The assumption that examinees either know the answer to a test item or else guess at random is usually totally implausible. A different assumption is outlined, under which formula scoring is found to be clearly superior to numberright scoring. Empirical studies to investigate the validity of the new assumption are called for. (13pp.)
