Vous êtes sur la page 1sur 6

SPECIALSECTION

LEARNING TO PROGRAM AND LEARNING


TO THINK: WHAT’S THE CONNECTION?
Focusing on thinking skills that are cognitive components of programming-
rather than on intellectual ability-can illuminate the relationship between
learning a programming language and learning more about thinking processes.

RICHARD E. MAYER, JENNIFER L. DYCK, and WILLIAM VILBERG

Many strong claims have been made concerning the solving in domains beyond the programming lan-
relationship between learning to program and learn- guage that is taught. Linn [13] has suggested three
ing to think. In the process of learning to program possible “cognitive accomplishments” from learning
a computer, it is assumed, students will also learn programming in a language such as Basic: (1) learn-
about their own thinking processes. This premise ing the features of the language, such as the state-
underlies many assertions concerning the usefulness ments LET, PRINT, and INPUT; (2) learning to solve
of teaching computer programming in schools. For programming problems, such as designing programs
example, Papert [21] claims that, when children are in Basic; and (3) learning problem-solving skills ap-
allowed to write Logo programs, “powerful intellec- plicable to other formal systems, such as problem
tual skills are developed in the process.” Similarly, solving in other languages. The third category,
Bork [4] sees “computer programming as a vehicle which represents “transfer” of learning to new
for... training. . analytic thinking applicable to domains, is the focus of this article.
broad classes of problems.” Nickerson [20] argues As a brief historical prelude, it must be pointed
that we should view “computer programming out that the search for methods to teach problem
as a vehicle for teaching thinking skills.” solving has been an elusive one [14, 181. For exam-
Despite these claims, there have been very few ple, the Latin School movement, which originated in
relevant research studies and almost no convincing the 1600s in the United States, was one of the first
support of this connection [7, 8, 13, 17, 221. This large-scale attempts to teach “proper habits of
article presents research on three assertions con- mind.” The curriculum focused on teaching students
cerning the relationship between learning to pro- to read, write, and speak Latin, as well as teaching
gram and learning to think, based on a cognitive some Greek and geometry [23], the aim being to
analysis of programming [17]. Each assertion is de- build logical and disciplined minds. However, the
fined, available literature is reviewed, and an empir- practical demands of an emerging industrialized so-
ical study from our laboratory in Santa Barbara, Cal- ciety and the negative results of educational re-
ifornia, is summarized to assess the current state of search studies eventually helped to bring on the de-
knowledge concerning the relationship between mise of Latin Schools. Thorndike’s classic “transfer
learning to program and learning to think. of training” studies also found that learning Latin
In order to accomplish this goal, it is necessary to did not produce strong transfer to other domains
define what is meant by “learning to program” and [26]. Similar failures to produce transfer have been
“learning to think.” In the context of this article, observed for modern curricula aimed at teaching
learning to program refers to the initial learning of general thinking skills [5,15] and for compensatory
a novice’s first programming language. In particular, training in general intellectual development [6].
we focus on changes in people who initially know Transfer is even rare when students who have
nothing about programming and who engage in ap- learned problem-solving strategies within one do-
proximately lo-50 hours of experience with Basic. main are asked to solve formally identical problems
Learning to think refers to improvements in problem presented within a different domain [5, 181. It is
from this historical context-of strong claims for
This project was supported by the National Science Foundation under grant transfer coupled with little or no research support-
MDR84-70248.
that we now address each of the three assertions
0 1986 ACM OOOl-0782/86/0700-0605 7% discussed below.

