You are on page 1of 20

Cognitive Development and the Structuring of Geometric Content

Author(s): Janet L. McDonald


Source: Journal for Research in Mathematics Education, Vol. 20, No. 1 (Jan., 1989), pp. 76-94
Published by: National Council of Teachers of Mathematics
Stable URL: http://www.jstor.org/stable/749099
Accessed: 29-02-2016 21:25 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/749099?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

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

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Journal for Research in Mathematics Education

1989, Vol. 20, No. 1, 76-94

COGNITIVE DEVELOPMENT AND THE

STRUCTURING OF GEOMETRIC CONTENT

JANET L. MCDONALD, Union College

From a pool of secondary students (n = 161), 20 students were chosen who had high (formal

operations) scores and 20 who had low (concrete operations) scores on two paper-and-pencil

measures of Piagetian formal reasoning (the Test of Logical Thinking and the Longeot Test).

The students made similarity judgments among all possible concept pairs from 13 geometric

concepts and 10 mathematical expressions from a unit on ratio, proportion, and similarity.

Multidimensional scaling procedures showed that despite idiosyncrasies in individual structures,

clear prototypical maps could be derived for both the formal and concrete groups. In addition,

formal operational students structured subject matter content significantly more like subject

matter experts than concrete operational students did.

As students process information about a subject matter domain, they restructure

concepts and the relations among concepts (Johnson, 1969; Shavelson, 1974; Thro,

1978). Initially, students differentiate among concepts on the basis of such rela-

tive attributes as shape (triangular or square) or type of figure (right or equilateral

triangle), and they construct class inclusion principles (all triangles are polygons).

As students are introduced to increasingly abstract concepts such as geometric mean

and exposed to relational operators such as equal, congruent, and similar, their

mental models of the relationships among concepts require continual restructur-

ing. From the perspective of a network model of memory (Collins & Loftus, 1975),

learning includes both increasing the number of concepts and elaborating hierar-

chical relations among concept types. Thus, in geometry a student may add more

types of triangles to memory while at the same time relationally subsuming the

concept of triangle by concepts such as similar or congruent. From the perspec-

tive of the van Hiele model of the development of geometric thought, the student

moves from observing and identifying the figure to a recognition of its properties,

and then to understanding the interrelationships of the properties of the figures and

the axiomatic system within which they are placed (Crowley, 1987; Hoffer, 1983).

Three major factors are thought to affect this restructuring: the complexity of

the subject domain (Piaget, 1972; Preece, 1978), the characteristics of the learner

(Piaget, 1972; Stasz, Shavelson, Cox, & Moore, 1976), and the method and organi-

zation of instruction (Burger & Shaughnessy, 1986; Fuys, Geddes, & Tischler,

1985). Typically, research related to the structuring of knowledge has been drawn

from content areas within mathematics and science because of their high level of

structure and the ability of experts to agree on the structural relationships between

This report is based on portions of a doctoral dissertation completed at the State

University of New York at Albany in 1982 under the direction of Margaret A. Farrell and Fred

W. Ohnmacht. I would also like to thank Elizabeth B. Leonardi, Vicky L. Kouba, and the

JRME reviewers for their assistance in the preparation of this article.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
77

the many concepts (Branca, 1980; Geeslin & Shavelson, 1975; Johnson, 1969;

Schoenfeld & Herrmann, 1982). Studies of learner characteristics related to knowl-

edge structures have included investigations of field independence (Stasz et al.,

1976) but have concentrated on differences between expert and novice problem

solvers (Champagne & Klopfer, 1981; Novak, 1983; Reif & Heller, 1982; Schoen-

feld & Herrmann, 1982; Silver, 1982). Studies attempting to verify the van Hiele

model have investigated materials and order of content in attempts to determine

the effects of curricular and instructional differences on these developing structures

(Burger & Shaughnessy, 1986; Crowley, 1987; Fuys et al., 1985; Hoffer, 1983).

