1

The
British
Psychological
British Journal of Educational Psychology (2011)
C 2011 The British Psychological Society
! Society

www.wileyonlinelibrary.com

Are diagrams always helpful tools? Developmental

and individual differences in the effect of
presentation format on student problem solving
Julie L. Booth1 ∗ and Kenneth R. Koedinger2
1
Temple University, Philadelphia, USA
2
Carnegie Mellon University, Pittsburgh, USA

Background. High school and college students demonstrate a verbal, or textual,

advantage whereby beginning algebra problems in story format are easier to solve
than matched equations (Koedinger & Nathan, 2004). Adding diagrams to the stories
may further facilitate solution (Hembree, 1992; Koedinger & Terao, 2002). However,
diagrams may not be universally beneficial (Ainsworth, 2006; Larkin & Simon, 1987).
Aims. To identify developmental and individual differences in the use of diagrams,
story, and equation representations in problem solving. When do diagrams begin to
aid problem-solving performance? Does the verbal advantage replicate for younger
students?
Sample. Three hundred and seventy-three students (121 sixth, 117 seventh,
135 eighth grade) from an ethnically diverse middle school in the American Midwest
participated in Experiment 1. In Experiment 2, 84 sixth graders who had participated in
Experiment 1 were followed up in seventh and eighth grades.
Method. In both experiments, students solved algebra problems in three matched
presentation formats (equation, story, story + diagram).
Results. The textual advantage was replicated for all groups. While diagrams enhance
performance of older and higher ability students, younger and lower-ability students do
not benefit, and may even be hindered by a diagram’s presence.
Conclusions. The textual advantage is in place by sixth grade. Diagrams are not
inherently helpful aids to student understanding and should be used cautiously in the
middle school years, as students are developing competency for diagram comprehension
during this time.

An abundance of evidence exists on the cognitive benefits of different types of external

representations, including that the addition of relevant diagrams to text enhances
learning (Mayer, 1989, 2005). However, several theories of diagrammatic reasoning
caution that the benefit of diagrams is dependent on their relevance for the task at hand,

∗ Correspondence should be addressed to Julie L. Booth, 1301 W. Cecil B. Moore Avenue, Philadelphia PA 19122, USA
(e-mail: [email protected]).

DOI:10.1111/j.2044-8279.2011.02041.x
2 Julie L. Booth and Kenneth R. Koedinger

the context of the representation, and the skill of the user (for a review, see Acevedo
Nistal, Clarebout, Elen, Van Dooren, & Verschaffel, 2009). For example, representations
that are not appropriate or compatible with the task that the user is asked are unlikely
to be helpful (Meyer, 2000), as they may not support the necessary types of cognitive
processing required for the task (Greeno & Hall, 1997). Diagram utility may also depend
on the spatial grouping of information that will be used together (Larkin & Simon, 1987),
the degree of contiguity between the text and associated diagram (Mayer, 2005), and
a lack of complete redundancy between the text and diagram (Chandler & Sweller,
1991; Kalyuga, Chandler, & Sweller, 1998). The effects of these characteristics may be
more or less pronounced depending on the prior knowledge and spatial ability of the
user (Mayer & Gallini, 1990; Mayer, Steinhoff, Bower, & Mars, 1995). However, user
characteristics may influence effective use of even compatible, well-designed diagrams.
For example, domain experts can interpret diagrams within their domain more accurately
than novices (Ainsworth, 2006), as they can more easily identify critical components of
the diagram (Larkin & Simon, 1987) and connections among diagram components and
the represented situation (Narayanan & Hegarty, 1998). Domain-general diagrammatic
reasoning and spatial skills may be also crucial for drawing inferences from the spatial
layout of the diagram (Larkin & Simon, 1987), and familiarity with the general form and
components of the representation may also influence correct interpretation (Ainsworth,
2006). Diagrams that are aligned with these learner characteristics and are designed to
support the cognitive processes of those learners are more likely to yield deep com-
prehension of the represented information (Butcher, 2006; Davenport, Yaron, Klahr, &
Koedinger, 2008).
One context in which diagrams may be especially useful is in supporting students’
understanding of key concepts within a domain. A commonly proposed strategy for
improving learning involves creating a smooth transition from what students already
know, which is often more concrete, to the desired knowledge, which is often more
abstract. For example, Piaget’s theory of cognitive development suggests that it is
necessary to provide concrete experience from which young students can abstract
higher order concepts (Kamii, 1974); a number of empirical studies support this
hypothesis that transitioning from grounded representations to more abstract ones is
an effective instructional technique (Bransford, Brown, & Cocking, 1999; Goldstone &
Son, 2005; Koedinger & Anderson, 1998; Moreno & Mayer, 1999; Nathan, Kintsch, &
Young, 1992; Nathan & Koedinger, 2000; Romberg & de Lange, 2011; Schwartz & Black,
1996). For example, instructors might ensure that early experiences are concrete, in
order to promote connections between the facets of a representation and their real-
world counterparts, and then fade the concreteness so learners are able to use more
abstract representations and transfer their knowledge more easily to different situations
(Goldstone & Son, 2005).
Much of the relevant empirical work in mathematics focuses on the transition
from concrete physical representations, or manipulatives, to abstract, symbolic ones
in elementary school (e.g., Kennedy & Tipps, 1994; Sowell, 1989). However, bridging
instruction may also be important later in development, when students are transitioning
from arithmetic to algebraic thinking. Despite the common belief that word problems are
inherently more difficult than equations, solving simple algebra problems (e.g., those that
refer to the variable only once) in a textual format can actually be easier for high school
(typically aged 15–18) (Koedinger & Nathan, 2004) and college students (Koedinger,
Alibali, & Nathan, 2008) than solving them in an equivalent equation format. However,
for more complex problems (e.g., those that refer to the variable twice), story problems
 Development of diagram use 3

