Reprinted with permission of copyright holder.
Use of Adaptive Information
for Advisement in Learning Concepts and Rules Using
Computer-Assisted Instruction
ROBERT D. TENNYSON
University of Minnesota
Findings from two experiments showed
that students can effectively manage their learning needs in a
computer-assisted instructional system when provided continuous,
updated advisement information about their achievement
(diagnosis) and instructional needs (prescription) in relation to
the objective. In Experiment 1, high school students receiving
instruction (learning four physics concepts) via a
learner-adaptive-control management strategy that included
advisement performed better on the posttest than students ii: a
learner-control strategy (p > .001) and needed less
instructional time than students in a program-managed,
adaptive-control strategy (p > .001). Experiment 2 replicated
the effectiveness of the learner-adaptive-control strategy by
showing that students were able to make increasingly better self
assessments and management decisions during three separate
instructional units (learning nine punctuation rules) than either
a learner-partial-control strategy (p > .001) or a
learner-control strategy (p >.001).
Research on computer-assisted instruction (CAI) has consistently shown that when students control the learning program they often terminate instruction too early and fail to learn the defined objective(s) (Tennyson & Rothen, 1979). In learner-control management strategies, the implicit, if not explicit, management principle is that the student is an adequate judge for selecting the learning strategy that will result in effective learning (C. L. Tennyson, & Rothen, 1980). However, instructional research (DiVesta, 1975; Rothen & Tennyson, 1978) and applied projects (Steinberg, 1977) dealing with variables of learner control (using rather large or complex teaming tasks) have failed to demonstrate that students can make and carry out decisions of content element selection and personal assessment. In particular, the poorest decision makers were students who knew little about the learning task or who were performing poorly at it (Holland, 1977).
To account for such problems in CAI management systems, Tennyson and Rothen (1979) designed and tested (Tennyson & Rothen, 1977) the Minnesota Adaptive Instructional System (MAIS). Research on the MAIS has demonstrated the effectiveness of a program-control management system in selecting the appropriate amount and sequence of instruction for individual students. Although sophisticated adaptive systems may eliminate the problem of premature termination of study, they neglect the important educational goal of student responsibility in learning.
Glaser (1977) suggests that students in a learner-control mode who receive meaningful information about their own learning development may adopt reasonable learning strategies. Using this proposition, and in an attempt to help students make and carry Out appropriate content selection and learning assessment, Tennyson (1980) and Tennyson and Buttrey (1980) combined a learner controlled mode of CAI management with the diagnostic and prescriptive information generated from the MAIS. The results of their research showed that a learner-control condition can be a valuable CAI management strategy if students receive sufficient information about their learning development-information that continuously shows them what progress they are making toward mastery of the objective and provides meaningful advice on appropriate stimuli necessary to obtain it. From these findings, it is possible to postulate a cognitive-instructional theory such that students, if given assessment information according to their current state of learning, will be able to make instructional management decisions leading them toward mastery of the given objective. This idea is associated with both mastery learning theory (e.g., Anderson & Block, 1977; Bloom, 1974) and cognitive learning theory (e.g., Jenkins, 1974; McKeachie, 1976), such that the student would have control of his or her own learning program while receiving substantial individual advice on instructional needs necessary to reach mastery.
To continue development of a theoretical basis for student self-assessment in reference to learning specific objectives, two research objectives were proposed: (1) to replicate and extend the research of Tennyson (1980) and Tennyson and Buttrey (1980) by testing directly the advisement information procedure in reference to both program and learner-control CAI management strategies; and (2) to test the learner-control strategies (with and without advisement) over several, separate units of instruction (in contrast to a conventional one-session research paradigm). Operationally, the advisement information variable functions as follows: First, after an initial assessment, the student is advised of (a) his or her initial level of achievement compared with the desired learning criterion (diagnosis), and (b) the amount and sequence of instruction necessary to obtain the objective (prescription). Second, the student would be continuously advised, while on task, of his or her learning development (updated diagnosis) and the instructional needs (updated prescription) necessary for mastering the objective.
