University of Twente Student Theses


Using data forensics to detect cheating in randomized computer based multiple-choice testing

Noord, S. van (2018) Using data forensics to detect cheating in randomized computer based multiple-choice testing.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Embargo date:23 May 2023
Abstract:Cheating is an existential problem in the testing industry; especially in high-stakes testing cheating examinees endanger the value of credentials. Data forensics, data analysis methods to identify aberrant behaviour patterns of cheating, have been around for decades. However, most of these methods focus on fixed paper-based exams. The Data Forensics Tool software, developed by eX:plain, is able to analyse randomized computer-based multiple-choice tests. Using and adjusting the Guttman model (1944), eX:plain has developed six indices that detect behavioural patterns. The quality, true detection rates and reliability of measurement of such indices and software has rarely been investigated in practice; researchers mainly use simulated data. The design of the current study is unique in its field: having known and instructed cheaters along with a control group of honest examinees take an existing test, with two years of information for benchmarking. Aside from being able to determine the true quality of the Tool, the software was finetuned to detect 37.5% of all cheaters, with 96.8% reliability within the detected sample. Based on current adjustments and results, further improvements of the software were suggested, including suggestions for automation of the analysis procedure and adapted behaviour measurements of response times and answer selection.
Item Type:Essay (Master)
eX:plain, Amersfoort, Netherlands
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:81 education, teaching
Programme:Educational Science and Technology MSc (60023)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page