are harder to solve than matched problems in equation format (Koedinger et al., 2008).
These results suggest that instruction may be more effective if it builds understanding
and skill with abstract symbolic representations on top of existing competence with
more concrete textual representations.
Failure to translate story problems into usable internal representations (De Corte,
Verschaffel, & De Win, 1985; Kintsch & Greeno, 1985; Zawaiza & Gerber, 1993) or
to produce appropriate mathematical representations of the problem (Heffernan &
Koedinger, 1997; Koedinger & Nathan, 2004) can each preclude successful problem
solving. For such difficulties in representation, providing a diagram could facilitate
solution by effectively scaffolding the representation process. External representations
(e.g., diagrams, graphs, tables, etc.) are recommended tools for math instruction
( National Council of Teachers of Mathematics [NCTM], 2000), and empirical work from
the field of mathematics education provides evidence of performance-enhancing benefits
of diagrams – Hembree’s (1992) meta-analysis concluded across 16 studies that students
provide more correct answers to word problems when they contain accompanying
diagrams. However, recent studies have called into question the effectiveness of diagrams
for word problems. For example, De Bock, Verschaffel, Janssens, Van Dooren, and Claes
(2003) report that diagrams may actually be harmful for high school students learning
geometry. Further, a detrimental effect of diagrams was found for both strong and
weak fifth grade math students when solving arithmetic word problems (Berends & Van
Lieshout, 2009); in some situations, solving problems with even a well-designed diagram
may increase the cognitive load placed on students (Berends & Van Lieshout, 2009; Lee,
Ng, & Ng, 2009).
Diagrams are often used in math instruction in Singapore (Beckmann, 2004) and
Japan (Murata, 2008), two countries in which mathematics achievement is consistently
outstanding by world standards (National Center for Education Statistics, 2003). The style
of diagrams used in these countries, sometimes called tape diagrams (Murata, 2008), strip
diagrams (Beckmann, 2004), or bar models (Hoven & Garelick, 2007), uses strings of
objects and/or strips of different lengths to represent the magnitude of and relationships
between the quantities in the problem. Like algebraic equations (but unlike symbolic
arithmetic problems), these diagrams are not meant to help users carry out operations,
but to help them decide what operations to use and to understand why those operations
are conceptually sound (Beckmann, 2004). Tape diagrams are also found in textbooks
and some educational software (e.g., Carnegie Learning, 2009) in the United States, but
their use is less frequent and inconsistent (Murata, 2008). There is, however, evidence
that American students can use these diagrams to solve algebraic word problems that
would ordinarily be quite challenging for them (Koedinger & Terao, 2002). For example,
when given a tape diagram with a word problem, students in their study answered it
correctly 71% of the time. On comparable problems without diagrams, middle school
students were only 4% correct (Bednarz & Janvier, 1996) and college algebra students
were only 54% correct (Koedinger & Alibali, 1999). Thus, a potential motivation for
including a diagrammatic representation with a problem might be to help younger
students or those with lower ability to solve the problem, as they would be the ones less
likely to be able to represent the problems accurately on their own.

The present study

Experiment 1 had three purposes. The first was to replicate the verbal or textual
advantage with younger students (Koedinger and Nathan, 2004; Koedinger et al.,
4 Julie L. Booth and Kenneth R. Koedinger

2008) by presenting middle school students (typically aged 12–14) with difficult algebra
problems (e.g., start-unknown and systems of equations problems) in equation and story
formats. We also extend these findings by investigating developmental differences in the
textual advantage, as older students’ increased familiarity with equations may improve
their problem solution in that format.
The second purpose was to determine whether a diagrammatic advantage also exists
for students learning algebra. Previous research does not yield a conclusive prediction:
diagrams may prove to be beneficial (Hiebert & Carpenter, 1992; Koedinger & Terao,
2002), but could be futile or even harmful (Berends & Van Lieshout, 2009; De Bock et al.,
2003; Gravemeijer, 1994; Uttal, Liu, & DeLoache, 2006). Alternatively, diagrams may be
helpful for some students and not others (e.g., Ainsworth, 2006; Larkin & Simon, 1987).
We examine whether diagrams are useful in the context of algebraic story problems and
investigate developmental differences in their effectiveness.
The third purpose was to examine individual differences in the textual and dia-
grammatic advantages based on students’ mathematics ability level. The optimal type of
presentation may vary for students with different background knowledge (cf. Kalyuga
et al., 1998). As low-ability students have particular difficulty representing word
problems (Montague, Bos, & Doucette, 1991), diagrams that eliminate or obviate the
necessity of creating a problem representation could be particularly beneficial for them.
Alternatively, high-ability students (domain experts) may have greater success than low-
ability students (domain novices) at interpreting the diagrams, in which case high-ability
students may show greater benefit from diagrams (Kozma, 2003; Lowe, 2003).
In Experiment 1, we presented sixth, seventh, and eighth grade students with algebra
problems in each of three presentations types: equation, story, and story + diagram. The
main predictions were that the textual advantage would be replicated (story problems
would be easier than equations) and that a diagrammatic advantage would be evident
(adding diagrams to story problems would facilitate solution); individual differences in
grade and math ability were expected for each type of advantage.

EXPERIMENT 1

Method
Participants
Participants were 373 students [121 sixth grade (age 12), 117 seventh grade (age 13), 135
eighth grade (age 14)] drawn from an ethnically diverse middle school in the American
Midwest in which 25% of attending students were African American, 7% Asian, 62%
Caucasian, and 5% Latino; 40% of attending students were from low-income families.
All classes used Connected Mathematics, a reform-based curriculum, which begins
developing algebraic skills in sixth grade, with increasing attention given to algebra
in seventh and eighth grade (Lappan, Fey, Fitzgerald, Friel, & Phillips, 2002). Students
are exposed to both symbolic and spatial representations throughout the curriculum,
though it does not introduce the type of diagrams tested in this study.

Measures and procedure

The three items that were the focus of this study were embedded in a written
comprehensive algebra assessment that included a variety of other items used for
 Development of diagram use 5

studies on student understanding of equality and variability, which have been published
elsewhere (Alibali, Knuth, Hattikudur, McNeil, & Stephens, 2007; Knuth, Alibali, McNeil,
Weinberg, & Stephens, 2005; Knuth, Stephens, McNeil, & Alibali, 2006). The assessment
was administered to students early in their fall semester by the classroom teacher
during normal class time; individual students were randomly assigned to receive one
of three alternate forms. Each form contained the same three problems, but varied the
presentation format: equation, story, or story + diagram. Each form contained one
problem for each presentation format, and each problem appeared once in each of
the three presentation formats across the three forms. One item was a single-reference
problem and one a double-reference problem, similar to those used by Koedinger et al.
(2008); the remaining problem used two variables. Figure 1 shows the three problems
in each format.
For approximately one-third of the students (n = 128: 43 sixth, 41 seventh, 44 eighth
grade), parents consented to collection of national percentile rankings for the math
section of the students’ most recent TerraNova tests. For sixth and seventh graders,
scores were taken from the current year; the school district does not administer the
TerraNova test in eighth grade, thus, eighth graders’ scores were from the previous year.
Percentile rankings on the Terra Nova test correlated positively with student accuracy
on the study problems, R(128) = .46, p < .001.

Results1
Textual advantage
A 3 (grade: 6, 7, 8) × 2 (presentation: equation, story) ANOVA2 on correct responses
yielded a main effect of presentation format F(1,371) = 31.95, p < .01, !p 2 = .08; more
correct responses were given when problems were in story form (24%) than in equation
form (9%). There was no main effect of grade or grade by presentation interaction.

Diagrammatic advantage
A 3 (grade: 6, 7, 8) × 2 (presentation: story, story + diagram) ANOVA on correct
answers yielded a main effect of grade F(2,370) = 3.17, p < .05, !p 2 = .02. Least
Significant Differences (LSD) post hoc analyses revealed that eighth graders answered
more problems correctly (30%) than sixth graders (21%). No main effect of presentation
format or interaction was found.

