Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
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
COMPREHENSION OF MATHEMATICAL
RELATIONSHIPS EXPRESSEDIN GRAPHS
FRANCESR. CURCIO,
Queens College of the City Universityof New York
This study was supported by grants from the National Institute of Education (Grant
No. NIE-G-80-0093) and St. Francis College. It is based on the author's doctoral
dissertation, completed at New York University in 1981 under the direction of Ed-
ward M. Carroll. Thanks are due to M. Trika Smith-Burke and Robert G. Malgady,
New York University, and Stephen M. Kosslyn, Harvard University, for their com-
ments on an earlier version. Also, thanks to the anonymous reviewers for their
suggestions.
FRAMEWORK
SCHEMA-THEORETIC
A knowledge of the topic, content, and form of general discourse depends
upon the amount of previous meaningful exposure to the topic, content,
and form of the discourse the reader has had. This exposure contributes to
the development, revision, modification, and editing of related schemata.
The topic of general discourse is usually identified by its title and assists
the reader in retrieving from memory relevant familiar information pertain-
ing to the passage to aid in comprehension (Bransford & Johnson, 1973;
Sjogren & Timpson, 1979). The topic of a graph, which is identified by the
title, labels on axes, and key vocabulary words used in the title and labels,
may be one of the factors that requires prior knowledge for comprehending
the mathematical relationships expressed in the graph (Culbertson & Pow-
ers, 1959; Harper & Otto, 1934; Washburne, 1927).
The content of text material is the relationship between words and ideas,
the familiarity of which allows the reader to recognize, for example, a cause-
effect relationship (Pearson & Johnson, 1978). The mathematical content
of a graph, which is the number concepts, relationships, and fundamental
operations contained in it, is a second factor about which prior knowledge
seems to be necessary for comprehension (Goetsch, 1936; Thomas, 1933;
Vernon, 1952).
The form of a reading passage is its structure or framework, which em-
ploys certain conventions. Knowledge about these conventions allows the
reader to make predictions and impose certain expectations about the text
(Royer & Cunningham, 1978). The form or type of graph, such as a line
graph, bar graph, pictograph, or circle graph, appears to be a third factor
about which prior knowledge is necessary (Janvier, 1978; MacDonald-Ross,
1977). (For an extensive discussion of topic, content, and form, see Curcio,
1981a, 1981b.)
The purpose of this study was to extend the schema-theoretic perspective
of understanding general discourse to graph comprehension by examining
the effect of prior knowledge on the ability to comprehend the mathematical
relationships expressed in graphs. Because reading achievement and math-
ematics achievement are general predictors of success in school-related
tasks, these were also examined to determine whether prior knowledge
contributes significantly to the ability to comprehend graphs over and above
the contribution of reading and mathematics achievement.
The study focused attention on the performance of fourth and seventh
graders. By the fourth grade, most of the elementary work with graphs
should have been accomplished, and children should have achieved a suffi-
cient command of reading and arithmetic skills, the tools of learning nec-
essary for reading graphs (Strickland, 1938/1972). By the seventh grade, I
expected that growth and achievement in graph-reading ability would have
occurred (Bamberger, 1942), and any sex-related differences would be man-
ifested (Armstrong, 1975; Callahan & Glennon, 1975; Fennema, 1974,
METHOD
Subjects
The sample, restricted to native speakers of English, consisted of 204
fourth graders (101 boys and 103 girls) and 185 seventh graders (102 boys
and 83 girls) from four elementary schools, two junior high schools, and
one K-8 school located in a stable, middle-class community in 1 of the 32
New York City School Districts. Native English-speaking children were
selected so that inability with language would not be a confounding factor.
The superintendent and principals expressed interest and granted permis-
sion for the study to be conducted in their schools. As required by the public
schools, parents gave written consent for their children to participate in the
study.