To test advisement as an instructional
variable in learner control of content selection and
self-assessment in relation to CAI management strategies, two
experiments were conducted: one for each of the above stated
objectives. Experiment 1 tested the advisement variable by
contrasting it directly with a conventional learner-control
management strategy and a program managed adaptive-control
strategy (which used the MAIS). Experiment 2 further extended the
test of the advisement theory by studying its effect on student
decision making over a series of instructional units (one hour
each), the purpose being to study the effect of various
learner-control management strategies after students had become
familiar with the process of decision making. In other words,
Experiment 1 tested the advisement variable in both a
learner-control management strategy and a program control
management strategy. Both of these strategies were tested against
a conventional learner-control system. In Experiment 2, using
only learner-control management Systems, the advisement variable
was contrasted with two forms of learner control. To be tested
was student adaptation to instructional decision making over an
extended period of time.
EXPERIMENT 1
In Experiment 1, using a
concept-learning task, three CAI management strategies were
tested: learner-adaptive-control, adaptive-control, and
learner-control. I hypothesized that the learner-adaptive-control
management strategy would not only be as effective in student
acquisition of the concept-learning task as the adaptive-control
strategy (that is, students in both of these management strategy
conditions would surpass the criterion of mastery on the
posttest), but also that it would be more efficient in terms of
student on-task learning time. I further hypothesized that both
of these management strategies (learner-adaptive-control and
adaptive-control) would be more effective than the
learner-control management strategy.
METHOD
Subjects
Participants N = 63) were 12th-grade students (female and male) enrolled in physics classes at a suburban high school in Minneapolis, Minnesota. From a random list of the three computer-assisted instructional treatment conditions, students were assigned to one treatment condition as they appeared for the experiment. They were informed that credit was given for participation and that their teachers would grade their posttest scores. This contingency was included to simulate an actual classroom-related incentive.
Learning Program
The coordinate concepts selected for this study, drawn from the field of physics, were force, power, velocity, and speed (Tennyson, 1980). Examples used in the learning program and accompanying tests were written according to the concept design strategy developed by Merrill and Tennyson (1977). Of 104 examples in the learning program, 56 (14 per concept) were used in the instructional lessons and 24 (6 per concept) each in both the pretest and posttest. The learning program used the same response format as the two tests, but provided feedback on whether or not a response was correct.
To validate the learning program, a
formative evaluation procedure was used (Tennyson, 1976;
Tennyson, 1978). First, several subject matter experts reviewed
the definitions and examples. Then, after appropriate revisions,
a one-to-one tryout of each treatment condition was conducted
with six randomly selected students from the sample population,
followed by simulation tryouts of each treatment condition (six
students per treatment). From this tryout data, final refinements
on the learning program and computer software were made.
MAIS
To study the adaptive control and
advisement-information variables, I used the Minnesota Adaptive
Instructional System (MAIS), a computerized Bayesian statistical
model developed by Tennyson and Rothen (1979). Detailed
descriptions of the MAIS adaptive model are presented in Rothen
and Tennyson (1978) and Tennyson and Rothen (1977).
Treatment Conditions
In the learner-control treatment
condition, the subjects decided (a) whether to continue receiving
examples or to go to the posttest, and (b) which concept they
wanted to study next. Subjects were informed in the program
directions that they had complete control over the amount and
sequence of instruction. In contrast to the learner-control
strategy, the adaptive-control treatment condition, using the
MAIS management system, continuously selected the number of
examples presented to each student based on the student's pretest
and on-task performance in relation to the learning objective,
then sequenced the concepts according to the student's response
pattern to each given example (Park & Tennyson, 1980).
Following the pretest, students in the adaptive-control condition
were given program directions and informed that they would
receive a posttest at the conclusion of the instruction. The
learner-adaptive-control treatment condition used a management
strategy that (a) allowed students to decide whether or not to
continue receiving instruction, and (b) advised them on the
number of examples needed to reach mastery for each concept
(diagnostic and prescriptive information provided by the MAIS).
Program directions informed the students of the learner-control
format and told them that the advisement information was
determined according to their individual learning development in
relation to mastering each concept and would aid them in deciding
the amount and the sequence of their instruction.