Differences by problem type

Because Koedinger et al. (2008) revealed that the advantages of different presentation
formats could vary based on problem difficulty, we examined differences in accuracy
by problem. Table 1 shows accuracy data on each problem type for the full sample

1 Analyses were conducted both across all three presentation formats and separately for the textual and diagrammatic
advantages. The pattern of results from both analyses were the same, thus, as the purpose of the paper is to test two
particular comparisons: story versus equation (to examine the textual advantage) and story versus story + diagram (to
examine the diagrammatic advantage); separate ANOVAs are presented for each comparison.
2 Data were analysed using both parametric and non-parametric tests (when available), which yielded similar results. Due to
the large sample size in the study and the fact that appropriate non-parametric tests did not always exist (e.g., there is no
non-parametric version of two-factor, repeated measures tests), we only report results from the parametric tests.
 6
Equation Story Story + Diagram
Solve the equation below to Mom won some money in a lottery. She kept Mom won some money in a lottery, She kept $45 for herself and gave each
find the value of N: $45 for herself and gage each of her 3 sons of her 3 sons an equal portion of the rest. If each son got $20.50, how
an equal portion of the rest. If each son got much did Mom win?
$20.50, how much did Mom win? (You can use the picture below to help you solve the problem.)
(N - 45) ÷ 3 = 20.50
<---$45 Mom kept---XX-------- equal parts each son got ------->

<--$20.50,-->
<----------------- how much did Mom win?------------------------>
Find values of S and C that John bought 3 t-shirts and 2 baseball caps for John bought 3 t-shirts and 2 baseball caps for $58. Sue bought 2 t-shirts
make these equations true: $58. Sue bought 2 t-shirts and 3 baseball and 3 baseball caps for $52. What is the cost of one shirt? What is the cost
Julie L. Booth and Kenneth R. Koedinger

caps for $52. What is the cost of one shirt? of one baseball cap?
3S + 2C = 58 What is the cost of one baseball cap? (You can use the picture below to help you solve the problem.)
2S + 3C = 52
$58

$52

Solve the equation below to Molly bought a coat on sale. It was 1/5 off Molly bought a coat on sale. It was 1/5 off the original price. She paid $30.
find the value of N: the original price. She paid $30. What was What was the original price of the coat?
the original price of the coat? (You can use the picture below to help you solve the problem.)
1
N- / 5
* N = 30 <------------------------$30 paid----------------------------><-- 1/5 off-->

<-------------------What was the original price?------------------------->

Figure 1. Problems displayed in each presentation format.

 Development of diagram use 7

Table 1. Experiment 1: Accuracy by condition for full sample and each grade level for each problem

Problem Overall Equation Story Story + diagram

Lottery
Full sample 49% 22% 64% 61%
Grade 6 45% 20% 63% 51%
Grade 7 44% 18% 58% 54%
Grade 8 57% 26% 70% 76%
T-shirt
Full sample 4% 0% 5% 7%
Grade 6 2% 0% 3% 2%
Grade 7 5% 0% 8% 8%
Grade 8 5% 0% 4% 11%
Sale
Full sample 9% 5% 4% 17%
Grade 6 2% 0% 2% 2%
Grade 7 16% 8% 8% 32%
Grade 8 8% 7% 2% 16%

and separately by grade for each presentation format. First, to determine problem-type
differences in difficulty level, we conducted a 3 (grade: 6, 7, 8) × 3 (problem: lottery,
t-shirt, sale) ANOVA with repeated measures on problem. This analysis revealed a main
effect of problem, such that the lottery problem (single reference: a single variable is
included only once in the equation) was easier than the sale (double reference: a single
variable is included twice in the equation) and t-shirt problems (two variables: two
separate variables are referred to in the problem), F(1,370) = 211.55, p < .001, !p 2 = .36.
The main effect of grade [F(2,370) = 3.99, p < .05, !p 2 = .02] and problem by grade
interaction (F(2,370) = 5.49, p < .01, !p 2 = .03) were also significant, but follow-up
repeated measures ANOVAs conducted separately by grade confirmed that the pattern
of difficulty was the same for each grade level (sixth grade, F(1,120 = 84.72, p < .01,
!p 2 = .41; seventh grade, F(1,116) = 30.54, p < .001, !p 2 = .21; eighth grade,
F(1,134) = 114.59, p < .001, !p 2 = .46).
To determine whether the textual advantage existed for each problem, we conducted
a 3 (grade: 6, 7, 8) × 2 (presentation: equation, story) ANOVA for each problem. For
both the lottery problem and the t-shirt problem, there was a significant main effect of
presentation, with accuracy higher in the story format than the equation format (lottery,
F(1,244) = 55.34, p < .001, !p 2 = .19; t-shirt, F(1,242) = 6.35, p < .01, !p 2 = .03).
However, the main effect of presentation for the sale problem was not significant
[F(1,242) = 0.06, ns, !p 2 = .00], and no other main effects or interactions were found.
These results replicate Koedinger et al.’s (2008) finding that stories are easier than
equations for single-reference problems (i.e., lottery), but that stories and equations
were equally challenging for double-reference problems (i.e., sale).
A parallel set of 3 (grade: 6, 7, 8) × 2 (presentation: story, story + diagram) ANOVAs
were conducted for each problem to test for the diagrammatic advantage. Main effects
of grade were found for both the lottery (F(2,242) = 3.36, p < .05, !p 2 = .03) and sale
problems (F(2,244) = 7.70, p < .001, !p 2 = .06). Results for the sale problem (double
reference) also yielded a main effect of presentation (F(1,244) = 12.19, p < .001,
!p 2 = .05), with accuracy higher on the story + diagram format than the story
format. The presentation by grade interaction was also significant (F(1,244) = 3.65,
8 Julie L. Booth and Kenneth R. Koedinger

Table 2. Exp 1: Distribution and analysis of high- and low-ability students by test form and grade

Sample Form A Form B Form C Analysis

High ability
Grade 6 8 5 4 X 2 (4, N = 64) = 5.17, ns
Grade 7 4 10 9
Grade 8 5 11 8
Low ability
Grade 6 10 10 6 X 2 (4, N = 64) = 7.39, ns
Grade 7 4 5 9
Grade 8 5 3 12

p < .05, !p 2 = .03). Follow-up t tests conducted separately for each grade level on the
sale problem indicated that story + diagram and story formats were equally challenging
for sixth graders (t(80) = 0.00, ns, d = .00), but that accuracy was higher for the story +
diagram format than the story format for seventh (t(76) = 2.78, p < .01, d = .65) and
eighth graders (t(88) = 2.33, p < .05, d = .50). No analyses involving presentation format
were significant for the lottery or t-shirt problems.

Individual differences in the textual and diagrammatic advantages

For the subset of students for whom math achievement scores were obtained, we
computed the median national percentile ranking and grouped students by whether
they were of high (greater than or equal to the median) or low (below the median)
mathematical ability. Approximately, half of the students in each grade fell into the
high-ability group. Table 2 shows the distribution of students in the high- and low-ability
groups in each grade to test form; chi-squared tests yielded no significant differences in
distribution.

Textual advantage
A 3 (grade: 6, 7, 8) × 2 (presentation: equation, story) × 2 (math ability: high, low)
ANOVA on correct answers yielded a main effect of presentation format, F(1,122) =
7.09, p < .01, !p 2 = .06; as in the parallel full sample analysis, students gave correct
answers more frequently for story problems (29%) than equations (15%). There was also
a main effect of math ability, F(1,122) = 23.20, p < .01, !p 2 = .16; students with high
math ability solved more problems correctly (33%) than peers with low math ability
(11% correct). There were no other significant main effects or interactions. Thus, there
were no individual differences in the textual advantage; story problems were easier than
equations for all students, and high math ability increased the likelihood of solving either
type of problem correctly.

