Académique Documents
Professionnel Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=nctm. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
National Council of Teachers of Mathematics is collaborating with JSTOR to digitize, preserve and extend
access to Journal for Research in Mathematics Education.
http://www.jstor.org
Journal for Research in Mathematics Education
1981, Vol. 12, No. 2, 119-141
EFFECT
INITIAL INELEMENTARY
OFCALCULATORS
SCHOOLMATHEMATICS
Purpose
The purpose was to determine the effect the availability of calculators to
students and the availability of calculator-related curriculum resources,
consultant resources, and in-service workshops for teachers had on the ele-
mentary school children's attitudes and achievement in mathematics,
Grades 2-6.
There were many possible effects one might reasonably expect, such as
changes in parent willingness to make calculators available in the home,
differences in teacher acceptance and use of calculators, modification of the
Characteristic 1 2 3a 4 5
Number of students in 780 348 620 (K-3) 745 420
building 550 (4-6)
Attendance rate (Percent) 97 96 92 (K-3) 96 95
93 (4-6)
Average class size (2-6) 27.2 23.2 28.7 25.3 25.5
Percent minority 0 6 50 6 2
Percent bussed 100 19 45 50 18
County population (1970) 109,378 72,127 907,872 80,911 833,249
School type Rural Small Large Small City Large City
City Urban Suburban Suburban
Income level Low- Middle Low High Middle
Middle
Building constructed 1968 1954,1963 1957 1964 1955,1958,
1968
a Site housed in two
buildings K-3, 4-6.
Table 2
Site Characteristics:
Grade Equivalents for October Standardized Testing by Site
Site
Grade
level Test 1 2 3 4 5
Gradea 2 Concepts 2.1 2.7 2.0 2.8 2.7
Computations 2.1 2.4 2.0 2.7 2.6
Application 1.9 2.6 2.1 2.9 2.9
Grade 3 Concepts 3.0 3.7 2.7 4.0 3.7
Computations 2.6 3.4 2.8 3.7 3.4
Applicatipn 2.8 3.5 2.9 3.9 3.8
Gradeb 4 Concepts 4.3 4.8 3.3 4.7 4.8
Computations 4.2 4.5 3.8 4.8 4.3
Application 4.2 5.0 3.7 4.5 4.9
Grade 5 Concepts 4.9 6.1 4.2 6.4 5.3
Computations 5.4 5.7 4.6 6.5 5.7
Application 4.7 5.8 4.0 6.0 5.4
Grade 6 Concepts 6.1 6.5 4.5 7.5 6.5
Computations 6.0 6.8 5.1 7.4 6.4
Application 5.4 6.7 4.3 7.1 6.3
a Grades 2-3 Stanford Mathematics Achievement
Primary IIA.
b Grades
4-6 Stanford Mathematics Achievement Intermediate IA.
Design
The basic design was experimental including two treatment groups with
pre- and posttesting. Figure 1 is a summary of the time and sequence of
events. The experiment was conducted from the second week in October
1977 through the second week of February 1978. The pretests were given
the first week in October 1977 and the posttests were given during the third
week of February.
The treatment occurred during an 18-week time period which included
regular school holidays and days missed due to heavy snowfalls and bliz-
zard conditions. The treatments were in effect for an average of 66.6 school
days, ranging from 56 days at Site I up to 74 days at Site 4. This is equiva-
lent to 13.32 weeks of school time.
A follow-up period was included in the project to allow for the remedia-
tion of any detrimental effects, should they occur. The follow-up period ex-
tended from the middle of February through the first week in May.
Instruments
The testing included responses from students in five categories. The cate-
gories were attitudes, basic facts, mathematics achievement, estimation and
special topics. Classroom observations were made throughout the treatment
period.
Attitudes. The attitude scales were two six-item semantic differentials with
five response options. One scale was used to measure mathematics attitude
and the other to measure calculator attitude of the children. The six pairs of
terms used were bad--good, sad--happy, boring--exciting, jump in--hold
back, hard--easy, and more--less. These terms were selected as a result of
consultation with teachers and children. The instruments were piloted and
children interviewed to determine how valid the scales were for reflecting
the students' attitudes. The same scales were used for all grades (2-6). The
instructions and the scales were read to Grade 2 and Grade 3 students but
not to Grade 4 through Grade 6 students.
The scales were scored by assigning scores of 5 through 1 to the student
responses. If the student response was in the space adjacent to a term signi-
fying a positive attitude a score of 5 was assigned; if the response was next
to a term signifying a negative attitude a score of 1 was assigned. For ex-
ample:
---score---
(1) (2) (3) (4) (5)
bad -_ _ - :good
The scores for the six items were added together resulting in an attitude
I rI -A= - 0 C ot I
0
0,0IC4
0
CII . u CL CL
4c C40 E r- -A
C4 _ -000.o
R I
<I 0 E % 00
%4
$.--0 - 4)
* - cc - I -
4, r - 1-
o N<IU
:u
•• U,
•••• • -,
. )
•• I
0 V) CL .j tn .
-
F
I • • "
I <•. • <
0U
c,4
C4 E r- C4
'a.
s C4UC
4 CE
CC4
i
C. _ _ _ <:
< <
2!.