Procedure
As students reported for the experiment,
each was assigned to one of three treatment condition programs.
The experimenter switched on the computer (an Apple II
microcomputer) and entered each student's treatment program
number. After receiving directions on operating the computer,
students were administered a 24-item pretest. When the pretest
was finished, students received a print copy of the four concept
definitions from the experimenter, to which they were able to
refer during the learning program. After studying the
definitions, students raised their hands to indicate readiness to
study the examples in the learning program. The experimenter
entered the appropriate command on the computer for students to
begin. When each student finished the learning program, the
experimenter took the definition sheet and entered the
appropriate command on the computer for the posttest to begin.
All student entries were single-letter alphanumeric responses to
multiple-choice questions. The test and learning program required
no other entries by the student. After the students had finished,
the experimenter thanked them and they left the room.
RESULTS
The data analysis consisted of a
multivariate analysis of variance (MAN-OVA) with univariate tests
(ANOVAs) on each dependent variable, followed by mean comparison
tests (Student-Newman-Keuls, SNK). Multivariate dependent
variables consisted of the correct score on the posttest and time
on task. The tests for homogeneity of variance of within-class
and between-class linearity were nonsignificant (p >.05).
Means and standard deviations for the posttest correct scores and
time on task are presented in Table I. The MAN OVA on the three
management strategies was significant, U(2, 1,61) = .36,p
<.001.
Posttest Correct Scores
The ANOVA on the posttest correct mean
scores showed a difference between the three management
strategies, F (2,61) = 41.64, p <.001. On the
SNK mean comparison test, the mean correct score for the
learner-adaptive control condition was equivalent to the
adaptive-control condition (p> .05), both of which were over
eight points higher than the learner-control condition (p
<.001; see Table I). A review of the standard deviations shows
more variability for the learner-control subjects: of interest
because a chance score on this test is 8.4. In terms of the
percentage of items correct, the learner control condition was at
52 percent correct, whereas the other two conditions
Note. Maximum posttest score 24.
were above 80 percent correct. The
pretest correct mean score F test was nonsignificant (p>
.05).
Time
Average time spent on the pretest was 7.5 minutes; time spent on the posttest averaged 10.1 minutes, with no time differences among groups on either test (p > .05). Time-on-task mean difference among the three management strategies was significant, F (2,61) = 25.06, p < .001. In length of time on task, the SNK test showed that the subjects in the adaptive-control condition spent more time than the subjects in either the learner-adaptive-control condition (14% greater; p < .05) or the learner-control condition (46% greater; p < .001); the learner-adaptive-control condition subjects spent 30 percent more time on task (p < .01) than the learner-control condition subjects.
In summary, the results for Experiment I
showed that the advisement information included in the
learner-adaptive-control condition successfully kept the subjects
on task until they mastered the concepts, and they also did it in
less instructional time than subjects in the adaptive-control
condition. These results replicate the Tennyson (1980) and
Tennyson and Buttrey (1980) findings and test directly the three
forms of MAIS management strategy. The implication seems to be
that students can successfully participate in the management of
their learning when provided with their own individual diagnostic
and prescriptive information. The second objective in this study
was to investigate the effect of practice in learner-controlled
CAI management strategies. That is, once students become
experienced with a learner-controlled CAI learning system, would
the use of advisement continue to have the same effect as in a
one-session research paradigm, and would students with practice
in conventional CAI learner-control systems do as well as
students receiving advisement information?
EXPERIMENT 2
Findings from Experiment I showed that subjects, if provided information concerning their own learning development while on task, can make effective instructional decisions. Given these findings, in Experiment 2 I propose to make two extensions. First, instead of a concept-learning task, I selected a more complex learning behavior; rule learning, a task that requires students to solve problems using learned rules (Gange, 1980). Second, I tested the learner control variable over an extended period of learning, when the subjects had become familiar with a learner-controlled management strategy. The purpose here was to test the theoretical notion that students can accept responsibility for their own learning program and can make meaningful decisions on instructional needs if provided appropriate diagnostic and prescriptive information. Additionally, the adaptive-control condition was not used in this experiment as a possible contrasting control group because, although it was as effective as the learner-adaptive-control condition, it was also significantly less efficient, and as such, would not contribute to a test of the advisement theory.
