Learner support options in computer-assisted learning

Comftut. E&r. Vol. 8, No. 4. pp. 349-354, Printed in Great Brttain

0360-131584 ~3.0~+0.~ Pergamon Press Ltd

1984

LEARNER SUPPORT COMPUTER-ASSISTED

OPTIONS IN LEARNING

G. M. BOYD, L. DOUGLAS and C. LEBEL Educational

Technology

Graduate Programme, Concordia Montrtal, Qukbec, Canada

University,

145.5 De Maisonneuve

West,

tj3G lM8

INTRODUCTION Attention is usually concentrated on instructional objectives and the presentation and sequencing, (or the dynamic simulation) of content, in the design of computer-assisted learning (CAL) activities. However, other factors are equally important: (1) the “sociostructure” (who works with whom and under what social rules) is crucially important for educational CAL where social cooperativeness and responsibility goals are always as important as the specific individual skills acquired; (2) the computerized “learner-support-system” (review, hints, calculate, printout, map, learning-strategy advice, etc.) is extremely important in all kinds of computer-assisted education and training. Lack of attention paid to the educational impact of the sociostructure leads to mis-educational phenomena such as that of the “closet computer queen”-the intelligent pupil whose social exchange skills are negligible who is sent down the hall to work on the Apple or PET in the broom closet and who develops ever more technical competence at the expense of social competence. Learner competition, pitting individual against individual is generally detrimental [ l] yet much of the current CAL studyware which does in fact involve more than one learner at a time, does so via direct competition. Studyware which requires pairwise or group coo~erat~un on-line needs to be developed and researched. Some of the work of one of the authors, Claude Lebel, in this direction will be described here. An obvious form of learner support is simply to provide enough time on the computer for the learning activity to make a real difference to the learner’s thinking and problem-solving. This is particularly important when micro-world languages such as LOGO, Micro-PROLOG and SMALLTALK are being used. Mistaken aspirations toward fairness often lead to the provision of only a few minutes a week of on-line time to each student. Such “gruel-&ble” CAL is merely a cosmetic, not a real educational enhancement. The Lincoln School project of Papert et af.[Z], indicates that about 40 h of on-line work is needed for most children to learn the “core programming” problem-solving skills which they believed to be essential. In both projects described here learners had what initially seemed like ample time on the machines to develop the target skills, but in both cases some learners clearly would have benefited from much more time, had it been available. Strategic, tactical and recapitulation related forms of learner support have been included in various CAL projects[3,4] but much research remains to be done to ascertain the most cost-beneficial forms of provision of these kinds of learner support. The work of one of the authors, Lionel Douglas, has been carried out with the intention of making a contribution in this area. The general question of how much, and what kind of advice, and from whom, should the on-line learner have is what we are addressing here. If you believe that the basic educational goal is the cultivation of globally responsible potent autonomy, then the concept of “learner-control” in CAL is inherently very appealing. A number of experiments have been done which on the surface seem to deprecate the use of learnercontrolf5,6], but when examined in detail they are seen to show that yearner-control without adequate learner-support is what is ineffectual. Learner-support is needed in the form of advance briefing and goal orientation, and advanced organizers. It is needed in the form of in-progress aids: entailment graphs, glossaries, calculators, memory-support and human buddy support. Finally it is needed in the form of diagnostics with explanations and advice as to what remedial investigation and exercises are needed when specific objectives prove difficult to reach. CAF 8,3 c

349

350

G. M. Born et al.

There is the question then that, given limited resources, which forms of learner-support are most important to make learner-control CAL successful? It is to be expected that the various forms of learner-support interact but multi-factor experiments to study all these interactions were beyond our resources; however experiments were done looking at some pairs of salient learner-support factors and their interaction.

MEMORY

SUPPORT

AND

ADVISORY

SUPPORT

Lionel Douglas’ experiment was with university students who one would expect to be fairly well organized autonomous learners and who therefore should need a minimum of learner-support to make use of learner-control CAL[7]. Sixty-one volunteers desirous of learning the use of “string-handling commands in BASIC” were the subjects and they were quasi-randomly assigned to four groups: (MS) a group with access to Memory Support in the (AS) a group with access to Advisory Support of the (MAS) a group with access to both of the above; (NMAS) a group with access to neither of the above but features of the lesson, including objectives rationale and The objectives of the CAL lesson were: It was expected that at the end of the lesson, subjects

forms listed in Table strategic and tactical

1; forms.

still with access to many learner-control some advice. would

have been able to:

(1) correctly employ the use of string functions to solve problem situations presented to them; (2) given pre-written BASIC statements, correctly evaluate the solution to these statments; (3) identify any irregularity or invalidity that may be present in statements in BASIC which utilize the functions covered in this lesson. On the basis of a pilot study with eight students, materials were revised and an adequate time frame for completing the task was deduced to a 2 h. Unfortunately in the main experiment only 54% of the students actually felt that the 2 h was enough time for them to complete the lessson. A pretest, post-test and delayed post-test design was used with twelve multiple choice and eight problem-solving questions on each form of the test. (A Spearman Brown reliability of 0.92 and KR21 reliability of 0.86 and a Wilks Lambda discrimination index of 0.003 were obtained.) The only statistically significant results of the study was that performance test scores were signiJicantly higher,for the groups with Memory Support (MS) and (MAS) than for the other two groups. The rate of learning (indicated by the time taken on-line before the successful students chose to take the post-test) also proved to be much greater for those using Memory Support options than for those not using them. Unfortunately the Advisory Support option either was not adequately designed or the students were unwilling to use it, because no significant differences in level or rate of performance were measured between the (AS) and (NMAS) groups. We feel that it is clearly demonstrated that for complex technical subject matter memory support, in the form of look-back and structured review, should dejinitely be provided for good learning performance. This to some extent simply confirms early findings that printer-type terminals have an advantage for most learners because of the easy access to earlier work. The advisory support question needs further work both to improve the design of advisory support tools and to determine the relative value of advice given by a peer or tutor versus advice provided by the machine.

SOCIOSTRUCTURE

SUPPORT

IN

LEARNER-CONTROL

CAL

An experiment was conducted, with English-speaking Grade X Mathematics students (1517-year-olds), by Claude Lebel to determine if pairs of pupils working together under a low-performance contingency situation[8,9] might not do better than solo students or students working together without the contingency condition on their marks. The underlying questions are: (I) How many learners per machine are cost-effective? (2) Is peer support helpful? and (3) Is peer

Learner

support

options

in computer-assisted

learning

351

support more forthcoming and helpful when the mark entered for both students in a pair will be the lowest post-test mark of either of them? The instructional objective was for the student to be able to qualitatively predict the change in shape of the graph of a quadratic function y = a(x - b)* + c when any or all of the three parameters (a,b,c) are changed, as in Fig. 1. The experiment used a two-way repeated measures design with 47 subjects randomly assigned to the three experimental groups. Each group interacted with the same CAL module (a model of the general quadratic function) for 1 h; however, this interaction took place on an individual basis for the first experimental group, in pairs for the second group, and also in pairs but under a low-performance contingency for the third group. Immediately following this activity, a post-test was administered to each of the subjects in order to measure their understanding of, as well as their ability to generalize, the concepts which were to be learned as a result of the exercise. A parallel version of the same test was administered two weeks later in order to detect possible differences in retention by the three experimental groups. The immediate and delayed post-tests were counterbalanced in order to increase their reliability by eliminating differences which might have occurred had the two forms been used respectively as post-test and delayed post-test. This had the added advantage of making it more difficult for students to pass answers to their peers. Rejection of the null hypotheses was based on the 0.05 level of statistical significance. The studyware written in Applesoft BASIC for the Apple II computer was essentially an interrogative graph-plotting program for the quadratic function with student usage record-keeping facilities. Three functions can be plotted at once on a single screen display. THE

LESSON

In order for the problem to make sense it had to be put into context. The first step consisted of an assignment requiring students to draw the graphs of three quadratic functions. These graphs served as the basis for a class discussion in which the general quadratic function was introduced. During this discussion students recognized that each function was represented by a parabola, that this was probably due to the fact that in each case y was expressed as a second degree expression

0 START

Fig. 1. Effects of parameter

variation.

Fig. 2. Basic program

structure.

G. M. BOYD et al.

352

in x, and that any differences were likely accounted for by the various constants which appeared in each function. It was at this point that the concept of a general function was introduced. Students were shown how each of these functions could be considered to be a special case of the general quadratic function. They easily identified the value of each parameter in all three functions. This introduction was followed by ,an exercise which required students to state the value of each parameter in various functions and to write the equation of a function given the value of its parameters. Upon completion and correction of this exercise students were ready to be introduced to the problem. The problem was presented to students in the form of a questionnaire. The purpose of the questionnaire was to reduce the original problem to a series of smaller problems in accordance with Polya’s[lO] model for problem-solving; the formative evaluation had revealed that most students could not do without such guidance. The problem required that the eflect of each of the three parameters on the graph of the function be discovered. Under normal circumstances a classroom discussion would have followed. However, because of the nature of this study, the problem-solving exercise was succeeded by a test designed to measure the extent to which students had been able to solve the problem and whether or not they could apply this knowledge to non-quadratic functions.

DEFINE FUNCTIONS -------__~------

Y * A(X -

In2+c

FUNCTION

1

Y = -3(X

FUNCTION

2

Y = 2(X

Fig. 3. Definition

s’;bltL

- n2

Fig. 4. Effects of varying

- 2

2

fi, 1‘tl DOTS 'C' TO CONT Y NUE.