]uly 1986 Volume 29 Number 7 Communications of the ACM


Special Section

ASSERTION 1: LEARNING A PROGRAMMING took the same thinking skills pretests at the begin-
LANGUAGE WILL ENHANCE A PERSON’S ning of the term, and were retested toward the end.
THINKING SKILLS Both groups consisted of nonengineering students
Does learning a programming language enhance who had no plans to become professional programmers.
thinking skills in domains beyond programming? The left column of Table I lists eight thinking
Preliminary studies involving Logo have offered skills tests: word problem translation required translat-
mixed results. Although Papert [Zl] offers case stud- ing word problems into equations, word problem solu-
ies and testimonials, the unreliability of such reports tion involved giving the correct numerical answer
is notorious. In a research study, Pea and Kurland for word problems, following procedures involved pre-
[ZZ~ failed to find support for the idea that a year of dicting the output for a procedure stated in English,
Logo activities improved children’s strategic plan- following directions involved predicting the conse-
ning skills. Similarly, Gregg [ll] found that four- and quences of following one or more directions, logical
five-year-old children had great difficulty both in reasoning involved solving a series of oddity prob-
learning to program a turtle and in transferring what lems, visual ability involved a series of paper folding
they had learned. Gorman and Bourne [lo] found, tasks, verbal ability involved decoding of verbal mes-
however, that third graders who learned Logo with sages, and arithmetic computation involved a series of
one extra hour of computer time per week per- addition and division problems. Test-retest reliabil-
formed better on tests of logical reasoning than third ity correlations were computed for each test, yield-
graders who learned Logo with just one half hour of ing a significant correlation at p < 0.001 for each
extra computer time per week. Apparently, gains in test. Sample items from each test are shown in the
thinking skills depend on the student being given sidebar.
heavy doses of Logo rather than just minimal expo- Some of the tests were designed to evaluate two
sure. The most encouraging study found that first skills that are specifically related to learning Basic:
graders who learned Logo during a ‘12-week course the problem translation skill-as measured by the
performed better on tests of creative problem solving word problem translation test and tHe word problem
than first graders who were exposed to computer- solution test; and the procedure comprehension
assisted instruction (CA]) over the same period [7]. skill-as measured by the following procedures test
This study involved very few students, however, so and the following directions test. These two skills,
replications are needed. which can be called specific thinking skills or spe-
Preliminary studies involving Basic have also cific cognitive components of Basic programming,
yielded mixed results. Bayman and Mayer [3] and were identified by carrying out a cognitive task anal-
Bayman [Z] have found that students who learn ysis of Basic programming (171. Other tests were de-
Basic in traditional hands-on, mastery courses often signed to evaluate general intellectual abilities such
harbor serious misconceptions of Basic statements. as logical reasoning, spatial ability, and verbal abil-
Furthermore, students who are able to use Basic cre- ity. Finally, the arithmetic computation test was in-
atively in problem solving tend to have fewer mis- cluded to evaluate a thinking skill-making rapid
conceptions than students who are unable to solve
problems [2]. Linn [IS] attempted to investigate the TABLE I. Eight Thinking Skills: Net Proportion Change after
idea that learning Basic would enhance students’ Learning Basic and Predictive Correlation with Basic Exam Score
problem-solving skills, but methodological problems
such as students’ difficulty in learning Basic pre-
cluded the study. One promising piece of evidence is
that students who know and use Basic are better Problem translation skill
able to comprehend word problems, such as “There Word problem translation +0.08 0.55
are six times as many students as professors at this Word problem solution +0.07” 0.56’
Procedure comprehension skill
university” [25]. These results suggest that there Following procedures +0.18 0.44’
may be a connection between programming and Following directions +0.04 0.44’
problem solving in other domains. General ability
In order to more closely examine the effects of Logical reasoning -0.01 0.29’
novices’ initial learning of Basic on their thinking Visual ability -0.05’ 0.31 l

Verbal ability -0.01 0.16


skills, we conducted a study using 57 computer- Other skills
naive college students who took a course in Basic. Arithmetic computation -0.01 0.26
Before taking the course, all students took a battery The asterisk (‘) indicates that gain score for Basic group is significantly
of thinking skills pretests; at the end of the term, all different from gain score for comparison group based on a t-test
students took versions of the same tests. A compari- (p < 0.05). or that correlation between pretest score and Basic exam
score is significant (p < 0.05).
son group, consisting of 54 computer-naive students,