Diagrammatic advantage
A parallel 3 (grade: 6, 7, 8) × 2 (presentation: story, story + diagram) × 2 (math ability:
high, low) ANOVA assessed individual differences in the diagrammatic advantage. Similar
to the full sample analysis, a main effect of grade was found (F(2,122) = 4.96, p < .01,
!p 2 = .08), as was a trend towards an interaction between the grade and presentation
format variables, F(2,122) = 2.31, p = .10, !p 2 = .04. Proportion of correct answers
 Development of diagram use 9

60%
50%
Percent Correct 40%
30%
20%
10%
0%
6th grade 7th grade 8th grade 6th grade 7th grade 8th grade
Low-Ability High-Ability

Story Story+Diagram

Figure 2. Experiment 1 performance on story and story + diagram problems by grade for low- and
high-ability students.

increased with grade, and sixth graders tended to answer more story problems correctly
than story + diagram problems (25% vs. 17%, respectively), while the opposite was
true for seventh and eighth graders (43% and 46% correct on story + diagram problems
vs. 39% and 23% on story problems). This suggests that some improvement in diagram
use is happening as students mature and/or experience more math instruction. Not
surprisingly, there was also a main effect of math ability F(1,122) = 5.47, p < .05,
!p 2 = .04. High-ability students answered more problems correctly (39%) than low-
ability peers (26%).
The interaction among the three variables was also significant, F(2,122) = 3.42,
p < .05, !p 2 = .05 (Figure 2). Separate 3 (grade) × 2 (presentation) repeated measures
ANOVAs were then conducted for each ability group. The analysis for high-ability
students yielded only a main effect of grade, F(2,61) = 3.43, p < .05, !p 2 = .10. In
contrast, the analysis for low-ability students yielded trends towards main effects of
presentation (20% [story] vs. 33% [story + diagram]; F(1,61) = 3.14, p < .10, !p 2 = .05)
and grade (F(2,61) = 2.45, p < .10, !p 2 = .07), and a significant presentation by grade
interaction, F(2,61) = 8.17, p < .001, !p 2 = .21. Paired-sample t tests comparing the
story and story + diagram formats were then conducted separately for the low-ability
students in each grade. Low-ability sixth graders scored higher on story problems (27%
correct) than story + diagram problems (4% correct; t(25) = 2.29, p < .05, d = .53).
The pattern was opposite for low-ability eighth graders (10% (story) vs. 55% [story +
diagram]; t(19) = 2.93, p < .01, d = .98); no difference was found for low-ability seventh
graders (22% (story) vs. 33% (story + diagram); t(17) = 1.37, ns, d = .33).

Discussion
Results from Experiment 1 replicated the textual advantage in a younger student
population: algebra story problems are easier for middle school students to solve than
matched equations. This same pattern holds for both single-reference and two-variable
problems. Previous findings suggest that for college students, the effect is reversed:
double-reference problems are easier in equation format than story format (Koedinger
et al., 2008). This effect was not found directly for middle school students in the present
study, likely because the students are not yet familiar with complex equations; however,
10 Julie L. Booth and Kenneth R. Koedinger

it is important to note that for double-reference problems, stories and equations are
equally challenging.
This experiment also explored the role of diagrams in algebraic problem solving.
Despite a number of reasons why diagrams should be beneficial, this study found no
general advantage for including diagrams with story problems, except for with the
more difficult double-reference problem. However, students’ ability level and age appear
to mediate potential benefits of diagrams. Sixth grade students tended to solve more
problems correctly in the story format but seventh and eighth graders solved more
problems correctly when a diagram accompanied the story. Low-ability sixth graders
appeared to be particularly confused by the diagram, but low-ability eighth graders
benefited from its presence.
Experiment 1 provides initial evidence of a diagrammatic advantage for some
students, but suggests that students must develop some diagram-relevant capability
before any such benefits appear. In Experiment 2, we use a longitudinal design to
examine whether and how the sixth graders from Experiment 1, who did not show a
diagrammatic advantage, come to benefit when they reach seventh or eighth grade. We
investigate the interaction between grade level and presentation format (story vs. story +
diagram), and also focus, in particular, on how the performance of low-ability sixth
graders changes over time.
An additional goal of Experiment 2 was to evaluate the types of errors students
make to begin to explore the nature of any observed benefit. One possible mechanism
is that by eliminating the need for students to represent the problem, diagrams may
make the problems seem easier, and students may be more likely to attempt to solve
them. Alternatively, the diagram may provide necessary supports that afford students
a better conceptual framing of the problem. For example, the diagram may make the
relations between problem components explicit, reducing students’ need to discern
those relations from the words (Larkin & Simon, 1987). The distribution of elements in
the diagram also provides visual cues about the appropriate size of the answer (Nunes,
Schliemann, & Carraher, 1993; Rittle-Johnson & Koedinger, 2005), and makes it readily
apparent that certain tempting solution paths are not valid (Larkin & Simon, 1987) (See
Figure 3), which should reduce the conceptual errors students make in solution attempts.
We explore age- and ability-related differences in no response and conceptual errors in
Experiment 2.

EXPERIMENT 2

Method
Participants
Participating in this study were 84 students who had participated in Experiment 1 as
sixth graders, including 23 of the 26 low-ability sixth graders and 14 of the 17 high-ability
sixth graders.

Procedure
The procedure was identical to that in Experiment 1. In the Fall of their seventh and
eighth grade school years, students were given the same form of the test they had taken
in sixth grade. Thus, they completed the same problems with the same numbers and in
the same format all three years.
 Development of diagram use 11

Conceptual error made on story problem Diagram supports correct coceptual strategy

9. Mom won some money in a lottery. She kept $45 for herself and 9. Mom won some money in a lottery. She kept $45 for herself and
gave each of her 3 sons an equal part of the rest. If each son got gave each of her 3 sons an equal parts of the rest. If each son got
$20.50, how much did Mom win? $20.50, how much did Mom win?

You can use the picture below to help you solve this problem

$45 Mom kept Equal parts each son got

$20.50
How much Mom won

10. Mally bought a coat on sale. It was 1/5 off the original price. She 10. Molly bought a coat on sale. It was 1/5 off the original price. She
paid $30. What was the original price of the coat? paid $30. What was the original price of the coat?

You can use the picture below to help you solve this problem.

$30 paid 1/5 off

Original Price

Figure 3. Examples of conceptual errors made by middle school students on story problems (left) and
conceptually correct strategies guided by the presence of a diagram (right).

Results
Textual advantage
A 3 (grade: 6, 7, 8) × 2 (presentation: equation, story) repeated measures ANOVA on
correct answers yielded a main effect of presentation, F(1,83) = 4.61, p < .05, !p 2 = .05.
More correct responses were given for problems in story form (22% correct) than in
equation form (11%). No main effect of grade or interaction was found.