Given the extended instructional time required in this experiment (three sessions at 1 hour each), I followed Park and Tennyson's (1980) guidelines for CAI design and dropped the pretest in favor of an initial assessment period included in the learning program. That is, the learner-adaptive-control management strategy program was modified so that an introductory section of instruction was under program control (MAIS) to obtain the initial advisement information, after which instructional control was given to the subjects. To experimentally control for this program control in the initial section of the learner-adaptive-control management strategy, a third management strategy, learner-partial-control, was designed such that the introductory instruction was under program control (same system as the learner-adaptive-control strategy), followed by direct subject control of the remaining instruction (but without the advisement information).
The hypothesis for Experiment 2 was as
follows: The learner-adaptive-control strategy would be more
effective for student learning than either the
learner-partial-control or the learner-control strategies (that
is, the subjects receiving advisement would master the
objectives). An assumption behind this hypothesis was that
students would make more effective use of advisement information
if given an opportunity to practice its application over an
extended period of time. A secondary hypothesis was that the
learner-partial control strategy, because of the introductory
instructional section, would be more effective than the
learner-control strategy.
METHOD
Subjects
Participants (N = 47) were 11th-grade male and female students enrolled
in English classes at a Minneapolis,
Minnesota, suburban high school. As subjects appeared for the
experiment, they were assigned to one treatment condition from a
random list of the three treatment conditions. Subjects
understood that their participation was recommended (their
achievement scores would count toward their semester English
grade), and that they were given the option of completing the
instruction using printed materials rather than the
computer-assisted instructional lessons.
Learning Program
Nine internal punctuation rules were selected for the learning program following a content analysis procedure that structures information elements according to both cognitive learning needs and relations of base level information with subordinate and prerequisite information (Nelson, Note 1). Structuring the internal punctuation rules in this manner provided four rational content sets, which were organized into three distinct units of instruction as follows: Unit 1, one comma rule, two semicolon rules, one colon rule, and a no-punctuation rule, all associated with independent clauses; Unit 2, two comma rules associated with dependent clauses and phrases; and Unit 3, one comma rule and one semicolon rule associated with items in a series, and one colon rule associated with final appositives. For purposes of acquainting the subjects with the computer and program/task directions, a presession training unit was also developed.
Of 189 examples in the learning program,
81 were used in the instructional lessons (Unit 1 = 56; Unit
2 = 28; Unit 3 = 42), 28 in the Unit 1 posttest, 14 in the Unit 2
posttest, and 21 in the Unit 3 posttest. The instructional units
were tested for readability using the text analysis methods from
the Minnesota Interactive Readability Approximation Program
(MNIRAP). The readability index was 10.7 equivalent reading grade
level. Each example, presented via computer terminal, was a
sentence requiring the student to enter the appropriate
punctuation (if needed) and identify which rule was used.
Following completion of each punctuated sentence, the subject
received: (a) knowledge of results (correct or incorrect) on each
punctuation mark (if any); (b) the correct answer(s); (c) the
correctly punctuated sentence; and (d) identification of the
appropriately used rule(s). In addition to the above, subjects in
the learner-adaptive-control condition received
diagnostic-prescriptive information about the number of examples
needed to reach mastery for each rule. The learning program was
validated and revised according to a formative evaluation
procedure for instructional materials (Tennyson, 1976, 1978).
Treatment Conditions
CAI treatment programs for the learner-adaptive-control and learner-partial-control conditions consisted of two sections: an introductory set of examples (4 examples per rule) presented by program control, followed by a second set of examples presented under learner control. Examples in both sections were presented using the adaptive sequence rule of the MAIS (Park & Tennyson, 1980). The third condition, learner-control, consisted of only one management strategy: complete subject control over the amount and sequence of instruction throughout the entire program.
