Unintentional clues to the correct response in objective test items can best be reduced by

The clues are given by the experimenters, so what can be done is not letting the experimenters know what stimuli the subjects are receiving. For example, the experimenters can not know whether they are giving the patients medicine or placebo - they can use codes that will be deciphered later.

This is called a "double-blind" experiment.