V,
-0 r-E 0. E
D
:* iS E m
U U <CS c ~
o -o,. -o • o uo
LA- a-
0 - <
r0> .0
0"- 0
r-
0
C U.I
0
l v
""' --. - .. 0" "-
41
r,,, p._ ''>
0. 4 0.tu'& 0. -0 "t
2 1 I I I
-- . I. I
*1 L.." .. )
0.00
4).9-,*IuU -%, E
-0 o--
o 00,.) 0 4.
0.. U0.- .! 00
-0ol U- la
OYO YFXC .-ooe0.1'
4)YY ~B
e' O ?P c
g >PY
0-0-0
-O M
-
: E
w
-5o
c E
o
-0. 'v ur- E -0.
~ f c
N>
Z
I.0.0
1-
i
1 CEaEpE" Pl-I.-t
eP~lgp~
ncC~ O cE888 ,
z•.
r-
.
c-•-.
.. p"
0 140- -0zi z I
o •. .•o ~ c
I - CNc- C -C-J
score with a maximum of 30 and minimum of 6. The internal consistency
reliability estimates of these scales ranged from 0.82 to 0.92 for Mathe-
matics Attitude (MAT) and from 0.66 to 0.88 for Calculator Attitude
(CAT).
Basic facts. The four basic facts tests (ADD, SUB, MUL, DIV) were each
made up of 20 randomly selected items (10 easy, e.g., 4 x 2, and 10 hard,
e.g., 8 x 7, basic fact combinations). These were read to students at all
grades (2-6) with a 5-second delay between items for responding. Calcu-
lators were not used.
The student responses were scored right or wrong. The scores on each
basic facts test could range from 0 to 20. The internal consistency reliability
estimates were from 0.55 to 0.89 for Addition, 0.88 to 0.90 for Subtraction,
0.84 to 0.94 for Multiplication, and 0.75 to 0.95 for Division. The low relia-
bility estimates for Addition were for Grade 6. The Addition basic facts
test was very easy for Grade 6 and the majority of errors were careless er-
rors resulting in low internal consistency estimates.
The teacher was instructed to read the item as follows: "Number 8. Which
number is smaller?" The teacher was instructed not to read the options.
An example of a Part II item is:
(circle one)
17. Mr. Jackson has 1253 2274 4780 4,419,331 28,148,443
cars to load on his boat.
Each car weighs 3527
pounds. How many pounds
will be added to the boat?
The teacher asked all children to turn on their calculators to see that the
calculators were working just prior to beginning Part II. For all Part II
items the teacher was instructed to read the item but not the options. The
total number of items answered correctly on Parts I and II was used as the
score on the primary level Special Topics test. The maximum possible
score was 18.
The Special Topics test for the intermediate level (SPCIA), Grades (4-6)
consisted of two parts (I, II). Part I consisted of 12 items and was taken by
students withoutthe use of the calculators. Part II consisted of 11 items and
was taken using calculators.
An example of a Part I item is:
5. Write the number for five
thousand seven hundred sixty- 5.
three and sixty-five thousandths.
The student was instructed to write the answer in the space provided. The
teacher did not read the questions to the students.
An example of a Part II item is:
(circle one)
14. Which number is bigger? 3/7 5/13
Results
The results include the analysis of pretest and posttest data relevant to
the hypotheses stated earlier. The purpose of the pretest analysis was to
identify the level of achievement and attitudes of students prior to treat-
ments and determine whether or not pretest differences would require co-
variance procedures in subsequent analyses.
Pretests
The pretests were given 3-6 October 1977. Tables 3 and 4 give margin
means for Grade and Treatment contrasts and summarize the multivariate
and univariate analyses of variance for the pretest measures. The data for
Grades 2-3 were analyzed separate from the data for Grades 4-6. Since no
interaction effects were significant, the interaction contrast was omitted
from the summary tables. In the primary grades (2-3) the expected grade
level differences were found on basic facts and mathematics achievement
while no attitude differences were found between grades. The treatment
groups were assumed to be equivalent prior to the beginning of the experi-
ment since no evidence of treatment group differences were found.
The pretest data for Grades (2-3), Table 3, result in the Grade Level dif-
ferences that are generally expected for ADD, SUB, and MUL basic facts.