606 Communications of the ACM ]uly 1986 Volume 29 Number 7


Special Secfior7

Examples of Items from Eight Cognitive Tests


PROBLEM TRANSLATION 4. Add the number in Box A and the number in Box C.
and put the result in Box A.
Word Problem Translation Test (6 problems)
5. Write down the numbers from Bojt A, B, and C.
A car rental service charges 20 dollars a day and
15 cents a mile to rent a car. Find the expression for What is the output of this program?
total cost C, in dollars, of renting a car for D days to
travel M miles. a. 5. 4. 9
a. C = 20D + 0.15M b. 14, 4. 9
b. C = 15D + 0.20M c. 14, 9, 9
c. C = 2019 + 15M d. 9. 4. 9
d. C = 0.15D + 20M e. None of the above
e. None of the above
GENERAL ABILITIES
Word Problem Solution Test (9 problems)
One day Mrs. Arnold worked 3% hours in the morning, Logical Reasoning Test (10 problems)
took a % hour for lunch. and worked 4% hours in the Draw an X through the set of letters that is different.
afternoon. If she began work at 8:30 A.M., at what time
BCDE FGHI JKLM PRST VWXY
did she finish?
a. 4:30 Spatial Ability (10 problems)
b. 5:oo Draw an X through the correct answer.
c. 5:30
d. 6:00
e. 6:30

PROCEDURE
Following
COMPREHENSION
Directions Test (8 problems)
ELI
L-
Column
Verbal Ability (9 problems)
1 2 3 4 5
Row 1 A B C D E black sheep = dag kip
Row 2 B D E A c white dog = tin bud
Row 3 c E II A B black cow = dag stam
Row 4 B A C E D white sheep =
Row 5 A C E B D
a. dag kip
Start in the lower left-hand corner. and follow the letters b. tin kip
up Column 1, down Column 2. up Column 3, and so on, c. stam dag
until you reach the upper right-hand corner. What is the d. bud tin
first letter to appear four times? e. tin bud

A B C D E
OTHER
Following Procedures Test (8 problems)
Computation (60 problems)
1. Put5inBoxA.
2. Put 4 in Box 8. 36
3. Add the number in Box A and the number in Box B. 20
and put the result in Box C. + 54

computations-that is not closely rela?ed to Basic thinking skills : problem translation (as measured by
programming. word problem translation and word problem solu-
Students in the Basic and comparison groups were tion) and procedure comprehension (as measured by
matched for pretest score, so that mean scores on following procedures and following directions). The
each pretest were identical for the two groups. Gain results of the word translation test are consistent
scores were computed by subtracting the proportion with the results of Soloway and his colleagues [25]
correct on the pretest from the proportion correct on in that the learning of Basic programming seems to
the second test. The first column of numbers in be related to improved skill in representation of
Table 1 shows the net gain score for each thinking word problems. In contrast, the Basic group did not
skill test for the Basic group, determined by sub- show significantly greater statistical gains than the
tracting the gain score for the comparison group comparison group on tests of genera1 intellectual
from the gain score for the Basic group. As indicated, ability, including logical reasoning, spatial ability,
the Basic group gained significantly more than the and verbal ability. As expected, learning Basic did
comparison group on the two specific component not tend to increase students’ computational speeds.