Relationships between representations and formal operations, however, have

received comparatively little attention (Preece, 1978). Yet, students without for-

mal operational schemata (Inhelder & Piaget, 1958) may not be capable of estab-

lishing the relatively complex coordinations among concepts characteristic of

semantically rich (Simon, 1979) and highly structured domains such as Euclidean

plane geometry. If we assume that understanding requires knowledge of the struc-

ture of the subject matter (Champagne, Klopfer, DeSena, & Squires, 1981; Piaget,

1970; Schwab, 1962) and that only people who are formal operators are able to

structure the abstract principles within a given domain, then the students' knowl-

edge representations may be constrained by their cognitive-developmental level

(Lawson & Renner, 1975; Strauss & Kroy, 1977).

The purpose of this study was to determine the relationship between high school

students' level of cognitive development and the way in which they structured

geometric content. Because significant numbers of students in high school mathe-

matics classes are not formal operational (Collis, 1975; Farmer, Farrell, Clark, &

McDonald, 1982; Renner et al., 1976; Shayer & Adey, 1981), the implications for

teaching may be substantial. And if the teaching occurs at a level beyond that of

the student, then the content may not be properly assimilated into the student's

memory (Hoffer, 1983).

Because structural knowledge is difficult to assess using conventional testing

techniques, this study employed a methodology that compares a person's hypotheti-

cal cognitive structure of a specific content domain to the idealized content struc-

ture as defined by experts. Various techniques are available that produce visual

representations (maps) of cognitive structure (Eignor, 1978), including multidimen-

sional scaling (MDS) (Kruskal, 1964a, 1964b). The maps are created from judg-

ments that students make about the similarity of certain items in the content do-

main. The resulting numerical values are placed in a matrix referred to as a

similarity or distance matrix. MDS reduces the similarity matrix to a spatial rep-

resentation of the concepts, typically in two or more dimensions. By comparing

the scaling of the structure created by experts to the matrix representing a given

student's structure, one can determine how closely the two mappings are matched

by computing correlations between the similarity matrices.

To compare structures, a distinction is made between content structure and

cognitive structure. Content structure is a hypothetical construct representing the

ideal organization of concepts. Operationally, content structure can be defined by

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
78 Structuring Geometric Content

experts because it is assumed that experts in a given field can generate a complex

and accurate model of the relationships in that content domain (Johnson, 1967; Thro,

1978). Cognitive structure is a hypothetical construct representing a person's

organization of concepts in long-term memory. It is the person's idiosyncratic,

internalized representation following learning (Geeslin & Shavelson, 1975; Hayes-

Roth, 1977; Shavelson, 1974). The degree of student knowledge of geometric re-

lationships can be determined by examining the goodness of fit between the

student's cognitive structure and the content structure defined by experts.

The study sought answers to the following questions:

1. Are the individual maps of the concrete operational students sufficiently alike

to allow for the creation of a prototypical map? If so, what does the prototypical

map suggest about the structuring of geometric content by these students?

2. Are the individual maps of formal operational students sufficiently alike to

allow for the creation of a prototypical map, and, if so, what does such a map suggest

about the structuring of geometric content by these students?

3. Are the cognitive maps of formal operational students more like the content

map of experts than the cognitive maps of concrete operational students?

4. If these maps differ, in what ways do they differ?

METHOD

Subjects

The sample for this study consisted of 20 students identified as concrete opera-

tional and 20 students identified as formal operational. The sample of 40 students

was drawn from a pool of 161 college-bound students in six 10th-grade geometry

classes in a suburban high school.

The sample was chosen on the basis of their scores on two tests of logical think-

ing that contained tasks requiring formal operations, the Test of Logical Thinking

(TOLT) and the Longeot Test of Formal Operations. To control for general abil-

ity as an intervening variable, IQ scores for the students in the sample were obtained.