Table 4
Cell Means,MarginMeans,and SignificanceLevels
for Math and CalculatorAttitudePretestsby Treatmentfor Grades2-3
Attitude
Treatment grade MAT CAT
Second 23.7 27.1
NC Third 21.3 28.1
Combined 22.5 27.6
Second 22.2 28.1
C Third 23.4 27.9
Combined 22.8 28.0
Overall 22.6 27.8 p .001
Table 6
Cell Means, Margin Means, and Significance Levels
for Math and Calculator Attitude Pretests by Treatment for Grades 4-6
Attitude
Treatmentgrade MAT CAT
Fourth 20.6 27.6
Fifth 21.3 27.3
NC Sixth 21.5 26.7
Combined 21.1 27.2
Fourth 21.8 27.3
Fifth 20.5 28.0
C Sixth 21.6 26.2
Combined 21.3 27.2
Overall 21.2 27.2 p - .001
level differences that are generally expected for MUL and DIV basic facts
are found in Table 5. The students at the higher grade levels do better than
those at the lower grade levels. A similar interpretation can be made con-
cerning the Stanford Achievement Test mathematics scores.
Posttests
The posttests were given 13-17 February 1978. The basic facts (+,-,x,+)
and mathematics achievement (Concepts, Computations, and Applications)
tests were analyzed for pre-posttest differences. The same basic facts test
was used for Grades (2-6) and for pre- and posttests. These scores were an-
alyzed directly. The mathematics achievement scores for Concepts, Compu-
tations and Applications were obtained from different batteries of the Stan-
ford Achievement Tests depending on the Grade Level of the students and
16 16
5
12 12
4
8 8
3
4 4
2 MUL DIV
20 5 _ 20
4 --6 6
5
3
16 16 :3
12 12 2
8 8
4 4
ADD SUB
0o I 0
Pre Post Pre Post
Figure 2. Basic facts grade by testing time.
rize the margin means and the multivariate and univariate analyses for
posttest difference for Grades 2-3. No interaction effects were found; there-
Table 8
Grade Equivalents for Mathematics Achievement
by Treatment by Testing Time
No Calculator Calculator
Pretest Posttest Pretest Posttest
Concepts 4.4 5.4 4.4 5.5
Computations 4.3 5.0 4.3 4.9
Applications 4.2 5.2 4.1 5.2
Table 9
Margin Means and Significance Levels for
Grade Level and Treatment Contrasts for Posttests-Grades 2-3
Gradelevel Test Treatment Test
U M U M
2 3 p< p< NC C p< p<
Attitude .595 .569
MAT 23.1 22.4 .591 22.3 23.2 .444
CAT 27.0 27.2 .635 27.1 27.0 .805
Basicfacts .001** .509
ADD 17.2 19.0 .009** 17.7 18.5 .210
SUB 15.0 18.2 .001** 16.7 16.6 .880
MUL 4.4 13.0 .001** 8.6 8.8 .915
DIV 2.5 9.7 .001** 5.9 6.4 .770
Mathematics
achievement .001** .850
CNP3A 13.3 19.4 .001** 15.8 16.9 .515
CMP3A 13.3 21.2 .001** 17.1 17.4 .840
APP3A 13.8 19.1 .004** 16.1 16.8 .671
Calculator-
related
mathematics
achievement .002** .258
EST 3.6 4.7 .001** 4.0 4.2 .508
SPCPA 3.6 6.6 .001** 4.5 5.7 .091
CMIlA 20.3 27.8 .001** 23.8 24.3 .748
Note. NC-No Calculator; U-univariate; C-Calculator; M-multivariate.
* < .05.
p
**p < .01.
fore, only the grade level and treatment contrasts are reported in the sum-
mary. There were no treatment effects identified for Grades 2 and 3 as evi-
denced by the significance tests reported in Table 9. None of the
multivariate tests for differences between treatment groups were significant.
Only one variable out of the 12 measured showed even the slightest in-
dication of a statistical difference and that was SPCPA, the Special Topics
test (p < .091). The scores for the Calculator classes showed some sign of
being greater than those of the No Calculator classes. The difference be-
tween Calculator and No Calculator classes in experience with negative
numbers could have accounted for such a difference.
Across the basic facts measures and all achievement measures Grade 3
scores were significantly higher than Grade 2 scores (p - .01), which is to be
expected.
The grade equivalents associated with the Calculator Aided Computa-
tions (CMIlA) were 5.0 and 6.0 for Grades 2 and 3, respectively. This level
of computation performance, when the calculator was used while taking the
test, was not dependent on the treatment condition. The children learned to
use the calculator for calculations with ease. Assuming that second graders
would be expected to have grade equivalent scores of 2.5 and third graders
scores of 3.5 on the Computations test without calculators in February, the
Calculator Aided Computations scores of 5.0 and 6.0 were found to be sig-
-
nificantly higher (p .001) than would be expected for both second- and
third-grade children.