/u/y 7986 Volume 29 Number 7 Communications of the ACM 607


Special Section

These results encourage the idea that learning a general verbal intellectual ability (r = 0.17). In addi-
programming language-even a language with as tion, tests measuring nonverbal logical reasoning
many critics as Basic has-can result in changes in and mathematics problem solving correlated with
thinking skills. The improvements appear to be lim- learning Basic (Y = 0.54). Similarly, Webb [27] found
ited to thinking skills that are specifically tied to that the best predictor of success in learning Logo
specific concepts underlying Basic, however, and was a mathematics test consisting of word problems
there is no evidence of any enhancement of intellec- and computation problems (r = 0.81). In addition,
tual ability in general. The conclusion that can be tests of nonverbal logical reasoning correlated
drawn concerning this assertion is a modest one: strongly with learning Logo (r = 0.49); spatial ability
Under appropriate conditions, learning to program correlated with learning Logo in Webb’s study, but
may result in increases for specifically related think- not with learning Basic in Snow’s report.
ing skills, but there is not strong support for the idea The picture that emerges from this work is that
that it will radically improve general thinking skills. success in learning a language such as Basic may
depend on such specific skills as ability to translate a
ASSERTION 2: CERTAIN THINKING SKILLS WILL word problem into an equation or answer (problem
ENHANCE THE LEARNING OF PROGRAMMING translation), and ability to follow directions listed as
The question of what students need to know in or- a procedure (procedure comprehension). In fact,
der to learn a programming language has motivated tests based on these specific thinking skills, or cogni-
many studies, which find that general measures of tive components, provide the basis for construct va-
nonverbal intellectual ability such as in the IBM lidity and may provide even better predictive valid-
Programmer Aptitude Test (PAT) or Aptitude As- ity than traditional measures of general skills, such
sessment Battery Programming (AABP) can correlate as logical reasoning, and spatial and verbal ability.
with programming test scores in the range of r = 0.3 In order to test these hypotheses concerning pre-
[l: 8, 9, 12, 191. However, as Webb [27] points out, dictive thinking skills for Basic, we conducted a se-
“it is unclear which specific abilities included in ries of studies that each generated similar results,
these tests relate most strongly to performance.” but will focus on the study previously described, in
It is not particularly surprising or useful to find which we administered a battery of tests to 57 col-
that measures of general intelligence are related to lege students before and after a course in Basic.
students’ learning of programming, as general intelli- The second column of numbers in Table I (p. 606)
gence tests are designed to predict success in aca- summarizes the correlations between pretest score
demic learning under a wide variety of situations. and Basic exam score for each pretest. Tests of
The fact that such tests tend to predict success in the two specific thinking skills and two of the three
initial programming ability simply points to the tests’ general abilities tests tended to predict success in
predictive validity: Success on the test is related to learning Basic. A subsequent stepwise multiple
success in learning to program. A theoretically more regression analysis revealed that tests measuring the
important form of validity is construct validity-that two specific thinking skills were better predictors of
is, determining the underlying cognitive mechanisms success than tests measuring general ability: word
to explain why performance on a test is related to problem translation, word problem solution, and fol-
success in learning to program. The search for con- lowing directions were selected for the regression
struct validity requires a search for tests that mea- equation. Performance on these three tests ac-
sure theoretically meaningful thinking skills, such as counted for approximately 50 percent of the vari-
specific component processes required for program- ance in Basic exam scores. It is significant that the
ming in Basic [l’i]. most highly predictive thinking skills are logically
Accordingly, our analysis of whether or not think- related to Basic programming (i.e., are specific think-
ing skills enhance programming ability is concerned ing skills or component skills), but not identical to
mainly with construct validity (in addition to predic- information taught in Basic instruction. These re-
tive validity). Another way of stating this focus is to sults are consistent with a single study reported by
say that our analysis of this assertion is concerned Snow [N], in which skill at problem representation
with identifying specific thinking skills (in addition was related to learning Basic.
to general thinking skills). This part of our study demonstrates how it is pos-
The issue of which specific thinking skills are re- sible to pinpoint specific thinking skills that are re-
lated to the learning of programming has been ex- lated to learning a programming language. As ex-
amined in only a few studies. Snow [24] reports that pected, success in learning Basic was related to gen-
succ:ess in learning Basic is more strongly related to eral intellectual ability, especially logical reasoning
“diagraming” (r = 0.66)-a problem representation and spatial ability. More importantly, this study
skill specifically related to programming-than to identified two specific thinking skills that are based