Variables
Graph comprehension was measured by a researcher-designedGraph Test
composed of twelve graphs: three bar graphs, three circle graphs, three line
graphs, and three pictographs. Six multiple-choice items were constructed
for each graph. The six items reflected three tasks of comprehension: two
questions were literal items (requiring a literal reading of the data, title, or
axis label); two questions were comparison items (requiring comparisons
and the use of mathematical concepts and skills to "read between the data");
and two questions were extension items (requiringan extension, prediction,
or inference to "read beyond the data"). One of the graphs with its six
comprehension questions is presented in Figure 1. (For a detailed description
of test construction and the instruments used in the study, see Curcio,
1981a, 1981b.)
Prior to the study, the reliability of the Graph Test was estimated. Seventy-
five fourth graders and 67 seventh graders from a K-8 school in the school
district of the main study were given the test in the spring of 1980. All the
w00 w
,-
0
a
-W0"
.~4
0
~4,1-4-1
.0-
"i =
U
-H r-4o0
4-4
-4 0
,- 4 C -4A 4-4
0o4) U u) u)
,-4)
W
0 41 oto ,0-
0 4 4
coo-2 0 U,
U c 0 C
4-1
,,'
I=
, •
co 0
4)
4-1
0
c
C14 0
C14J
U
1)
ci
H
)N
o• 4-4 4-J1
W4) to
*H
~
4.1-4-1co- CO
4) O 41 0 4-.
*H r -4 4) r-4 W
,4) 4
U , 4 ,
,•r-.
U, r- ~0e-.
4- C
,
H
-A
co
4-1
r-4
r-4
co
r-4
to
4-1
r-4
,-A
CO
0
U,
U
4)
0
0
41-
4
4)
>
c: CO I
•$4
4)r
0
0
Uh
?4 U4 co 0 U
CJ 4) ?4 41 41 $ 4 >c
4) 4 ? 4 4 4-1 W4) co 0 I= 0 o
4. 4.J ) In 4) r-4 ?Q 4) O a) a) (au *H
44 4) 4)
co 4) ,- ? 4 .14 -J
.1 4-1 0 r-4 ?Q 1 10 4-E
am a) ) a) a) 0 -H co ,- 4-1,- 0
4-1 r-4 0 0 0 ?t 4-1 4-1 , - , A - ,-A Uf
U 4-1 4-1 ,-A H -H a) *H 0 r, *H r-4 -H r-4 0 ,0 f0
co U U 0 0 a) co 4-1 4-1 c 4-1 s-4 a) a) -H t *H U0 H -H U a) *H H
at 0 ) 4) C 4-1 U
i w U Q -H w
H
,-.4
H U U 4-1 0
o \-A
co
c
a
U u
r) ,
*H
a
u
)oo ca 0 r-4 '
c
rla) 4)
.Co
\H 4) 0
w
0
0 0
?4
co
4 ? 6
,-4 f) 00 f
CJ
U~ CO
c0
'0
)
0
0 c
.1 U\H
I
Ln 0 iL i
.-4 ,-4
*H
.1
C
U)
0
.1 U),
o'0
H
?
'0
aw
10
w)-H
U) 10
w)
1
w
0r-w00
P4 c'j un r- r- ctv rm 4) 0 P4 pw c P4 P4
0 0 0 '-4
** 0 1 1, 4 -C 4-4 0
SC4 C4ULA
42
00
m0-
0z-H $4 $
z
'a . ~ 40 0 0 0
41 *H m
a) 0 a)m H0 am C4 m
4-
Ii
HI I
04)04)
I0I401 ua3 ) 4
a)
-4
H
4
a
-4
I=H~
)*
I= 0
)*
U
0I=
Cr-
0Cr-CUCJC'C U, C CJ
0 co co
HH H ' - - t64 -
>
-W-a4)o 4) - ) ) 4
$4$a) $4 W $
children were native speakers of English. KR-20 reliabilities of .89 and .88
were obtained.
