A sample of 208 students from grades ten through twelve were randomly assigned to one of four groups in a study of the effects of multiple-choice, fill-in-the-blank, and constructed modes of response to embedded questions on the learner's ability to answer application-type questions. The three experimental groups received five instructional passages on psychological principles, including three adjunct questions on each passage; treatments differed in the mode of response: multiple choice, fill in the blank, or constructed. The control group received only the criterion measure. Half the members of each group were randomly assigned to one of two delayed criterion measures, lower-order questions or higher-order questions. Results indicated that the response modes produced selectively different learning outcomes. If the instructional objective is the acquisition of factual information, the multiple-choice mode will initiate effective processing activities. If the objectives are concerned with the acquisition of higher-order application skills, than the constructed response mode contributes to the necessary deeper-level processing. (Author/AA)