PPIG 2012 - 24th Annual Workshop
Observing Mental Models in Novice Programmers
Richard Bornat, Saeed Dehnadi, David Barton
Abstract: A test which partitions subjects into those who appear to use an algorithmic model of program execution and those who do not has been automated. Experiments have been conducted in a UK school with a year cohort of students aged 13-14 and in a Mexican university with a cohort of novice computer scientists. In the school about a third of subjects appeared to use an algorithmic model, which we find surprisingly many; in the university there were around half in the same category, which is in line with university results in the UK. Operation of the online test and the analysis tool is described. Interviews with subjects in the school revealed some ways in which the algorithmic classification may be expanded. End-of-course results are not yet available for the test subjects, so statistical associations have not yet been explored.