606 Communications of the ACM ]uly 1986 Volume 29 Number 7


Special Section

on a cognitive task analysis of Basic [17]: ability to 3. Write down the number from Box B.
translate word problems into equations or answers 4. Stop working on this.
(problem translation skill), and ability to predict the If the student gave the correct answer, the next
outcome of a procedure or set of directions that is problem was presented. If an error was made, the
stated in English (procedure comprehension skill). correct answer was given, and the student could re-
The search for additional specific thinking skills fer to an English version of the manual. After solving
represents a potentially fruitful direction for future 60 similar problems, students in the pretraining
research. group were transferred to learning Basic by predict-
ing the output of 40 simple Basic programs, as
ASSERTION 3: PRETRAINING ON CERTAIN
described above.
THINKING SKILLS WILL ENHANCE THE
The results indicated that the students who were
LEARNING OF PROGRAMMING
The first two sections of this article provide empiri- given pretraining in predicting the output of English
procedures learned Basic much faster than those
cal support for the idea that the ability to learn Basic
with no pretraining. For example, on the first set of
is predicted in part by two specific thinking skills,
40 Basic problems, the pretrained group averaged
and that an outcome of learning Basic is improve-
about 6 seconds per answer to predict the output of
ment in these two skills. The next logical step is to
Basic programs compared to over 12 seconds per an-
determine if direct instruction in these predictive
thinking skills can foster the learning of Basic. swer for the students who had received no pretrain-
ing. A t-test revealed that this difference was statisti-
This issue has not been convincingly addressed in
cally significant at p < 0.001. When we compared the
existing published research. In some of our previous
work [2, 161, we have given pretraining in appropri- group with no pretraining after 60 Basic problems to
ate mental models for various programming lan- the pretraining group (who received 60 equivalent
guages. Pretraining tended to enhance students’ sub- English procedure problems), the pretrained group
sequent learning of programming languages, espe- averaged about 6 seconds per answer to predict the
cially for those who lacked computer programming output of Basic programs, whereas those with no pre-
aptitude. This line of research only indirectly in- training averaged about 6.5 seconds. A t-test here
failed to reveal any significant difference between
forms our analysis of this issue, since the pretraining
the groups. The results may be summarized by say-
is not on specific thinking skills.
A preliminary study, conducted in our labs at ing that pretraining in procedure comprehension is
at least as effective as isomorphic pretraining in
Santa Barbara by Jenny Dyck, addresses this issue. In
Basic. These findings are interesting because they
Dyck’s study, 23 randomly selected college students
show that pretraining in procedure comprehension
(no pretraining group) learned Basic by reading a
(involving English) provides a foundation for learn-
manual and through exercises in predicting the out-
ing Basic. A straightforward conclusion is that proce-
puts of simple Basic programs. For example, a typi-
dure comprehension is a component skill in learning
cal problem was the following:
Basic, and that this skill can be taught to novices.
Determine the output of this program:
10 LET A = 3 CONCLUSIONS
20 LET B = A + 5
Several scholars, including many proponents of
30 PRINT B
Logo, have asserted that learning to program will
40 END
enhance thinking skills in domains outside of pro-
gramming. Anecdotal and personal introspective
If the student gave the right answer, the next pro- data are the two principal sources of evidence. Un-
gram was given. If the student made an error, the fortunately, both are notorious for their unreliability
correct answer was given, and the student could and thus their unsuitability as scientific evidence.
refer to the manual. In all, students solved 100 such Methodologically sound experimental studies in this
problems. area are almost nonexistent. Our study encourages
In contrast, 23 other randomly selected college the idea that learning to program can have positive
students (pretraining group) first received practice in effects on thinking skills that are directly related to
predicting the output of procedures that were stated the language to be learned. At present, however,
in English (see also the following procedures test there is no convincing evidence that learning a pro-
described in the sidebar). For example, a typical gram enhances students’ general intellectual ability,
problem was the following: or that programming is any more successful than
1. Put the number 3 in Box A. Latin for teaching “proper habits of mind.”
2. Add 5 to the number in Box A; The assertion that certain thinking skills will en-
put the result in Box B. hance a person’s learning of programming has also