Diagrammatic advantage
A 3 (grade: 6, 7, 8) × 2 (presentation: story, story + diagram) repeated measures ANOVA
on correct answers yielded a main effect of grade F(2,82) = 5.75, p < .05, !p 2 = .07;
probability of correct responses increased with grade. There was no main effect of
presentation, but the grade × presentation interaction was significant, F(2,82) = 5.75,
p < .05, !p 2 = .07 (Figure 4). Follow-up repeated measures ANOVAs conducted sepa-
rately by presentation revealed no grade differences within the story presentation, but
performance on story + diagram problems improved with grade F(1,83) = 9.85, p < .01,
!p 2 = .11.

Development of a diagrammatic advantage in low-ability students

To examine how low-ability sixth graders’ performance on the two types of problems
changed over time, a 3 (grade: 6, 7, 8) × 2 (presentation: story, story + diagram)
12 Julie L. Booth and Kenneth R. Koedinger

40%
35%
30%

Percent Correct
25%
20%
15%
10%
5%
0%
Story Story + Diagram

6th grade 7th grade 8th grade

Figure 4. Experiment 2 performance on story and story + diagram problems for sixth, seventh, and
eighth graders.

repeated measures ANOVA for the 23 low-ability students yielded a main effect of
grade and an interaction between the grade level and presentation format variables,
both F(1,22) = 11.73, p < .01, !p 2 = .35 (Figure 5). Probability of correct responses
increased with grade, and separate follow-up repeated measures ANOVAs on grade for
each presentation format revealed that low students improved with age on diagram
problems (F(1,22) = 11.73, p < .01, !p 2 = .35) but not story problems (F(1,22) = 2.10,
ns, !p 2 = .09).

Dissecting the diagrammatic advantage: Differences in error patterns by grade

and math ability
Following Koedinger and Nathan (2004), we conducted a qualitative analysis of the types
of errors students made on story and story + diagram problems. Cases in which students
did not attempt a problem at all were coded as no response errors. Errors that indicated
a failure to correctly translate the problem situation into an appropriate solution path
(e.g., adding two numbers that should be multiplied) were coded as conceptual errors;

70%
60%
Percent Correct

50%
40%
30%
20%
10%
0%
6th grade 7th grade 8th grade

Story Story+Diagram

Figure 5. Experiment 2 performance on story and story + diagram problems by grade for low-ability
students.
 Development of diagram use 13

Table 3. Exp 2: Frequency of error types by grade on story and story + diagram problems for all
students, low-ability students, and high-ability students

Sixth grade Eighth grade

Sample Story Diagram Story Diagram

All students
No response errors 41% 46% 21% 19%
Conceptual errorsa 58% 60% 63% 47%
Low ability
No response Errors 39% 65% 26% 22%
Conceptual errorsa 64% 75% 65% 44%
High ability
No response errors 50% 29% 21% 29%
Conceptual errorsa 43% 60% 64% 40%

Note. a Percentage computed out of attempted problems (e.g., for students in sixth grade, 58% of
attempted story problems, or 34% of all story problems, contained a conceptual error).

this is to be distinguished from errors in computation that occurred when a student tried
to solve a problem using a correct solution path (e.g., correctly decided to multiply the
two numbers, but made an error in carrying out the multiplication).
Frequency of no response errors is computed out of all trials, correct or incorrect. In
order to isolate the effects of each possible mechanism, frequency of conceptual errors
is then computed out of all problems that were attempted (i.e., eliminating no response
errors). We report error patterns for participants when they were in sixth grade (and
least likely to benefit from diagrams) versus eighth grade (most likely to benefit), and
then describe error patterns of students of varying math ability.
In eighth grade, students are equally likely to attempt diagram and story problems
(19% vs. 21%, respectively), but are less likely to make conceptual errors on diagram
problems than on story problems (47% vs. 63%; see Table 3, top right). In contrast, for
sixth graders, diagrams do not reduce conceptual errors (60% vs. 58%) and, if anything,
may increase no response errors (46% vs. 41%). When we separate the low- and high-
ability sixth graders (Table 3, lower left), we find that the low-ability students are much
more avoidant of diagram problems than story problems in sixth grade (65% vs. 39% no
response errors). The high-ability students attempt diagram problems in sixth grade, but
are not able to use them, as both low- and high-ability students do in eighth grade, to
reduce conceptual errors.

GENERAL DISCUSSION
The longitudinal findings from Experiment 2 largely confirmed and strengthened the
findings from Experiment 1. First, the textual advantage was replicated: story problems
were easier than equations in both experiments, across all grade levels. The fact that there
were no developmental or individual differences in the textual advantage is surprising,
given that students experience increasingly more exposure to symbolic equations as
they progress through the Connected Mathematics curriculum. However, the equations
used in this study were, by design, difficult; perhaps students are developing symbolic
capacity, but it is not transferring to more difficult equations. A textual advantage had
14 Julie L. Booth and Kenneth R. Koedinger

previously been demonstrated with high school (Koedinger & Nathan, 2004) and college
students (Koedinger et al., 2008) on beginning algebra problems, but not for early
elementary school students3 (Cummins, Kintsch, Reusser, & Weimer, 1988) on arithmetic
problems. Thus, development of the textual advantage appears to occur between second
grade and ninth grade, perhaps as reading comprehension improves. Results from the
present study suggest that the textual advantage is in place by sixth grade, and that story
problems may be useful as a transition for students learning to solve abstract, symbolic
equations.
Evidence for a diagrammatic advantage was also found, but only for older and
high-ability middle school students. Results from error analysis suggest that diagrams
improve these students’ performance by increasing the likelihood that they will generate
a conceptually correct understanding of the problem situation. Overall, diagrams are
beneficial additions to story problems for more accomplished students, suggesting that
their use as a transition to abstract, symbolic representations may be constructive for
these students.
Unfortunately, results from both experiments also suggest that younger middle school
students, and especially those with low math ability, do not benefit from the diagrams.
Our error analysis suggests that the main barrier to successful diagram use in sixth grade
was the inability to extract a correct conceptual understanding of the problem from
the diagram. Whether this is due to misinterpretation of the diagram itself or a failure to
accurately map the story problem to the diagram, the diagrams did not reduce conceptual
errors.
Low-ability sixth graders also experience an additional barrier to success, as they
were even less likely to attempt the diagram problem compared with the story problem.
Unlike their more accomplished peers who were likely to find the diagram a helpful tool,
for low students, diagrams made the problems less approachable than the story problems
alone. This is perhaps not surprising, as low-ability students generally perceive problems
as more difficult than high- or average-ability students and are thus more likely to shut
down and not attempt the problems (e.g., Ericsson & Simon, 1980). In the present study,
low-ability sixth graders may have given up on problems because they experienced or
perceived an increase in cognitive load when examining the problem, perhaps due to an
(incorrect) assumption that they would need to attend to both the diagram and the story
simultaneously in order to solve the problem (Chandler & Sweller, 1991). Because they
had limited diagram comprehension skills, their confusion over the diagram, it appears,
led them to miss the fact that they could simply ignore it and work from the story alone.
Why did young, low-ability students, who potentially have the most to gain (Murata,
2004), fail to benefit from the diagrams, while older students found success? We join with
Ainsworth (2006) and Larkin and Simon (1987) in asserting that a complex interaction of
student abilities and task characteristics drives development of successful diagram use.
In the case of the present study, it seems likely that sixth graders have not sufficiently
developed the necessary component skills or capacities to reap the benefits of diagrams.
A number of changes – both developmental and instructional – occur between
sixth and eighth grade that could influence diagram comprehension. First, students
develop greater cognitive capacity during adolescence, including increases in working

