Administration Reliability With These 3 Tips

Students like taking exams online. Whether driven by convenience or comfort, an overwhelming number of students will choose online options over in-person testing when given a choice. At Western Governor University, for example, 90 percent of its 80,000 students choose to take all tests online despite offering 2,000 test centers for in-person.

This increased preference for and prevalence of online exams can threaten academic integrity unless rigorous oversight, such as proctoring, is established to protect online test administration reliability. Many institutions have dealt with this by using human proctors to monitor exams, a solution that leaves much to be desired.

• Cost. Hiring human proctors to monitor students in real time is expensive. With higher education already being a significant financial burden to many students, extra fees may discourage students from completing exams online and even compromise their ability to participate in education.
• ‍Scalability. Human proctoring lacks the scalability needed to accommodate the growing demand for remote testing. In addition to being prohibitively expensive, it can be extraordinarily difficult to find qualified proctors, particularly when proctoring needs change rapidly.
• ‍Reliability. Practical and economic barriers make it difficult to maintain a manageable proctor-to-student ratio to ensure high-quality proctoring.
  

An AI-driven proctoring system can offer a cost-efficient, scalable and reliable solution without the need for continuous human oversight. An advanced system is trained to detect potential violations based on movements, sounds, and data input. All of these are vital to detecting potential violations in an online testing environment, and they can be easily missed by human proctors, particularly those with an abundance of students to monitor. Just as importantly, proctoring systems with advanced AI means you no longer have to bear the cost or negotiate the logistical challenges that come with relying exclusively on human proctors. This allows you to better accommodate student needs without compromising their educational experience or adding to their financial burden.

Online test administration reliability begins with identity verification. The Higher Education Opportunity Act of 2008 (HEOA) requires institutions receiving Title IV funding to verify the identities of students by using a secure login and password, proctored examinations, or other technologies and practices. While any of those should ideally allow you to verify identity, in practice, choosing the right approach to verification is essential.

A human proctor may effectively identify a student based on a picture I.D. and a live webcam stream. But happens when someone else steps in after the verification process and conducts the test in place of the student identified? When a proctor is dividing their attention amongst multiple students, a switch can easily be missed. What happens if an automated system relies solely on flawed facial recognition and fails to recognize a student? A test-taker could be stripped of their right to education based on poorly trained AI.

Choosing a proctoring platform that identifies a student at the beginning of the exam and continuously throughout can be a robust measure to prevent academic dishonesty. However, AI alone should not be relied upon to do this. Combining human oversight with purpose-built AI offers a more reliable way of verifying identity.

In an online exam, the value of human oversight goes far beyond identity verification. Environmental factors play a large role in online test administration reliability, and when students are taking exams at home, those environments can vary greatly. Lighting, sounds, and even room décor can all play a role in how a proctoring solution perceives suspicious behaviors or concerning environments. For example;

• Dim lighting can prevent AI technology from identifying a student.
• Benign sounds can be flagged as a test violation.
• Room decor can be marked as a distraction.
  

Even the most robust AI may fail to create a level playing field when faced with such diverse environments.

A human-in-the-loop system that combines AI with human wisdom and leaves final decision-making to human proctors can help you avoid this. Should AI flag a student erroneously, a human proctor can review the event and decide if a violation truly occurred. They can determine if a student was speaking to someone off camera or simply reading a question aloud to themselves, if the test-taker had to pause to attend to their child or if they were looking up answers off screen. The best proctoring solutions allow you to modulate sensitivity to minimize AI inaccuracies but even then, there is no substitution for human insight.

Student expectations are changing. The Covid-19 pandemic has shown them that classrooms aren’t the only place to learn or take exams, and many don’t want to return to the way things were. When online tests can be administered reliably, securely, and cost-effectively, they may not have to.

Rosalyn’s human-in-the-loop system is built for scalability and offers features that go far beyond standard proctoring solutions.

• System lockdown and ID verification remove the temptation for common test violations.
• Identity is verified at the start of the exam and continuously throughout.
• All events flagged by AI are sent to human proctors for review.
• Sophisticated queuing enables Rosalyn to efficiently distribute flagged events to proctors, which eliminates bottlenecks and supports unlimited scale.
  

Our student-focused solution is designed to deter academic dishonesty while providing user-friendly exam experiences that promote fair, equitable assessment.

Online assessments can present real challenges to students, educators, and institutions. An online proctoring solution developed to address them allows for reliable online test administration and helps remove barriers to education.