Prior knowledge was measured by a researcher-designedPrior Knowledge
Inventory, consisting of three subtests (Topic, Mathematical Content, and
Graphical Form). The inventory was designed to match the topic, mathe-
matical content, and graphical form of each of the twelve graphs. Items that
match the graph in Figure 1 can be found in Table 1.
Prior to the study, the reliability of the Prior Knowledge Inventory was
estimated. Sixty-seven fourth graders from the K-8 school in the district
were given the inventory during the spring of 1980. A KR-20 reliability
coefficient of .97 was obtained. The subtests (Topic, Mathematical Content,
and Graphical Form) had reliabilities of .93, .97, and .93. Seventy-three
seventh graders were given the same test and a KR-20 reliability coefficient
of .96 was obtained. The subtests had KR-20 reliabilities of .90, .95, and
.86.
Reading and mathematics achievement were measured by Level D and
Level F (for fourth and seventh grades, respectively) of the Reading and
Mathematics tests of the SRA Achievement Series (Naslund, Thorpe, &
Lefever, 1978).
Sex was measured by a dichotomous variable (0 for boys; 1 for girls).
Procedure
Nineteen seniors in a college teacher-training program were recruited and
trained as test proctors. Each had completed a course in tests and measure-
ments and attended one orientation session to insure that uniform testing
conditions and procedures were followed.
During the fall of 1980, each of the four tests was administered by the
proctors during one of four testing sessions. Data on sex and native-English-
speaking status were collected on the cover sheet of the Prior Knowledge
Inventory, the first test given.
Data Analysis
Correlational and multiple regression analyses were computed by grade.
To avoid having the results confounded with other cognitive components
(S. M. Kosslyn, personal communication, 21 January 1981), second-order
partial correlations were computed to determine the unique contribution of
prior knowledge of topic, of mathematical content, and of graphical form
to graph comprehension, partialing out reading and mathematics achieve-
ment. First-order partial correlations of graph comprehension with reading
and mathematics achievement, controlling for mathematics and reading
achievement, respectively, were also computed. These coefficients can be
found in Curcio (1981a, 1981b).
Table 1
Sample Items From the Prior Knowledge Inventory
Prior Knowledge of Topic
1. What does "height" mean in the following sentence?
In school today, the teacher measured Tommy's height.
a. How much Tommy weighs
b. How old Tommy is
c. How smart Tommy is
d. How tall Tommy is
2. How can we determine who is the tallest in the class?
a. By eating a lot of food
b. By sleeping the most
c. By standing next to one another
d. By dressing properly
Prior Knowledge of Mathematical Content
3. Which of the following is a correct statement?
a. 1 centimeter is greater than 1 inch
b. 1 inch is less than 1 centimeter
c. 1 inch equals 1 centimeter
d. 1 centimeter is less than 1 inch
4. 105 - 85=
a. 20
b. 25
c. 80
d. 190
Prior Knowledge of Graphical Form
Use the following picture to answer questions 5 and 6:
100-
25-T
A B C D
Prediction of Comprehension
The results of the regression analyses are reported in Table 4 for Grade
4 and Table 5 for Grade 7. As expected, for both grades, reading and
mathematics achievement, the general predictors of success in school-related
tasks, account for the greatest portion of the variance of graph comprehen-
sion. As Equations 2 and 3 in each table show, the percentage of the variance
accounted for by reading achievement is similar to that accounted for by
mathematics achievement.
Table 4
Regression Analysis With Graph Comprehension as the Dependent Variable,
Grade 4
Variables
Equation in equation R2 AR 2 df F
1 1 .000 (1,202) 0.098
2 1,2 .491 .491 (2,201) 96.856*
3 1,3 .454 .454 (2,201) 83.472*
4 1,2,3 .598 .598 (3,200) 99.153*
5 1,2,3,4 .618 .020 (4,199) 80.569*
6 1,2,3,5 .656 .058 (4,199) 95.039*
7 1,2,3,6 .620 .022 (4,199) 81.172*
8 1,2,3,4,5,6 .681 .083 (6,197) 70.129*
Note. Codefor variables:1 = sex; 2 = readingachievement;
3 = mathematicsachievement;
4 = topic; 5 = content; 6 = form.