3Though a pure textual advantage between word problems and numerical problems has not been found for young students,
beginning elementary students do fare better on word problems than on symbolic ones such as ‘a + _ = b’(Carpenter &
Moser, 1984; De Corte & Verschaffel, 1981).
 Development of diagram use 15

memory (Luna, Garver, Urban, Lazar, & Sweeney, 2004), which should reduce any
deleterious effects of cognitive load, and executive control (Luciana, Conklin, Hooper,
& Yarger, 2005). With increased capacity, students should be better able to coordinate
the information in the diagram and problem statement and make decisions about where
to focus; in fact, a recent study shows that success on diagram problems was indeed
associated with increased working memory capacity (Lee et al., 2009). Formal reasoning
skills, such as the ability to coordinate information gained from multiple sources or to
reason logically about possible outcomes, also show considerable development across
adolescence, enhancing students’ ability to solve problems in a variety of domains
(Inhelder & Piaget, 1958; Marini & Case, 1994).
Instruction helps students build a foundation in pre-algebraic conceptual understand-
ing during their middle school years. This increased domain expertise likely allows older
students to make better inferences and connections between the different elements of
the problem (Larkin & Simon, 1987; Narayanan & Hegarty, 1998); lack of knowledge
about mathematical equality, fractions, or the relationship between number of items
and cost could inhibit sixth graders’ performance. Further, as previously mentioned,
the Connected Math curriculum aims to increase students’ representational fluency by
helping students see connections between text and representations. Even though the
style of diagrams presented in the current study are not included in the curriculum,
increased exposure to instruction that illustrates the connections between different
types of diagrams and text may account for much of the improvement seen in this study.
It follows that with a sample that does not use Connected Math, even worse performance
on diagram problems compared with story problems and less improvement over time in
diagram problems may be found.
In contrast to younger students, older, more experienced students appear to have
developed a number of skills necessary to take advantage of the beneficial diagrammatic
features. It may be surprising to some readers that older middle school students (seventh
and eighth graders in this sample) can benefit from diagrammatic representations without
any substantial instruction on how to use diagrams in problem solving. On other hand,
advocates for use of diagrams in mathematics education (NCTM, 2000) may be surprised
that younger students (sixth graders) do not easily and naturally make use of diagrams.
One limitation of this study is that we cannot determine much about why sixth graders
are failing, in particular, whether there is a developmental barrier of some kind or
whether there is particular prerequisite knowledge for diagram processing that they
lack. Our results indicate, however, that further research is warranted to understand
how this promising practice of Asian curricula might be better introduced in or before
sixth grade to support pre-algebraic reasoning or whether it should be delayed to later
grades.
Interestingly, there is evidence of similar trends for adolescents learning to compre-
hend graphs and maps, in which younger adolescents had more difficulty extracting
implicit or conceptual information from the representations; a combination of increases
in domain-specific content knowledge and age-related development of ‘graphicacy’ is
purported to account for these changes (Postigo & Pozo, 2004). Is there a developmental
threshold that should be reached before which any of these representations should
be introduced with caution, or perhaps not used at all? There is a period of time in
elementary school in which students must learn to read before they can be expected
to gain new information from text (Chall, 1979). Similarly, perhaps there is a period
of time before which students cannot be expected to glean new information from
diagrammatic and graphical representations; they must first learn to comprehend and
16 Julie L. Booth and Kenneth R. Koedinger

use these representations. Further research is needed to tease apart the different aspects
of development and knowledge acquisition that are occurring during the middle school
years and determine which component skills or capacities are most crucial for successful
use of diagrams and other visual representations.

Improving students’ use of diagrams as problem solving aids

If a representation (concrete, abstract, diagrammatic) is ever going to be helpful as a
scaffold for learning and transfer, it needs to at least be easier. Can instruction improve
students’ ability to use the diagrams when solving algebraic story problems? Research
in both mathematics education and cognitive development suggests that students need
effective instruction in order to use any type of external representations correctly (Fueyo
& Bushell, 1998;Sowell, 1989; Uttal, Scudder, & DeLoache, 1997). Brief instruction on
the nature of the particular visual representation may not be sufficient (Rittle-Johnson
& Koedinger, 2001), but more effective representation-specific instruction could be
time consuming, and may not transfer to other novel representations that students
encounter. However, such instruction on different types of diagrams in the Connected
Math curriculum may partially account for the growth found in the present study.
Having students construct diagrams to represent the story problems themselves may
improve their use of given representations to solve problems. Middle school students
are able to use representations they created themselves to successfully solve algebraic
story problems (Koedinger & Terao, 2002), and constructing diagrams from scratch or
completing partial diagrams has been shown to increase student learning in a variety of
domains (e.g., Lewis & Mayer, 1987; see Van Meter & Garner, 2005 for a review). The
construction process likely helps students learn to coordinate and integrate text with
visual representations (Ainsworth, 2006; Easterday, Aleven, & Scheines, 2007), which
may in itself improve students’ use of diagrams to solve algebra story problems and their
subsequent ability to correctly represent the problem in an equation format. These types
of activities also have the potential to increase students’ meta-representational fluency
(diSessa & Sherin, 2000) such that students may come to build a broader understanding
of representations in general, to understand the advantages and disadvantages of
different representations, and to appreciate the usefulness and relevance of certain types
of representations for problem solving (di Sessa, Hammer, Sherin, & Kolpakowski, 1991).
Further research is needed to test whether such instruction can impact student learning
to use diagrams for algebraic word problems.
Regardless of the type of instruction necessary to make the diagrams accessible to
younger and low-ability students, such efforts may be well warranted. Crucial concepts
can be learned via grounded, concrete experiences that help to scaffold students to
more abstract understanding (Bransford et al., 1999; Goldstone & Son, 2005; Koedinger
& Alibali, 1999; Moreno & Mayer, 1999; Schwartz & Black, 1996). Struggling students
are arguably the most in need of this type of scaffolding, and may well be the reason
that well-meaning teachers include diagrams as aids to understanding. However, in
order to derive the intended benefit from the diagrams, these students must have gained
enough familiarity and competence with diagrammatic conventions that adding diagrams
enhances performance. Teachers (and even instructional designers) should implement
simple formative assessments (e.g., a few quiz problems with and without diagrams)
to determine when a particular kind of diagram may or may not be a scaffold for
mathematical understanding for a particular student level. Instructors and designers
should exercise caution when using diagrams in attempt to ease the transition to more
 Development of diagram use 17

abstract, symbolic representations; those diagrams may confuse or hinder the progress
of the very students who are in most need of assistance.

Acknowledgements
Funding for this research was provided by the Interagency Education Research Initiative
(IERI), NSF Award Number REC-0115635, “Collaborative Research: Understanding and
Cultivating the Transition from Arithmetic to Algebraic Reasoning”. Portions of this work
were presented at the biennial meeting of the Society for Research in Child Development in
Boston, MA, USA.

