A lecturer has highlighted flaws in the electronic plagiarism detection software used by some 80 per cent of UK universities after testing his own work for copied material.
Simeon Yates, director of Sheffield Hallam University's Culture, Communication and Computing Research Institute, found that the web-based software, Turnitin, failed to recognise that the work he submitted for analysis was 100 per cent plagiarised from his own previously published output.
Turnitin's developer iParadigm claim that the program has access to "a vast database" of 4.5 billion website addresses and a number of subscription sites.
The program provided Dr Yates with an "originality report" that wrongly highlighted 28 per cent of one piece of work, suggesting that those segments may have been plagiarised.
"When we generated a report from Turnitin from previously published materials, the [only] sentences it picked up were fairly standard and not plagiarism at all," Dr Yates said.
Turnitin suggested that another paper had 18 per cent of suspect material, highlighting sentences such as "The remainder of this paper is structured as follows" - E but it failed to spot that the entire essay was lifted from a published paper.
One article submitted to Turnitin had been published both as a chapter in a book and in a peer-reviewed journal.
Dr Yates said: "The fact that the [Turnitin] website is there and students have to check it means that they have to think about it.
"They are forced by the software to check their citations, but it does not stop students from buying essays. We shouldn't just assume that it is a simple solution, and it is not going to work 100 per cent every time."
Gill Rowell, a plagiarism advisor at the Joint Information Systems Committee, said: "Turnitin does not have an extensive database of full-text peer-reviewed journals. We are aware of gaps in the coverage, and it is something we are trying to develop, but it's a long process of trying to persuade publishers to put their material into the database."
Ms Rowell added: "The website concentrates on showing matching texts, and it is very effective at highlighting collusion and cut-and-paste plagiarism."