DOTS

- 6

to be graphed.

2

2 PRESS

of functions

- ,I2

L';;& DOTS

“a” alone.

Learner

support

options

in computer-assisted

learning

353

2

Y=1X2 IlEDIUtlDO S ;aiL-a!TS PRESS 'C' TO CONTINUE. i Fig. 5. Effects of varying

!%:+#TS “c” alone.

INSTRUMENTATION

The test consisted of fourteen multiple choice questions divided into two parts. The first seven questions tried to determine how well students understood the individual and combined effects of the three parameters on the graph of the quadratic function, while the purpose of the second set of questions was to find out if students could transfer their knowledge to non-quadratic functions. The test was developed concurrently with the program and was modified twice before arriving at the final version. A second form of the test was also prepared in order to reduce the effect of learning (as a result of taking the first test) on the delayed post-test. A final analysis was performed on both tests using the experimental data. The analysis included the calculation of indices of difficulty and discriminability for each item on both forms of the test as well as the Kuder-Richardson index of reliability for each of these[ 111. Pupils were given 1 h to solve the problem working with the computer program and then they were tested for 30 min. No statistically

.

Fig. 6. Effects of varying

“b” and “c”

together.

354

G. M. Bov~ et al.

significant differences (using r-tests) were found between the means of the three treatment groups on either the immediate or the delayed post-test. From this it may safely be concluded that two pupils working together, on one computer on a problem-solving exercise with a time limit, do just as well as single pupils. It was noted that even though students were assigned to each other in pairs on a random basis

they all became sufficiently involved in the problem to put aside any personal differences. No conflicts of any sort were observed in the paired sessions. Part of the reason for the null result on the experiment was the relationship between task difficulty and time available. The success rate for the solution of the problem used in this experiment was low, with roughly 50% of students receiving a mark of less than 30% on the immediate post-test. This lack of success can be attributed to poor problem-solving strategies. The suggestions included in the program were clearly insufficient, and the one recommending the graphing of two or three functions per experiment (as opposed to a single function) may even have been harmful. It could be that students require formal instruction in the use of “conservative focusing”. How successful such instruction would be is worthy of investigation Even though the hypotheses for this study were rejected, it may yet be found that computerassisted problem-solving and other types of non-instructional CAL are better served by a “cooperative goal structure” than solo study arrangements. Meanwhile, the use of pairs is recommended since it makes limited resources available to a greater number of students, while reducing the cost per student by 50%. REFERENCES 1. Ames C., Competitive versus cooperative reward structures: The influence of individual and group performance factors on achievement attributions and-affect. Am. educ. Res. J. 18, 273-287 (1981). 2. Pauert S. A.. Watt D., DiSessa A. and Weir S.. Final renort of the Brookline LOGO Proiect Part II: Proiect summarv and data analyses. A. I. Memo No. 545. MIT Press, Cambridge, MA (1979). 3. Burton R. R. and Brown J. S., An investigation of computer coaching for informal learning activities. Ini. J. Man-Mach. Stud. 11, 5-24 (1979). 4. Gentner D. R., Toward an intelligent computer tutor. In Procedures for Instructional Systems Development (Edited by O’Neil Jr H. F.). Academic Press, New York (1979). 5. Seidel R. J., Wagner H., Rosenblatt R. D., Hillelsohn M. J. and Stelzer J., Learner control of instructional sequencing within an adaptive tutorial CA1 environment. Instr. Sci. 1, 37-80 (1978). 6. Steinberg E. R., Review of student control in computer-assisted instruction. J. Comput. Eased Instr. 3, 8490 (1977). 7. Douglas L. L., An examination of the effect of memory support and advisory support in a learner control computerassisted instruction program. Master of Arts in Educational Technology Thesis. Concordia University, Montreal (1982). 8. Lebel C., Cooperation between adolescents in computer-assisted algebraic problem solving. Master of Arts in Educational Technology Thesis, Concordia University, Montreal (1982). 9. Johnson D. W. and Johnson R. T., Learning Togeiher and Alone: Cooperation, Compelition, and Individualization. Prentice-Hall, Englewood Cliffs, NJ (1975). IO. Polya G., How to Solve It. Princeton University Press, Princeton NJ (1973). Il. Tuckman B. W., Conducting Educational Research. Harcourt, Brace & Jovanovich, New York (1972). 12. See also: Boyd G. M., Four ways of providing computer-assisted learning and their probable impacts. Compur. Educ. 6, 305-310 (1982); Boyd G. M., Education and miseducation by computer. World Yearbook of Education 1982/83 Computers and Education (Edited by Megarry J., Walker D., Nisbett S. and Hoyle E.), Chap. 4. Kogan Page, London (1983).