References
Acevedo Nistal, A., Clarebout, G., Elen, J., Van Dooren, W., & Verschaffel, L. (2009). Conceptualis-
ing, investigating and stimulating representational flexibility in mathematical problem solving
and learning: A critical review. ZDM-The International Journal on Mathematics Education,
41, 627–636.
Ainsworth, S. (2006). DeFT: A conceptual framework for considering learning with multiple
representations. Learning and Instruction, 16(3), 183–198.
Alibali, M. W., Knuth, E. J., Hattikudur, S., McNeil, N. M., & Stephens, A. C. (2007). A longitudinal
examination of middle school students’ understanding of the equal sign and equivalent
equations. Mathematical Thinking and Learning, 9, 221–246.
Beckmann, S. (2004). Solving algebra and other story problems with simple diagrams: A method
demonstrated in grade 4–6 texts used in Singapore. The Mathematics Educator, 14, 42–46.
Bednarz, N., & Janvier, B. (1996). Emergence and development of algebra as a problem-solving tool.
In N. Bednarz, C. Kieran, & L. Lee (Eds.), Approaches to algebra (pp. 115–145). Dordrecht,
The Netherlands: Kluwer.
Berends, I. E., & Van Lieshout, E. (2009). The effect of illustrations in arithmetic problem-solving:
Effects of increasing cognitive load. Learning and Instruction, 19, 345–353.
Bransford, J., Brown, A., & Cocking, R. (Eds.). (1999). How people learn: Brain, mind, experience,
and school. Washington, DC: National Academy Press.
Butcher, K. R. (2006). Learning from text with diagrams: Promoting mental model development
and inference generation. Journal of Educational Psychology, 98(1), 182–197.
Carnegie Learning (2009). Bridge to algebra. Retrieved from www.carnegielearning.com/AR2009
Carpenter, T. P., & Moser, J. M. (1984). The acquisition of addition and subtraction concepts in
grades one through three. Journal for Research in Mathematics Education, 15, 179–202.
Chall, J. S. (1979). The great debate: Ten years later, with a modest proposal for reading
stages. In L. B Resnick & P. A. Weaver (Eds.), Theory and practice of early reading
(Vol. 1, pp. 29–55). Hillsdale, NJ: Lawrence Erlbaum Associates.
Chandler, P., & Sweller, J. (1991). Cognitive load theory and the format of instruction. Cognition
and Instruction, 8, 293–332.
Cummins, D. D., Kintsch, W., Reusser, K., & Weimer, R. (1988). The role of understanding in
solving algebra word problems. Cognitive Psychology, 20, 405–438.
Davenport, J., Yaron, D., Klahr, D., & Koedinger, K. (2008).When do diagrams enhance learning? A
framework for designing relevant representations. Proceedings of the 8th International Con-
ference on the Learning Sciences (pp. 191–198). Mawah, NJ: Lawrence Erlbaum Associates,
Inc.
De Bock, D., Verschaffel, L., Janssens, D., Van Dooren, W., & Claes, K. (2003). Do realistic contexts
and graphic representations always have a beneficial impact on students’ performance?
Negative evidence from a study on modelling non-linear geometry problems. Learning and
Instruction, 13, 441–463.
18 Julie L. Booth and Kenneth R. Koedinger

De Corte, E., & Verschaffel, L. (1981). Children’s solution processing in elementary arith-
metic problems: Analysis and improvement. Journal of Educational Psychology, 73, 765–
779.
De Corte, E., Verschaffel, L., & De Win, L. (1985). Influence of rewording verbal problems on
children’s problem representations and solutions. Journal of Educational Psychology, 77,
460–470.
diSessa, A. A., Hammer, D., Sherin, B., & Kolpakowski, T. (1991). Inventing graphing: meta-
representational expertise in children. Journal of Mathematical Behavior, 10, 117–160.
diSessa, A., & Sherin, B. (2000). Meta-representation: An introduction. Journal of Mathematical
Behavior, 19, 385–398
Easterday, M., Aleven, V., & Scheines, R. (2007). ‘Tis better to construct than to receive? The
effects of diagramming tools on causal reasoning. In R. Luckin, K. Koedinger, & J. Greer (Eds.),
Proceedings of the 13th International Conference on Artificial Intelligence in Education
(pp. 93–100). Amsterdam, The Netherlands: IOS Press.
Ericsson, K. A., & Simon, H. A. (1980). Verbal reports as data. Psychological Review, 87, 215–251.
Fueyo, V., & Bushell, D. (1998). Using number-line procedures and peer tutoring to improve
the mathematics computation of low-performing first graders. Journal of Applied Behavioral
Analysis, 31, 417–430.
Goldstone, R. L., & Son, J. Y. (2005). The transfer of scientific principles using concrete and
idealized simulations. The Journal of the Learning Sciences, 14, 69–110.
Gravemeijer, K. (1994). Educational development and developmental research in mathematics
education. Journal for Research in Mathematics Education, 25, 443–471.
Greeno, J. G., & Hall, R. P. (1997). Practicing representation: Learning with and about representa-
tional forms. Phi Delta Kappan, 78, 361–367.
Heffernan, N. T., & Koedinger, K. R. (1997). The composition effect in symbolizing: The role
of symbol production vs. text comprehension. In Proceedings of the Nineteenth Annual
Conference of the Cognitive Science Society (pp. 307–312). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Hembree, R. (1992). Experiments and relational studies in problem solving: A meta-analysis.
Journal for Research in Mathematics Education, 23, 242–273.
Hiebert, J., & Carpenter, T. P. (1992). Learning and teaching with understanding. In D. A. Grouws
(Ed.), Handbook of research on mathematics teaching and learning (pp. 65–97). New York:
Macmillan.
Hoven, J., & Garelick, B. (2007). Singapore math: Simple or complex? Educational Leadership,
65, 28–31.
Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from childhood to adolescence.
New York, NY: Basic Books.
Kalyuga, S., Chandler, P., & Sweller, J. (1998). Levels of expertise and instructional design. Human
Factors, 40, 1–17.
Kamii, C. (1974). Pedagogical principles derived from Piaget’s theory: Relevance for educational
practice. In M. Schwebel & J. Raph (Eds.), Piaget in the classroom (pp. 199–215). New York,
NY: Basic Books.
Kennedy, L. M., & Tipps, S. (1994). Guiding children’s learning of mathematics (7th ed.).
Belmont, CA: Wadsworth.
Kintsch, W., & Greeno, J.G. (1985). Understanding and solving word arithmetic problems.
Psychological Review, 92, 109–129.
Knuth, E., Alibali, M., McNeil, N., Weinberg, A., & Stephens, A. (2005). Middle school students’
understanding of core algebraic concepts: Equality & variable. International Reviews on
Mathematical Education, 37, 68–76.
Knuth, E., Stephens, A., McNeil, N., & Alibali, M. (2006). Does understanding the equal sign
matter? Evidence from solving equations. Journal for Research in Mathematics Education,
37(4), 297–312.
 Development of diagram use 19