Program directions for the learner-adaptive-control condition informed subjects that (a) an initial section of the instruction would, in addition to presenting examples of the rules to be learned, assess their learning progress; (b) following the initial section of program control, they would be advised of that assessment for the purpose of helping them select examples in the second (learner-control) section; and (c) the assessment and advisement would be continuous until they decided when to end the instruction and take the posttest. Prior to the beginning of the learner-control section, the advisement procedure was illustrated and discussed.
Program directions for the learner-partial-control condition simply informed subjects that the introductory section of program control would be followed by a learner-control section. Prior to the beginning of the second section, subjects were presented further directions on the learner-control format and the options they could exercise.
Program directions for the
learner-control condition informed subjects that from the
beginning they would select the amount and sequence of
instruction and, finally, decide when to begin the posttest.
Procedure
As subjects reported for the experiment,
they were assigned to a treatment program from a random list of
the three treatment conditions. The experimenter switched on the
microcomputer and entered each subject's treatment-program
number. After receiving about 10 minutes of on-line training on
how to operate the computer, how to punctuate sample sentences,
and what to consider in making learning decisions, subjects began
Unit 1 instruction. Once they began the instruction, subjects
were given a printed booklet containing best examples and the
punctuation rules for that given unit (Tennyson, Chao, &
Youngers, 1981). Although the program was not timed, subjects
were told they had to complete both instruction and the posttest
before the end of the 50-minute class period. Subjects were
permitted to study the rule definitions whenever they wanted
during the instruction. When each subject finished with the
learning program, the experimenter took the rule booklet and
entered on the computer the appropriate command for the posttest
to begin. After completing the posttest for Unit 1, subjects were
reminded of their next instructional period for which they were
scheduled. This general format was used for Units 2 and 3, except
that the training section required only review of the response
procedures.
RESULTS
The data analysis consisted of a MANOVA
and ANOVAs on each dependent variable followed by SNK mean
comparison tests. Dependent variables was the correct score on
each of the three unit posttests and task time on each
instructional unit. The tests for homogeneity of variance of
within-class and between-class linearity were nonsignificant (p>.05).
Means and standard deviations for the dependent variables of unit
posttest correct scores and time on task are presented in Table
II. The MANOVA showed a significant difference between the three
management strategies, U (2,5,40) =.62,p <.001.
Posttest Correct Scores
An ANOVA on the posttest mean correct
scores (Table II) for each unit of instruction was significant as
follows: (I) Unit 1: [F (2,44) = 9.47, p < .005]
subjects in the learner-adaptive-control condition (82% correct)
performed better (SNK, p < .001) than either the
learner-partial-control condition (71% correct) or the
learner-control condition (53% correct)-which were different
(SNK, p < .001); (2) Unit 2: [F (2,44) = 11.86, p
< .005] the learner-adaptive-control condition posttest
score (88% correct) was over three points higher than the
learner-partial-control condition (65% correct; SNK, p <
.01), which in turn was three points higher than the
learner-control condition (43% correct; SNK, p<.001);
and (3) Unit 3: [F (2,44) = 21.73, p < .001] the
learner-adaptive-control condition posttest score (90% correct)
was over five points higher than the learner-partial-control
condition (62% correct; SNK, p < .001), but with no
statistical difference between the
Note. Maximum posttest scores for
Unit 1 = 28, Unit 2 = 14, and Unit 3 = 2.1.
learner-partial-control condition and
the learner-control condition (49% correct; SNK, p > .05). In
terms of average percentage correct for the three units of
instruction, the learner-adaptive-control condition was 87
percent correct, whereas the other two conditions were below the
70 percent program objective at 66 percent correct for the
learner-partial-control condition and only 48 percent for the
learner-control condition.
Time
The time-on-task ANOVAs were significant
for each unit: Unit 1, F (2, 44) = 15.6, p <.001;
Unit 2, F(2,44) = ll.9, p <.005; and Unit 3, F(2,44) =
26.5, p < .001. Table II, which presents the time means
and standard deviations, clearly shows that subjects in the
learner-control condition ended their instruction relatively soon
in each instructional unit. These results replicate earlier
findings in learner-control research that shows subjects exiting
from the instruction rather quickly regardless of learning
progress. The data also show that once subjects moved from the
program control to conventional learner control
(learner-partial-control condition) they again left the
instruction quickly. However, when subjects were given advisement
concerning their progress toward mastery, they stayed on task
until the objective was learned. Overall, subjects in the
learner-adaptive-control condition stayed on task approximately
12 percent longer than the subjects in the learner-partial
control, while achieving 24 percent better posttest performance.