The second- and third-grade children did not do well on the Estimation
test. The mean score was 4.1, while the chance score was 3.0. Similar differ-
ences between attitudes toward mathematics and calculators observed from
the pretest scores were also observed in the posttest scores. A breakdown of
the attitude score means by Grade Level and Treatment are given in Table
10.
There were no Treatment effects identified for Grades 4-6 as evidenced
by the significance tests reported in Table 11. None of the multivariate tests
for differences between treatment groups were significant.
Statistically significant Grade Level effects were identified for the mathe-
matics achievement measures. These differences reflect an increase in per-
formance level from Grade 4 to Grade 6 which is to be expected. The grade
equivalents associated with the Calculator Aided Computations (CMADA)
were 5.8, 7.0, and 7.8 for Grades 4, 5, and 6, respectively. This level of com-
putational skill when the calculator was used while taking the test was not
dependent on the treatment condition. Performance was virtually the same
for Calculator and No Calculator groups. It was assumed that the fourth
graders would be expected to have grade equivalent scores of 4.5, the fifth
graders scores of 5.5, and the sixth graders scores of 6.5 on the Computa-
tions test without use of calculators in February.
The Calculator Aided Computations scores of 5.8, 7.0, and 7.8 for Grades
4, 5, and 6, respectively, were found to be significantly higher than would be
expected for fourth graders (p < .001), fifth graders (p -< .01), and sixth
graders (p < .01).
The Estimation scores show increased skill from fourth grade to sixth
grade. The means range from 2 to 3 points above chance. This is lower than
expected.
The multivariate Grade Level effect for attitude approached significance
(p < .065). On further inspection of the univariate tests of significance it can
be seen that the Calculator Attitude scores were primarily responsible for
this effect (CAT, p < .02). Inspection of the means across Grades (4-6) re-
veals a decrease in positive attitude from Grades 4 through 6. The older stu-
dents still viewed the calculator as something they liked, but perhaps they
learned that the calculator does not solve all of the problems for them.
Mathematics still requires thinking even if you use a calculator. This effect
was not dependent on treatment.
Table 12
Cell Means,MarginMeans,and SignificanceLevels
for Mathand CalculatorAttitudePosttestsby Treatmentfor Grades4-6
Attitude
Treatment grade MAT CAT
Conclusions
Widespread concern over the use of calculators by young children
prompted a broad, year-long study of the impact of calculators on elemen-
tary school mathematics learning. In the context of the level of calculator
use elementary teachers are likely to implement in the first year with calcu-
Table 14
Critical Treatment Differences Required
for 0.95 Power in Grades 4-6 (a = .05)
Possible raw
score range Pre Post
MAT 6-30 3.49 3.51
CAT 6-30 1.68 2.34
ADD 0-20 1.30 1.21
SUB 0-20 2.57 2.17
MUL 0-20 - 2.33
DIV 0-20 5.96 4.33
CNIIA 0-32 6.83 NG
CNI2A 0-35 NG 7.82
CMI1A 0-40 - NG
CMI2A 0-45 NG 10.21
API A 0-40 8.57 NG
API2A 0-40 NG 9.50
EST 0-12 NG 1.94
SPCIA 0-23 NG 5.36
CMADA 0-45 NG 7.39
Note. NG-Test not given.
lators available for all children and limited supplementary materials for stu-
dent use, the following conclusions seem warranted:
1. Children's attitude toward calculators is more positive than their atti-
tude toward mathematics.
2. Children grow significantly on basic fact and mathematics achieve-
ment tests taken without the use of calculators regardless of whether or not
calculators were used during instruction.
PhilosophicalQuestions
Although we did not attempt to do so, it would seem possible to use cal-
culators in such a way that algorithms such as the long division algorithm
need not be taught and would not be learned. The impact of such deletions
from the curriculum is unknown to us but should be considered. Societal
need of such algorithms would be one factor, mathematics curriculum need
of such algorithms another factor, and psychological need in learning
mathematics yet another factor. The replacement of these traditional al-
gorithms by others should be considered. Such issues cannot be resolved ef-
fectively by experimental research alone but rather in conjunction with sur-
vey, philosophical, and clinical research. We endorse such efforts as a
needed next step.
REFERENCES
Madden, R., Gardner, E., Rudman, H., Karlsen, B., & Merwin, J. Stanford Achievement Test,
Intermediate Level II Battery: Norm Booklet Form A. New York: Harcourt Brace Jovano-
vich, 1973.
Shumway, R. J. Hand calculators: Where do you stand? Arithmetic Teacher, 1976, 23, 569-572.
Stanford Achievement Tests. Mathematics Tests. New York: Harcourt Brace Jovanovich,
1972-1973.
Suydam, M. N. The Use of Calculators in Pre-College Education: A State-of-the-Art-Review.
Columbus, Ohio: Calculator Information Center, May 1979.