Proctoring or Proctored Test is defined as a mechanism to ensure the authenticity of the test taker and prevent him/her from cheating via a proctor is present during the duration of the test.
In Online Exam Proctoring or Online Video Test Proctoring, a candidate is monitored online during the test duration typically with the help of a webcam, mic and access to the screen of the candidate.
To take an online proctored test from a remote location candidates need:
1. a suitable device (Desktop PC/Laptop/Tablet/Mobile)
2. an Internet connection with at least 512kbps speed
3. a functional webcam and a mic
4. In PC/Laptop, latest version of Google Chrome, Mozilla Firefox, IE: Version 9 and above and Safari Browser.
Mettl provides 3 types of proctoring
1. Auto Proctoring:
Automated proctoring is the most advanced form of proctoring. Here again, the audio-video and screen share feeds of the test candidates are recorded during the test.But in addition to recording, the system also monitors the feeds for any suspicious activity using advanced video and audio analytics.It ensures candidate focuses on test screen during the test; there is enough light in the room and checks for suspicious objects in video and background voice activity to red flag the test. It also uses face recognition to do student authentication.
2. Live Online Proctoring:
The candidate can be viewed or monitored in the real time.
3. Record and Review Proctoring:
In this method, no proctor is monitoring the feed in real time. Instead, the audio-video and screen share feeds of the test candidates are recorded during the test.
A representative plays back these recordings in a fast-forwarded way (3X to 20X speed) and red-flags any suspicious activity through annotations.
Mettl’s AI based remote proctoring technology generates a variety of flags during an exam session to catch any suspicious behavior on behalf of the candidate. The choices of flags available are as follows,
Candidate not present
Additional person present
Suspicious candidate activities
Mobile phone detected