DISCUSSION
Research on learner-control variables for CAI has not produced instructional design variables of a generic nature. That is, learner control seems to be a useful management format once the correct contingency is identified, and this seems to be successful only in highly defined occupational areas. Too often the contingencies associated with school-related learning-such as grades, praise, or rewards-vary in relation to individual variables such as sex, race, age, and home environment rather than instructional format. This situation results, therefore, in variables and instructional design conditions often too confusing for practical application or theoretical development. My purpose in this study was to introduce a variable to the conventional CAI learner-control management strategy unlike that of previous research variables. This variable dealt with actual on-task learning development: advising students of both their learning progress (diagnosis) and their individual learning need (prescription) in reference to mastering the instructional objective. Students would thus have meaningful information on which to make learning assessments concerning the amount and sequence of instruction needed to learn a given content.
The variable of advisement, as operationally defined, was highly significant in providing subjects in the learner-adaptive-control conditions in both
experiments with meaningful information with which to make appropriate decisions about acquisition of the learning tasks. For example, in Experiment 2, on the three-unit posttests, subjects in the learner-adaptive-control condition did significantly better than the subjects in the learner-partial-control condition, averaging over 87 percent correct compared with 66 percent correct. In Experiment 1, subjects in the learner-adaptive-control condition did as well as the subjects in the adaptive-control condition. The importance of these two results is apparent when they are contrasted with the performance of subjects in the learner-control conditions, who in both experiments responded correctly to less than 50 percent of the posttest items. Experiment 2 further showed the effect of practice on each of the management-control strategies. My second objective was to study this effect over several separate instructional sessions. By the third instructional unit, subjects in the learner-partial-control condition were performing no better (below criterion) than the subjects in the conventional learner-control condition. Simply presenting more mandatory instruction does not seem to be an adequate design strategy for improving performance. As reported in an earlier study by Tennyson and Rothen (1977), boredom can be a major factor in deterioration of performance in program-controlled instruction. However, subjects who received the advisement information continued throughout the three instructional units to perform above the given criterion level. Possible boredom interference seemed to be minimized by the advisement information, which kept subjects actively involved in the management and assessment of their learning. These two experiments clearly indicate that when students in learner-controlled CAI programs receive diagnostic information and prescriptive advisement they can perform as well as students in a highly sophisticated, adaptive program-controlled strategy and that with experience they can further improve their learning performance over conventional or partial learner-controlled programs.
The dependent variable of on-task learning time is important to consider in the study of CAI management strategies because subjects in learner controlled conditions consistently learn instruction before mastery of the objective(s). In contrast to this basic finding, subjects in this study who received advisement while in a learner-controlled instructional program stayed on task long enough to obtain mastery. In Experiment 2 they were on task approximately 40 percent longer than the subjects in the conventional learner-control condition and 12 percent longer than the partial-learner-control subjects. It was my thesis that if subjects in a learner-control strategy were given advisement in the form of adaptive diagnostic/prescriptive information they would stay on task long enough to master the objectives. The assumption was that the information-processing strategy that students use in learning concepts and rules would further refine the adaptive information presented by the MAIS and provide the motivation to stay on task.
The findings of both experiments support this notion: subjects in the learner-control conditions without the adaptive information ended their learning prior to achievement of mastery, whereas subjects with advisement (learner-adaptive-control) apparently used the diagnostic/prescriptive information and stayed on task long enough to achieve mastery of the objectives and perform 47 percent better.
In conclusion, a learner-control
condition can be a valuable instructional management system,
especially for computer-assisted instruction, if students receive
sufficient information and advisement about their learning
development: information that continuously shows them what
progress they have made toward mastery of the objective and
provides meaningful advice on appropriate stimuli necessary to
obtain mastery. Further research should extend this advisement
variable by investigating the interactive effects with both
individual learner characteristics and content structure. For
example, students with low aptitude (or low prior achievement) or
low motivation (or low interest) may require more program
management support than students with high aptitude and
motivation. Also, student ability to make individual assessments
should be further studied to identify a means of transferring
such skill to learning situations in which advisement is not
provided.