July 1986 Volume 29 Number 7 Communications of the ACM 609


Special Sectiotl

spawned a line of research that is subject to both 4. Bark. A. Leaming with Computers. Digital Press, Bedford, Mass.. 1961.
5. Bransford. I.D.. Arbitman-Smith. R.. Stein, B.S., and Vye. N.1.
methodological and logical flaws. A common meth- Improving thinking and learning skills: An analysis of three
odological flaw is the “shotgun approach,” in which approaches. In Vol. 1. Thinking and Learning Skills, J.W. Segal.
SF. Chipman. and R. Glaser, Eds. Erlbaum, Hillsdale. N.J.. 1965.
many predictor variables are used so that a small pp. 133-206.
number might reach statistical significance. A com- 6. Caruso. D.R.. Taylor, J.J.. and Detterman. D.K. Intelligence research
and intelligent policy. In How and How Much Can Intefligeuce Be
mon logical flaw is the “correlation implies causa- Increased? D.K. Detterman and R.J. Sternberg. Eds. Ablex. Norwood,
tion fallacy”: If A predicts B, it does not mean that A NJ.. 1982, pp. 45-65.
7. Clements. D.H., and Cullo. D.F. Effects of computer programming
causes B. To avoid these problems, we suggested us- on young children’s cognition. 1. Educ. Psycho/. 76, 6 (Dec. 1964).
ing predictor tests that are selected on the basis of 1051-1058.
6. Dalbey. 1.. and Linn. M.C. The demands and requirements of com-
construct validity, that is, theoretically related to puter programming: A literature review. I, Educ. Compuf. Res. I, 3
learning a particular language. Although prior stud- (Summer 1985), 253-274.
9. Denelskv. G.Y.. and McKee. M.G. Prediction of comnuter oroaram-
ies have often found evidence that general abilities, 1 .I

