Examining large student cohorts - a question of questions

With pressure on academic courses worldwide to increase student numbers, the trends in exam marks distribution
for subject modules become more meaningful, with more distinct pattern characteristics reflecting student choice, topic and exam
question difficulty and lecturer marking severity, refinement and consistency. The practice of representing the overall exam results
for a module through histograms enables Normality, skewness and randomness to be identified, interrogated and understood better.
However, when dealing with large numbers of exam candidates (of the order of 1000 or more), an investigation of the averages
and histograms for individual exam questions can further reveal refined explanations for unusual student performance. This paper
investigates the outcomes of 1st and 2nd year examinations for modules on an engineering degree course in another jurisdiction,
with class sizes of circa 2400 and 1700 students, respectively, in order to develop a deeper understanding of exam dynamics
amongst students and academics setting and marking those papers. It can involve many tens of thousands of items of data in a
histogram for just one module, in which trends are not random and have potential causes, intended or accidental. It emerges that
at least four different question mark patterns may exist from a range of modules types investigated. These indicate the importance
of examiners having a better appreciation, when delivering lectures, planning exam papers, question structure and marking scripts,
of the different factors which give rise to unusual examination trend outcomes.