Learner control of instructional sequencing within an adaptive tutorial CAI environment

The study described in this report was designed to test effects of learner control at the level of instructional sequencing within a self-contained tutorial course, administered by an adaptive computer program. Applicability to other levels is speculative and clearly requires further research.The experimentation has as independent variables four features of our CAI system that afford the student a specified degree of control over the sequencing of instructional material. Three of the variables are student options that control remedial activity and acceleration. The fourth variable allows control over sequencing of topics at specified points in the course. The purpose is to assess the relative contributions and interactions of these variables with respect to instructional effectiveness and efficiency.Following an entry test period, students were administered tutorial CAI instruction, a COBOL course (an average of 30 hours long), with four possible types of learner-control variables. These students were assigned at random to one of 24 factorial treatment conditions. Sessions were approximately three hours long per day with breaks left up to the individual. Following the instructional period, students were administered an “exit questionnaire” covering their opinions about course administration, content, and instructional environment.During the conduct of the experiment, three types of measures were taken on each student: (a) entry characteristics, including information processing (Guilford's Structure of Intellect), affective, and biographical data; (b) learner strategies, including type and frequency of control usage and the circumstances of their use; and (c) achievement and other performance-related measures including quiz scores, transit times, programming errors, opinions of topics, and Level of Aspiration (LOA) prior to the quiz of each topic. Assessments were made of the relative contributions and interactions among the learner-control and entry characteristic variables with respect to instructional effectiveness and efficiency as represented by the dependent measures.The result and implications can be described as follows. First, the study developed a well-tested instructional vehicle that meets the criterion of student mastery, a prerequisite for valid research in an instructional environment. Secondly, the study was performed in a rich instructional environment, preferred for generalizing results of Aptitude by Treatment Interaction (ATI) studies to a real instructional world. The third significant aspect of the current study has been the development of a very useful means to characterize high and low performers with an operationally defined set of criteria that has highlighted the value of discriminant function analyses in instructional research settings. Of great importance is the finding that the particular individuals designated high or low performers differed depending upon the particular instructional tasks. Yet, the phenomenon of high and low performance was consistent across two divisions of the course. High and low performers differed with respect to the usage of options, as well as their Level of Aspiration settings concerning their performance. Research is needed to identify more specifically the taxonomic characteristics of instructional tasks related to student profiles of high and low performers.Another significant finding in our study was that self-assessment can make a significant contribution to instructional management, whether the latter be by students or by the learning system. The next step that should be taken is to use the instructional options based on expectations, as part of the decision-making process in an adaptive instructional environment. A proposed prescriptive use of LOA as an Expectancy Operator is described. Lastly, research requirements for systematic study of levels and types of self-managed learning paradigms are discussed.