THE USE OF DIAGRAMS IN SOLVING NON ROUTINE

PROBLEMS
Marilena Pantziara, Athanasios Gagatsis & Demetra Pitta-Pantazi
Department of Education, University of Cyprus

This paper explores the role of diagrams in a specific problem solving process. Two
types of tests were administered to 194, 12 year old students, each of which consisted of
six non-routine problems that could be solved with the use of a diagram. In Test A
students were asked to respond to the problems in any way they whished whereas in Test
B problems were accompanied by diagrams and students were asked to solve these
problems with the use of the specific diagrams presented. The results revealed that there
was no statistical significant difference between the two tests. The result also revealed
that it was not the same group of students that were successful in the two tests.

INTRODUCTION
The mathematics education community has espoused the importance of developing
children’s problem solving skills (for example, National Council of Teachers of
Mathematics, 2000; Shoenfeld, 1992). In the same vein, research in mathematics
education discusses the importance of using multiple representations in the problem
solving process (Lesh, Behr, & Post, 1987; English, 1996). Markmann (1999)
interprets the term “representation” as the concept that includes the represented
world, a representing world, a set of rules that map elements of the represented world
to elements of the representing world, and a process that uses the information in the
representing world. Diagrams are considered as one kind of such representations
(Novick & Hurley, 2001). The represented world in this case is a description of a
problem to be solved, while the representing world contains the spatial diagrams as
an abstract form, along with their applicability conditions. Specifically, a diagram is a
visual representation that displays information in a spatial layout (Diezmann &
English, 2001). In problem solving a diagram can serve to represent the structure of a
problem. Diagrams are considered structural representations, in which the surface
details are not important and this is their main characteristic and differentiation from
pictures and drawings (Veriki, 2002). Diagrams typically rely on conventions to
depict both the components of the situation being represented and their organization.
These conventions must be learned and understood before the diagrams can be
understood and successfully used (Diezmann & English, 2001).
According to a number of researchers the ability to use diagrams is a powerful tool of
mathematical thinking and problem solving, because they are used to simplify
complex situations, they concretize abstract concepts, and they substitute easier
perceptual inferences for more computationally intensive search processes and
sentential deductive inferences (Novick & Hurley, 2001; Diezmann & English, 2001;

Proceedings of the 28th Conference of the International

Group for the Psychology of Mathematics Education, 2004 Vol 3 pp 489–496
Vol 3–32
 Kidman, 2002).This paper discusses a part of a larger study which aims to investigate
the impact of diagrams in solving non-routine problems.
THEORETICAL BACKGROUND
Networks, Matrices and Hierarchies
The efficient use of a diagram depends on its suitability for a given situation. The
appropriateness of a diagram depends on how well it represents the structure of the
problem. Novic and Hurley (2001) propose three general–purpose diagrams that suit
a range of problem situations, networks, matrices and hierarhies. These diagrams
highlight structural commonalities across situations that are superficially very
different. These diagrams are especially useful in elementary non-routine problems
(Booth & Thomas, 2000; Diezmann & English, 2001). Networks consist of sets of
nodes with one or more lines emanating from each node that link the nodes together.
Networks do not have any predefined formal structure (Fig.1). Matrices use two
dimensions to represent the relationships between two sets of information. A cell in
the matrix denotes the intersection of value i on one variable and value j on the other
variable (Fig.2). Hierarchies comprise diverging or converging paths among a series
of points. A single node gives rise to at least two other nodes (Novick & Hurley,
2001; Diezmann & English, 2001) (Fig.3).

Figure 1: Networks Figure 2: Matrices Figure 3: Hierarchies

In Networks the nodes specify values along a single variable. Any node may be
linked to any other node and all of them have identical status. In Matrices the rows
and columns specify values along two distinct variables. Values on the same
dimension may not be linked. All the rows have identical status, as do all of the
columns. In Hierarchies the nodes at a given level have identical status, but the nodes
at different levels differ in status. While in Networks the links between the nodes may
be unidirectional, in Matrices the links between the nodes are non directional, while
in Hierarchies the links are directional (Novick & Hurley, 2001; Diezmann &
English, 2001). Networks and Hierarchies may show paths connecting subsets of
nodes, while matrices do not show subsets.
Similar diagrams are often used in solving non routine problems. Non routine
problems are the problems that do not involve routine computations but the
application of a certain strategy, in this case a diagram, is required in order to solve
the problem (English, 1996). Non routine problems are considered more complicated
and difficult than routine problems in which only the application of routine
computations is involved in their solution (Shoenfeld, 1992). Research concerning
the efficiency of the use of diagrams in solving non routine problems is often

