Randomised online assessment

Discipline Area

Mathematics and Statistics (but technique can be used in any discipline).

Key assessment issue addressed

Assessment of large classes, plagiarism, time saving, reliability.

The initiative

It is usually the case that class sizes exceed the number of computers in a laboratory and this has been one of the problems which needed to be overcome before computer based examination could become reality. HyperTester overcomes the problem by allowing a "Random Test" feature. These are tests where for each question, every student gets a random but equivalent alternative to the question. So at 'sit down' time, random questions are down-loaded and two students, whether sitting next to each other or taking the test on different days, are very likely to have different questions. The Mathematics department of UWA has courses with over 600 students enrolled and a Macintosh Lab of 128 machines and conducts tests over multiple sessions and usually over a number of days.

Computers are far more reliable at mundane tasks than people. Any human error in marking is removed with computer based tests. The man-hours to do the marking is also eliminated as the computer can mark all the papers in seconds and at the same time produce statistics of the students results. Even if tests are not run on the computer, the students results are usually entered into some database (eg an excel spreadsheet). Again this process, which is subject to error and takes time to enter and check, is eliminated with a computer based test.

Initially computer based tests can require more time to set up than paper equivalents. However if 'Random' tests are created, questions can be reused in subsequent years and so take less time than conventional tests to set up. Paper copies of the test don't need to be duplicated for each student, saving time and paper.

Finally, HyperTester's big advantage is in the way it presents its statistics. When it marks a test, it produces the usual mean and standard deviation and a detailed, though easy to understand, report of each question. From the report one can ascertain quickly which questions the students had problems with, which questions they found difficult, which distracters were not working and perhaps which distracters were working too well. Details for each student are also readily accessible. All these being produced automatically every time HyperTester does some marking.

The reason for the initiative's effectiveness

The time saved in the preparation, marking, recording of marks and production of statistics as well as the reliability of the accuracy of the marks stored in a final database.

Further details

Mr Con Savas
Department of Mathematics and Statistics
The University of Western Australia
con@maths.uwa.edu.au