*p < .01.
Table 5
Regression Analysis With Graph Comprehension as the Dependent Variable,
Grade 7
Variables
Equation in equation R2 AR2 df F
1 1 .022 (1,183) 4.173*
2 1,2 .474 .452 (2,182) 81.951**
3 1,3 .478 .456 (2,182) 83.198**
4 1,2,3 .600 .578 (3,181) 90.311**
5 1,2,3,4 .608 .008 (4,180) 69.905**
6 1,2,3,5 .644 .044 (4,180) 81.564**
7 1,2,3,6 .604 .004 (4,180) 68.758**
8 1,2,3,4,5,6 .647 .047 (6,178) 54.359**
Note. Codefor variables:1 = sex; 2 = readingachievement;
3 = mathematicsachievement;
4 = topic; 5 = content; 6 = form.
* < .05.
p
**p < .01.
When each of the three aspects of prior knowledge (topic, content, and
form) was entered (see Equations 5, 6, and 7 in Tables 4 and 5), although
the contribution to the variance was small, it was significant (p < .01) in
each case.
To determine the optimal linear combination of sex, reading and math-
ematics achievement, and prior knowledge of topic, of mathematical con-
tent, and of graphical form in predicting graph comprehension, the beta
weights for each grade were calculated (see Table 6). Predictors of graph
Table 6
Variables in the Graph Comprehension Regression Equations at
Grades 4 and 7
Grade 4 Grade 7
Variable Beta F(1,197) Beta F(1,178)
Sexa -0.05 1.49 0.01 0.03
Mathematics
achievement 0.21 14.13** 0.30 21.67**
Reading
achievement 0.29 23.76** 0.33 23.32**
Topic 0.17 9.25** 0.04 0.30
Content 0.26 28.53** 0.26 17.95**
Form 0.11 3.98** 0.05 0.78
aBoys = 0; girls = 1.
*p < .05.
**p < .01.
Table 7
Zero-Order Partial Correlations for Grade 4 (n = 204) and Grade 7 (n = 185)
Reading Mathematics Graph
Variables Sexa Topic Content Form achievement achievement comprehension
Sexa - .00 .09 .00 .10 .09 .02
Topic .11 -
.37** .53** .65** .55** .63**
Content .06 .54** - .48** .43** .52** .61*
Form .01 .45** .39** - .60** .58** .62**
Reading
achievement .16* .70** .50** .47** - .58** .70*
Mathematics
achievement .23* .46** .62** .40** .58** - .67*
Graph
comprehension .15* .57** .65** .44** .69** .69** -
DISCUSSION
The results of this study support previous research indicating that there
are no significant sex-related differences with respect to graph comprehen-
Suydam, M. N., & Riedesel, C. A. (1969). Interpretive study of research and development in
elementary school mathematics (Vol. 1). University Park: Pennsylvania State University.
(ERIC Document Reproduction Service No. ED 030 016)
Thomas, K. C. (1933). The ability of children to interpret graphs. In G. M. Whipple (Ed.), The
teaching of geography (32nd Yearbook of the National Society for the Study of Education,
pp. 492-494). Bloomington, IN: Public School Publishing Co.
Vernon, M. D. (1952). The use and value of graphical methods of presenting quantitative data.
Occupational Psychology, 28(1), 22-34.
Washburne, J. N. (1927). An experimental study of various graphic, tabular and textual
methods of presenting quantitative material, Part 1. Journal of Educational Psychology, 18,
361-376.
AUTHOR
FRANCES R. CURCIO, Assistant Professor, School of Education, Queens College of the City
University of New York, Flushing, NY 11367