3–490 PME28 – 2004

 contradicting and thus inconclusive (Diezmann, & English, 2001; English, 1996;
Booth & Thomas, 2000). Thus, the present study purports to throw some light to the
contribution of the three diagrams in the solution of non-routine problems. More
specifically, the study investigated two questions:
(a) Does the presence of diagrams in non-routine problems increase students’
ability in solving them? and
(b) How does the presence of diagrams in non-routine problems affect students’
reponses?
(c) Does the use of networks, hierarchies and matrices (diagrams) facilitate students
in solving non-routine problems?
(d) Does the use of diagrams facilitate all students?
METHOD
To examine the impact of diagrams in the problem solving process we constructed
two tests. Test A consisted of six problems. The problems used in the study were
chosen based on three criteria. First, the problems were taken from previous pieces of
research, which showed that children used diagrams similar to the ones that we
intended to use in this study in order to solve them (Diezmann & English, 2001;
Booth & Thomas, 2000). Second, students were familiar to similar problems since
they appear in their mathematics textbooks, and third, the teachers’ guide book
suggests that these problems should be solved using similar diagrams with the ones
we indebted to use in the study.
The first problem inquired about the distance between four trees. Specifically,
students had to find out the distance between two trees while the distance of the other
trees was given. In the second problem students had to find the step that the cleaner
was standing when he started cleaning the windows. Problems 1 and 2 could both be
solved with the use of Network diagrams, because the problems had identical status
items (trees and steps) that may be linked with each other. In both problems it was
important to show paths connecting the items.
The third problem concerned combinations A � B � C. Children were asked to find out
all the possible combinations of two cards (Christmas or Easter cards), their colour
(green or yellow) and their ribbon’s colour (red or blue). The fourth problem asked
children to find out the wining team among four teams. Problems 3 and 4 could be
solved using the Hierarchy diagram because they involved items that were
distinguished according to different levels. In addition, the links between the items
were directional and only one route existed between any two items.
The fifth problem asked children to match each of the four friends with their own
favourite kind of books according to some given information. The sixth problem was
a combination problem A � B (four flavours of ice cream and three types of cones).
These two problems could be solved using the Matrices diagram because both of
them had two sets of items. The items within a set could not be combined amongst

PME28 – 2004 3–491

 themselves (for example the 3 different cones). All possible combinations of the
items across sets should be considered (cones and flavours).
Test A was administered to 194 Cypriot students in grade 6. Students were asked to
solve the six problems (a1-a6) in any way they wished. Test B contained isomorphic
problems, this means problems that had the same structure as those in the first test
and was administered to students a week after the administration of Test A. Each
problem in Test B (b1-b6) was accompanied by a diagram (bgi-bg6) that students
were asked to use in order to solve the problems.
For the analysis and processing of the data collected, statistical analysis was
conducted by using the computer software SPSS and CHIC (Bodin, Coutourier, &
Gras, 2000). A t-test analysis was produced to examine the difference between
students’ achievement in the two tests. The statistical package CHIC produces three
diagrams. The similarity diagram which represents groups of variables which are
based on the similarity of students’ responses to these variables. The implication
graph which shows implications A � B. This means that success in question A
implies success in question B. Finally the hierarchical tree which shows the
implication between sets of variables. In this study we use only the first two.
RESULTS
In regard to the first question of this study the mean score of Test A was 14,36 (20)
and the mean score of Test B was 14,55(20). The t-test analysis revealed that there
was no statistical significant difference between students’ achievement in the two
tests (M�-��=-0,196, df=193, p=0,636).
In regard to the second question of the study, according to the similarity diagram
(figure 4), students’ responses to the problems of the two tests are separated except
from the problem b1. The emergence of these two clusters of variables
(a1,a2,a3,a5,a6) and (b2,b6,b3,b4,b5) show that students perceive the two tests as
completely different tasks failing to realise that the two tests included isomorphic
problems. The similarity diagram (Figure 4) shows that students have not realized the
structural resemblance of the problems in the two tests. With the presence of the
diagrams in Test B they encountered the problems as completely different tasks. It is
not clear why b1 joined cluster a. It may be conjectured that the high level of
difficulty of problems a1 and b1 caused this similarity between them.
The similarity diagram also shows statistically significant similarity at level 99%
(Note: Similarities presented with bold lines are important at significant level 99%)
only between the problems in test A. Particularly, there is statistically significant
similarity between the variables a1-a3. The similarity between students’ responses in
problems a1 and a2 may be due to their similar structure. There is also a statistically
significant similarity between the group of variables a1-a3, and the group of variables
a5-a6. This can be explained in certain extend to their similar structure (a5 and a6, a1
and a2) and the level of difficulty of these non-routine problems. As it was pointed
out earlier on, these problems are considered difficult and therefore are usually solved