mer training and job performance using the AABP Test. Pers.
such as logical reasoning, are predictive of learning Psycho/. 27 (1974). 129-137.
IO. Corman H.. and Bourne. L.E. Learning to think by learning LOGO:
programming, such research does not explain the Rule learning in third grade computer programmers. Bull. Psycho-
underlying mechanisms of that learning process. nomic Sot. 21 (1983). 165-167.
11. Gregg. L.W. Spatial concepts, spatial names. and the development of
The exemplary predictor study presented in this ar- exocentric representations. In Children’s Thinking: What Develops!
ticle demonstrates that it might be possible to iden- R. Siegler. Ed. Erlbaum, Hillsdale. N.J.. 1976, pp. 275-290.
12. Hollenbeck. G.P., and McNamara, W.J. CUCPAT and programming
tify specific information processing skills, based on a aptitude. Pus. Psychol. I8 (1965). 101-106.
cognitive analysis of a programming language, that 13. Linn, MC. The cognitive consequences of programming instruction
in classrooms. Educ. Res. 74, 5 (May 1985). 14-16. 25-29.
serve as predictors of learning programming. At 14. Lochhead. 1. An introduction to cognitive process instruction. In
present, problem representation and procedure com- Cognifiue Process Instruction. J. Lochhead and J. Clement, Eds. Frank-
lin Institute Press. Philadelphia, Pa., 1979. pp. l-4.
prehension are two likely specific thinking skills re- 15. Mansfield, R.S.. Busse. T.V., and Krepelka, E.J. The effectiveness of
lated to learning Basic; future research should be creativity training. Rev. Educ. Res. 48, 3 (Summer 1976). 517-536.
16. Mayer. R.E. The psychology of how novices learn computer pro-
directed at lengthening the list. gramming. Conlpul. Sure. 13. 1 (Mar. 1961). 121-141.
The issue of pretraining follows from the foregoing 17. Mayer. R.E. Learning in complex domains: A cognitive analysis of
computer programming. Psychol. Leanr. Motiv. 19 (1985). 89-130.
two assertions, Although the concept of “readiness 18. Mayer. R.E. The elusive search for teachable aspects of prob-
skills” has received wide acceptance in mathematics lem solving. In Hisfoy of Eduratimal Psychology. I. Glover and
R. Ronning. Eds. Academic Press, New York. To be published.
and language arts, very little is known concerning 19. McNamaer. W.J.. and Hughes, J.L. A review of research on the selec-
what a person needs to know to successfully learn to tion of computer programmers. Pers. Psycho/. 14 (1961). 39-51.
20. Nickerson. R.S. Computer programming as a vehicle for teaching of
program. At the present time, it appears that pre- thinking skills. Third&g 4 (1982). 42-48.
training in procedure comprehension skills transfers 21. Papert. S. Mindstorm. Basic Books. New York. 1960.
22. Pea. R.D.. and Kurland. M.K. On the cognitive effects of learning
to learning of Basic. Additional research is needed, computer programming. New Ideas in Psychol. 2. 2 (Spring 1965). 137-167.
however, to determine which other “predictor skills” 23. Rippa. S.A. Education in a Free Society: An American History.
Longman. New York. 1960.
might also serve as “readiness skills.” 24. Snow. R.E. Aptitude processes. In Vol. 1. Aptitude, Learning, and
Careful empirical research can inform the contro- Insfruction, R.E. Snow. P. Federico. and W.E. Montague. Eds.
Erlbaum. Hillsdale. N.]., 1960. pp. 27-63.
versy concerning the teaching of Basic in schools, by 25. Soloway. E.. Lochhead, I.. and Clement, J. Does computer program-
evaluating assertions concerning the relation be- ming enhance problem solving ability? Some positive evidence
on algebra word problems. In Computer Literacy. R.J. Seidel.
tween programming and problem solving. The em- R.E. Anderson. and B. Hunter, Eds. Academic Press. New York,
pirical research presented in this article suggests 1962. pp. 171-185.
26. Thorndike, E.L. The influence of first-year Latin upon the ability to
that there is an important-albeit limited-relation- read English. Sch. Sot. 17 (1923). 165-168.
ship between a person’s thinking skills and ability to 27. Webb, N.M. Microcomputer learning in small groups: Cognitive re-
quirements and group processes. 1. Educ. Psycho/. 76, 6 (Dec. 1964).
learn Basic. These preliminary results suggest that 1076-1088.
the most fruitful way to search for a relation be-
CR Categories and Subject Descriptors: H.1.2 [Models and Princi-
tween thinking skills and programming is to focus ples]: User/Machine Systems-human factors: K.3.2 [Computers and
on thinking skills that are cognitive components of Education]: Computer and Information Science Education-informafion
systems educafion
programming-specific thinking skills that are ele- General Terms: Human Factors
ments in a cognitive task analysis of programming- Additional Key Words and Phrases: Basic, human-computer interac-
tion, psychology
rather than to focus on general intellectual ability.
Authors’ Present Addresses: Richard E. Mayer. Jennifer L. Dyck, and
REFERENCES William Vilberg. Dept. of Psychology, University of California. Santa
1. Bauer, R.. Mehrens. W.A., and Visonhaler, J.R. Predicting perfor- Barbara, CA 93106: Jennifer L. Dyck, Dept. of Computer Science, Califor-
mance in a computer programming course. Educ. and Psychol. Meas. nia State University, Fresno. CA 93740.0109.
28 [1968), 1159-1643.
2. Bayman, P. Effects of instructional procedures on learning a first Permission to copy without fee all or part of this material is granted
programming language. Ph.D. dissertation, Dept. of Psychology, provided that the copies are not made or distributed for direct commer-
Univ. of California. Santa Barbara. 1963. cial advantage, the ACM copyright notice and the title of the publication
3. Bayman, P., and Mayer, R.E. A diagnosis of beginning programmers’ and its date appear. and notice is given that copying is by permission of
misconceptions of Basic programming statements. Commun. ACM 26, the Association for Computing Machinery. To copy otherwise. or to
9 (Sept. 1983). 677-679. republish, requires a fee and/or specific permission.

610 Communications of the ACM july 1986 Volume 29 Number 7

Vous aimerez peut-être aussi