Currently, great efforts are being made to prevent cheating attempts during online exams. Microphones and cameras are used to monitor exam takers' behavior during the exam, and control software tracks keyboard input and open programs. The use of such methods and tools must remain within a narrow framework for data protection reasons; however, there are no uniform rules for educational institutions to date. To make matters worse, individual monitoring of examinees is nearly impossible in exams with hundreds of participants - bandwidth and space on the observers' screens are limited, after all. In addition, examinees are aware of all the technical tricks for deception. This includes methods such as playing back a recorded video via a virtual webcam, using a secondary device outside the camera's field of view, or communicating with other exam takers to share answers and results. To ensure that online exams are conducted successfully, the goal must therefore be to prevent exam cheating (such as copying and the unauthorized exchange of information) not by means of costly, time-consuming, and possibly unlawful monitoring, but by using intelligent procedures to make it impossible from the outset.
At this point, the software of examU UG can help. The program makes it possible to "program" exam tasks. This means that the tasks are created in the form of code or algorithms. From this input, a desired number of individualized and randomized tasks or exams are automatically generated. Within a task, numbers, images, diagrams, texts and also the queried solution methods can vary. The exams generated by examU are therefore inherently deception-proof, since the examinees are not able to exchange correct results with each other. In addition, the generated tasks are checked for plausibility via predefined boundary conditions to ensure that all generated tasks are mathematically valid and make sense in terms of content. At the same time, a sample solution is also generated fully automatically for each task.
The topic of "anonymized exams" to increase fairness and avoid unconscious bias is discussed again and again. Electronic exams, e.g. in Moodle, can be corrected automatically in most natural science subjects. Discrimination by the computer can be ruled out, since no artificial intelligence is used here either, which may take other factors into account in addition to the results due to incorrect training. In the case of manual evaluation of texts, the answers of the examination candidates can be exported together with their user ID, even without names. After the tasks have been corrected, the scores can then be transferred back from the IDs to the exams in reverse. The many advantages of electronic exams can be used not only in exams on your own device at home, but also in digital face-to-face exams.
The German state of North Rhine-Westphalia has published a guideline for digital exams (https://www.ldi.nrw.de/handreichung-zu-online-pruefungen-hochschulen), which clearly highlights the data protection hurdles involved in monitoring exam candidates. Many of the measures listed are no longer necessary through the use of examU.
The generated exams can be imported into various LMS systems. It is particularly easy to use moodle as an LMS exam platform.Details
The description language of examU is based on Python - therefore any Python code can be used. Support for Matlab is also integrated, as well as a number of functions to simplify data exchange with Matlab.
Tasks can be randomized using random numbers. This allows to create complex tasks with random parameters, so that each examinee gets an individual exam.
Easy to use
The examU editor provides a simple interface with many wizards and dialogs for code generation.
Exams created in the examU editor can be recycled. New exams can be created quickly in the following exam phases. If a switch is made from online to face-to-face, face-to-face exams can also be generated with XMU using LaTeX. In addition, the change of LMS systems is possible.
In addition to standard task types, we provide support for LMS plugin extensions such as KPrime.
If an exam is programmed in multiple languages, e.g. in German and English, it can be output either in German, in English or bilingually. Alternatively, the language set in moodle can be used.