3–492 PME28 – 2004

 a a a a a a b b b b b b by mathematically able students.
1 2 3 5 6 4 1 2 6 3 4 5 On the contrary in Test B there
a a a a a a b b b b b b
does not appear to be any
statistically significant similarity
1 2 3 5 6 4 1 2 6 3 4 5

at level 99% between any of the

problems, suggesting that the
presence of each diagram in the
problems had a determinative role
to students’ responses. One
explanation that may be given is
that the appearance of the
diagrams was causing different
Figure 4: Similarity diagram of the behaviour to students resulting to
responses to the problems in the two tests different responses. Whereas in
Test A success in the problems
depended solemnly on students
mathematical abilities, in Test B students also had to interpret and use the diagrams
efficiently. This may have caused the lack of any similarity between students’
responses in Test B.
The implication graph in figure 5 shows significant implicative relations between the
problems in Test A. Specifically, it suggests that success in a1 implies success in a3.
Therefore in Test A, a1 was the most difficult and a4 the easiest problem. While in
Test A there is a hierarchy of difficulty, shown in the implication graph in figure 5,
this hierarchy does not exist in Test B. This implies that with the presence of the
diagram some problems became easier for some students, while other problems
became more difficult. In addition, the implication graph shows that students who
solved correctly a problem in Test A did not necessarily solved correctly the
isomorphic problem in the Test B and reversely.
In regard to the third question whether the use of the diagrams facilitates students
ability in solving non-routine problems, figure 6 gives some interesting information.
The implication graph shows that students who used the networks in b1 and b2 also
used the hierarchy in b3 and matrices in b6. However, this use does not imply the
successful solution of the foremost mentioned problems.

PME28 – 2004 3–493

 a1 bg1 bg2

a3

a6
bg4 bg5 bg3

a2

a5
bg6
a4
Figure 6: implication graph illustrating
Figure 5: implication graph illustrating
relations among the use of the diagrams
relations among the twelve variables,
bg1-bg6 and the correct answers to
a1-a6 and b1-b6.
problems b1-b6.
In regard to the fourth question, Table 1 shows in detail the students’ responses to the
two tests.
Wrong Responses Responses in Correct Responses in Responses in
in Test A Test B Test A Test B
A1 W:112 (58%) A1 C: 82 (42%)
B1 W: 84 (75%) B1 W: 41 (50%)
B1 C: 28 (25%) B1 C: 41 (50%)
A2 W: 77 (40%) A2 C: 117 (60%)
B2 W: 43 (56%) B2 W: 37 (32%)
B2 C: 34 (44%) B2 C: 80 (38%)
A3 W: 97 (50%) A3 C: 97 (50%)
B3 W: 57 (59%) B3 W: 33 (34%)
B3 C: 40 (41%) B3 C: 64 (66%)
A4 W: 22 (11%) A4 C: 172(89%)
B4 W: 12 (55%) B4 W: 23 (13%)
B4 C: 10 (45%) B4 C: 149 (87%)
A5 W: 62 (32%) A5 C: 132(68%)
B5 W: 27 (44%) B5 W: 41 (31%)
B5 C: 35 (66%) B5 C: 91 (69%)
A6 W: 75 (39%) A6 C: 119 (61%)
B6 W: 33 (44%) B6 W: 15 (13%)
B6 C: 42 (66%) B6 C: 104 (87%)
Table 1: Pupils’ responses to problems a1-a6 and their responses to the
isomorphic problems b1-b6. (W: Wrong responses, C: Correct responses)
The important aspect of the Table 1 is that although these problems are considered
difficult for primary school level and usually they are expected to be solved only by
mathematically able students, the presence of the diagram in Test B has been very
helpful for some students. However these students were not the same as those who
solved the isomorphic problem in Test A. This is evident by the fact that a number of
students that have solved the problem correctly in Test A they were not able to solve