Koedinger, K. R., & Alibali, M. W. (1999). A developmental model of algebra problem solving.
Paper presented at the annual meeting of the American Educational Research Association in
Montreal, Quebec.
Koedinger, K. R., Alibali, M. W., & Nathan, M. J. (2008). Trade-Offs between grounded and abstract
representations: Evidence from algebra problem solving. Cognitive Science, 32, 366–397.
Koedinger, K. R., & Anderson, J. R. (1998). Illustrating principled design: The early evolution of a
cognitive tutor for algebra symbolization. Interactive Learning Environments, 5, 161–180.
Koedinger, K. R., & Nathan, M. J. (2004). The real story behind story problems: Effects of
representations on quantitative reasoning. The Journal of the Learning Sciences, 13, 129–
164.
Koedinger, K. R., & Terao, A. (2002). A cognitive task analysis of using pictures to support pre-
algebraic reasoning. In C.D. Schunn & W. Gray (Eds.), Proceedings of the Twenty-Fourth
Annual Conference of the Cognitive Science Society (pp.542–547). Mawah, NJ: Lawrence
Erlbaum Associates, Inc.
Kozma, R. (2003). Material and social affordances of multiple representations for science
understanding. Learning and Instruction, 13, 205–226.
Lappan, G., Fey, J., Fitzgerald, W., Friel, S., & Phillips, E. (2002). Getting to know Connected
Mathematics: An implementation guide. Glenview, IL: Prentice Hall.
Larkin, J., & Simon, H. (1987). Why a diagram is (sometimes) worth ten thousand words. Cognitive
Science, 11, 65–100.
Lee, K., Ng, E., & Ng, S. F. (2009). The contributions of working memory and executive functioning
to problem representation and solution generation in algebraic word problems. Journal of
Educational Psychology, 101, 373–387.
Lewis, A. B., & Mayer, R. E. (1987). Students’ misconceptions of relational statements in arithmetic
word problems. Journal of Educational Psychology, 79, 363–371.
Lowe, R.K. (2003). Animation and learning: Selective processing of information in dynamic
graphics. Learning and Instruction, 13, 157–176.
Luciana, M., Conklin, H. M., Hooper, C.J., & Yarger, R. S. (2005). The development of nonverbal
working memory and executive control processes in adolescents. Child Development, 76,
697–712.
Luna, B., Garver, K.E., Urban, T.A., Lazar, N.A., & Sweeney, J.A. (2004). Maturation of cognitive
processes from late childhood to adulthood. Child Development, 75, 1357–1372.
Marini, Z., & Case, R. (1994). Development of abstract reasoning about the physical and social
world. Child Development, 65, 147–159.
Mayer, R. E. (1989). Systematic thinking fostered by illustrations in scientific text. Journal of
Educational Psychology, 81, 240–246.
Mayer, R. (2005). Cognitive theory of multimedia learning. In R. Mayer (Ed.), The Cambridge
handbook of multimedia learning (pp. 31–48). Cambridge: Cambridge University Press.
Mayer, R. E. & Gallini, J. K. (1990). When is an illustration worth ten thousand words? Journal of
Educational Psychology, 82, 715–726.
Mayer, R. E., Steinhoff, K., Bower, G., & Mars, R. (1995). A generative theory of textbook
design: Using annotated illustrations to foster meaningful learning of science text. Educational
Technology Research and Development, 43, 31–44.
Meyer, J. (2000). Performance with tables and graphs: Effects of training and a visual search model.
Ergonomics, 43, 1840–1865.
Montague, M., Bos, C., & Doucette, M. (1991). Affective, cognitive, and metacognitive attributes
of eighth-grade mathematical problem solvers. Learning Disabilities Research & Practice, 6,
145–151.
Moreno, R., & Mayer, R. C. (1999). Multimedia-supported metaphors for meaning making in
mathematics. Cognition and Instruction, 17, 215–248.
Murata, A. (2004). Paths to learning ten-structured understanding of teen sums: Addition solution
methods of Japanese grade 1 students. Cognition and Instruction, 22, 185–218.
20 Julie L. Booth and Kenneth R. Koedinger

Murata, A. (2008). Mathematics teaching and learning as a mediating process: The case of tape
diagrams. Mathematical Thinking and Learning, 10, 374–406.
Narayanan, N. H., & Hegarty, M. (1998). On designing comprehensible interactive hypermedia
manuals. International Journal of Human-Computer Studies, 48, 267–301.
Nathan, M.J., Kintsch, W., & Young, E. (1992). A theory of algebra-word-problem comprehension
and its implications for the design of learning environments. Cognition & Instruction, 9,
329–389.
Nathan, M. J., & Koedinger, K. R. (2000). Moving beyond teachers’ intuitive beliefs about algebra
learning. Mathematics Teacher, 93, 218–223.
National Center for Education Statistics (2003). Teaching mathematics in seven countries: Results
from the TIMSS 1999 video study. Washington, DC: Author.
National Council of Teachers of Mathematics (NCTM) (2000). Principles and standards for school
mathematics.. Reston, VA: Author.
Nunes, T., Schliemann, A. D., & Carraher, D. W. (1993). Street mathematics and school
mathematics. New York: Cambridge University Press.
Postigo, Y., & Pozo, J. I. (2004). On the road to graphicacy: The learning of graphical representation
systems. Educational Psychology, 24, 623–644.
Rittle-Johnson, B., & Koedinger, K. (2001). Using cognitive models to guide instructional design:
The case of fraction division. In J. Moore & K. Stenning (Eds.), Proceedings of the 23rd Annual
Conference of the Cognitive Science Society (pp. 857–862). Mahwah, NJ: Erlbaum.
Rittle-Johnson, B., & Koedinger, K. R. (2005). Designing knowledge scaffolds to support mathe-
matical problem solving. Cognition and Instruction, 23, 313–349.
Romberg, T. A., & de Lange, J. (2001). Monitoring student progress. In T. A. Romberg &
J. de Lange (Eds.), Insight stories: Assessing middle school mathematics. New York: Teachers
College Press.
Schwartz, D., & Black, J. B. (1996). Shuttling between depictive models and abstract rules:
Induction and fallback. Cognitive Science, 20, 457–497.
Sowell, E. J. (1989). Effect of manipulative materials in mathematics instruction. Journal for
Research in Mathematics Education, 20, 498–505.
Uttal, D. H., Liu, L. L., & DeLoache, J. S. (2006). Concreteness and symbolic development. In
L. Balter & C. S. Tamis-LeMonda (Eds.), Child psychology: A handbook of contemporary
issues (2nd ed., pp. 167–184). New York: Psychology Press.
Uttal, D. H., Scudder, K. V., & DeLoache, J. S. (1997). Manipulatives as symbols: A new perspective
on the use of concrete objects to teach mathematics. Journal of Applied Developmental
Psychology, 18, 37–54.
Van Meter, P., & Garner, J. (2005). The promise and practice of learner-generated drawing:
Literature review and synthesis. Educational Psychology Review, 17, 285–325.
Zawaiza, T. R. W., & Gerber, M. M (1993). Effects of explicit instruction on math word-
problem solving by community college students with learning disabilities. Learning Disability
Quarterly, 16, 64–79.

Received 14 April 2010; revised version received 3 June 2011