Summary. Candidates taking multiple choice question (MCQ) papers in which marks are deducted for wrong answers often omit many items for fear of losing marks. In an MCQ paper in 1985 candidates who omitted many items made relatively few errors; they were not necessarily much less able than those who omitted few items.
Those who had omitted many items were interviewed and advised to answer more, including those about which they were uncertain but not totally ignorant; the same advice was given to the whole group before they took another MCQ paper in the same subject in 1987.
In 1987 there were large changes in the numbers of items omitted and, in particular, those who had omitted many items in 1985 answered many more in 1987. They also made more errors; despite this they tended to improve their performance in relation to their colleagues. A clear relationship was shown, for the whole group, between an increase in the number of items attempted and a rise in the rank order.
The effects on candidates' behaviour of deducting marks for wrong answers and of permitting them to omit items are reviewed and the nature of the advice which should be given to candidates taking such examinations is discussed.