3–494 PME28 – 2004

 the isomorphic problem in Test B. Some students seem to have failed to see the
representing structure in the diagram of the problem, while for other students the
diagram proved to be very helpful.
DISCUSSION
The present study is within the framework of the ongoing discussion about the role of
mathematical representations, and more specifically of diagrams, in the problem
solving process. The results of the study suggest that the presence of the diagrams in
Test B did not increase students’ ability in solving the non routine problems. This is
exemplified by many students’ failure to “see through” the diagram the structure of
the problem even though similar diagrams are used for the solution of these kinds of
problems in their classroom.
The presence of each diagram had a determinative role in students’ responses to each
problem. While students’ responses to Test A depended solemnly on students’
mathematical ability, in Test B students also had to interpret and use the diagrams
efficiently. It appears that not all students were able to do so. Success or failure to use
the diagrams correctly might be due to a number of reasons. First, students’ inability
to interpret the diagram correctly and second, some students’ lack of experience of
solving problems with the presence of diagrams.
The results of the study show that the efficient use of a diagram did not imply the
successful solution of a problem and reversely the successful solution of a problem
did not imply the efficient use of the accompanied diagram. It can be argued that
students perceived the problems with the accompanied diagrams as different tasks
failing to perceive each diagram as an additional aid for the solution of the problems.
The results of the study show that the diagrams make the problems easier for some
students while they make the problems more difficult for some other. For this reason,
students who solved the problems in Test B were not the same students who solved
the problems in Test A. It can be claimed that pupils with different visuo-spatial
abilities (Booth & Thomas, 2000) responded differently in the tests.
Diagrammatic literacy is an essential component of students’ mathematical
development (NCTM, 2000). In order to use diagrams efficiently, students must
develop the ability to translate the word problem into a diagrammatic representation
and the ability to interpret a diagram in terms to a given word problem (Novick &
Hurley, 2001; Diezmann & English, 2001). However, what is clearly suggested in
this study is that this is a skill that needs to be developed and is not inherent to all
students. A specific diagram does not have the same impact on all students. It is
likely that the diagrams presented may be incompatible to some students’ mental
representations. Therefore, it is very important for students to be engaged in multiple
representations in the problem solving process. Development of students’
diagrammatic literacy may only be achieved though carefully designed instructional
activities. These may turn diagrams into an effective tool for thinking.

PME28 – 2004 3–495

 References
Bodin, A., Coutourier, R., & Gras, R. (2000). CHIC : Classification Hierarchique
Implicative et Cohesitive-Version sous Windows – CHIC 1.2. Rennes: Association pour
le Recherche en Didactique des Mathematiques
Booth, R., & Thomas, M. (2000) Visualization in Mathematics Learning: Arithmetic
Problem-Solving and student Difficulties. Journal of Mathematical Behavior, 18(2),
169-190.
Diezmann, C., & English, L. (2001). . The Roles of Representation in School Mathematics:
2001 YearBook (pp.1-23). Virginia: NCTM
English, L. (1996). Children’s Construction of Mathematical Knowledge in Solving Novel
Isomorphic Problems in Concrete and Written Form. Journal of Mathematical Behavior
15, 81-112.
Kidman, G. (2002). The Accuracy of mathematical diagrams in curriculum materials. In
Cockburn A and Nardi E. (Eds) Proceeding of the PME 26 Vol.3 201-208.
UK:University of East Anglia.
Lesh, R., Post, T., & Behr, M. (1987). Representations and Translations Among
Representations n Mathematics Learning and Problem Solving. In C. Janvier (Ed.),
Problems of Representation in the Teaching and Learning of Mathematics (pp.33-40).
Hillsdale, NJ: Laurence Eribaum.
Markmann, A.B. (1999). Knowledge Representation. Mahwah, NJ: Erlbaum.
National Council of Teachers of Mathematics (2000). Principles and Standards for School
Mathematics, Reston, VA.: NCTM, 2000.
Novick, L., & Hurley, M. (2001). To Matrix, Network, or Hierarchy: That is the Question.
Cognitive Psychology, (42), 158-216.
Shoenfeld, A. (1992). Learning to think mathematically:Problem–solving, metacognition,
and sense making in mathematics. In D.A. Grouws (Ed.), Handbook of research on
mathematics teaching and leaning (pp.334-368). New York: Macmillan.
Vekiri, I., (2002). What is the value of Graphical Displays in Learning? Educational
Psychology Review, Vol. 14, No. 3, 261-312.

3–496 PME28 – 2004