REFERENCE NOTE
1. NELSON, J. A. Embedded structure
model: A content analysis. Unpublished master's thesis,
Department of Curricular and Instructional Systems. University of
Minnesota, 1980.
REFERENCES
ANDERSON, L. W., & BLOCK, H. H. Mastery learning. In D. J. Treffinger, J. K. Davis, & R. E. Ripple (Eds.), Handbook on teaching educational psychology. New York: Academic Press, 1977.
BLOOM, B. S. An introduction to mastery learning theory. In J. H. Block (Ed.), Schools, society, and mastery learning. New York: Holt, Rinehart, & Winston, 1974.
DIVESTA, F. J. Trait-treatment interaction, cognitive processes, and research on communication media. A V Communication Review, 1975, 23, 185-196.
GAGNE, R. M. Learnable aspects of problem solving. Educational Psychologist, 1980, 15, 84-92.
GLASER, R. Adaptive education: Individual diversity and learning. New York: Holt, Rinehart, & Winston, 1977.
HOLLAND, J. P. Variables in adaptive decisions in individualized instruction. Educational Psychologist, 1977, 12, 146-161.
JENKINS, J. J. Remember that old theory of memory? Well, forget it! American Psychologist, 1974, 29, 785-795.
MCKEACHIE, W. J. Psychology in American's bicentennial year. American Psychologist, 1976, 31, 819-833.
MERRILL, M. D., & TENNYSON, R. D. Teaching concepts: An instructional design guide. Englewood Cliffs, N.J.: Educational Technology, 1977.
PARK, 0., & TENNYSON, R. D. Adaptive design strategies for selecting number and presentation order of examples in coordinate concept acquisition. Journal of Educational Psychology, 1980, 72, 362-370.
ROTHEN, W., & TENNYSON, R. D. Application of Baye's theory in designing computer-based adaptive instructional strategies. Educational Psychologist, 1978, 12, 317-323.
STEINBERG, E. R. Review of student control in computer-assisted instruction. Journal of Computer- Based Instruction, 1977, 3, 84-90.
TENNYSON, C. L., TENNYSON, R. D., & ROTHEN, W. Content structure and instructional control strategies as design variables in concept acquisition. Journal of Educational Psychology, 1980, 72, 499-505.
TENNYSON, R. D. The role of evaluation in instructional development. Educational Technology, 1976, 16, 17-24.
TENNYSON, R. D. Evaluation technology in instructional development. Journal of Instructional Development, 1978, 2(1), 19-26.
TENNYSON, R. D. Instructional control strategies and content structure as design variables in concept acquisition using computer-based instruction. Journal of Educational Psychology, 1980, 72, 525-532.
TENNYSON, R. D., & BUTTREY, T. Advisement and management strategies as design variables in computer-based instruction. Educational Communications and Technology Journal, 1980, 28, 169-176.
TENNYSON, R. D., CHAO, J., & YOUNGERS, J. Concept learning effectiveness using prototype and skill development presentation forms. Journal of Educational Psychology, 1981, 73(3), 326-334.
TENNYSON, R. D., & ROTHEN, W. Pretask and on-task adaptive design strategies for selecting number of instances in concept acquisition. Journal of Educational Psychology; 77, 69, 586-592.
TENNYSON, R. D., & ROTHEN, W.
Management of computer-based instruction: Design of an adaptive
control strategy. Journal of Computer-Based Instruction,
1979, 5, 126-134.
AUTHOR
Tennyson, R. D. (1981). Use of adaptive information for
advisement in learning concepts and rules using computer-assisted
instruction. American Educational Research Journal, 18,
425-438.
ROBERT D. TENNYSON, Professor and Chairman, University of Minnesota, 178 Pillsbury Dr., S.E., Minneapolis, MN 55455. Specializations:
Instructional psychology, educational
computing, and educational research.