The mean IQ score (Lorge-Thorndike, Form L) for the 20 concrete operational

students was 110.1 (SD = 8.21), and the mean IQ score for the 20 formal opera-

tional students was 113.9 (SD = 9.13). The differences between these scores were

not significant at the .05 level, t(38) = 1.37. Correlations between IQ and the TOLT

and IQ and the Longeot Test were .19 (p < .05) and .23 (p < .05), respectively.

Although statistically significant, the modest magnitude of these coefficients sug-

gests that Piagetian tasks contain components that are distinct from general intel-

ligence tests (Humphreys & Parsons, 1979; Humphreys, Rich, & Dewey, 1985).

Approximately half the 161 students in the pool were enrolled in one of three

geometry classes taught by Teacher A using the text Geometry by Hirsch, Roberts,

Coblentz, Samide, and Schoen (1979). The remaining students were enrolled in

one of three classes taught by Teacher B using the text Geometry by Jurgensen,

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 79

Donnelly, Maier, and Rising (1975). On the basis of an analysis of the scope and

sequence of topics and the axiomatic development, a team of content experts judged

these two texts to be parallel in their axiomatic structure and symbol use. The 40

students in the sample were distributed across teachers and classes as shown in

Table 1.

Table 1

Frequency Distribution of Students Across Teachers and Classes

Teacher A Teacher B

Subjects Class 1 Class 2 Class 3 Class 1 Class 2 Class 3

Pool 28 26 28 27 24 28

Concrete 4 3 4 3 2 4

Formal 3 5 2 4 4 2

Experts

Five mathematics educators, each with a minimum of 15 years teaching experi-

ence, served as the content experts. Two were high school geometry teachers; three

were university professors who had also taught high school geometry. One pro-

fessor was a geometer and the author of two geometry textbooks. The second was

an author and mathematics educator whose courses included upper level geome-

try and the study of the structure of mathematics. The third was a mathematics

educator.

Instruments

Assessment of cognitive level. Each student's cognitive level was assessed by a

mean standardized score on two paper-and-pencil tests of Piagetian formal reason-

ing. The Longeot Test (Sheehan, 1970) measures propositional thinking, propor-

tional reasoning, and combinatorial thinking. The reliability of the Longeot test

has been reported in several studies for a variety of sample sizes and characteris-

tics. Reliability estimates for samples that most closely match that of this study

range from .72 to .85. Substantial work has also been reported related to the con-

struct, factorial, and criterion-related validity of the Longeot Test (McDonald &

Sheehan, 1983). The TOLT (Tobin & Capie, 1981) assesses the ability to control

variables and probabilistic, correlational, combinatorial, and proportional reason-

ing. Tobin and Capie have reported high test reliability (coefficient alpha = .85),

high factorial validity, and correlations of .80 with the test and clinical interviews

using five Piagetian and Piagetian-like tasks.

Measures ofcognitive structure. Two instruments (Tasks A and B) were designed

to measure the students' understanding of similarity. For each task the student was

required to judge on a scale of 1 to 10 the similarity of two given terms or sym-

bolic expressions. The first instrument (Task A) was built from 13 terms chosen

from the key concepts from the similarity unit: ratio, proportion, equal, similar,

congruent, equilateral triangle, right triangle, parallel, corresponding sides, cor-

responding angles, isosceles triangle, geometric mean, and polygon.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
80 Structuring Geometric Content

For this task, the student compared all possible pairs of the terms. The student

was given each pair of terms in turn and asked to make a numerical judgment of

their relationship. The numbers ranged from 1 to 10, where 1 indicated very little

or no relationship, 4 a limited relationship, 7 a strong relationship, and 10 that the

terms were identical. For example, when asked to give a number that would rep-

resent how similar the term ratio was to the term proportion, a student might re-

spond with the number 8 or 9 because the two terms are highly related.

The second instrument (Task B) was built from 10 mathematical symbolic rela-

tionships from the same unit:

AABC ~ ADEF AB A D

DE DF

AABC ~ AEDF AC DF

AC D AC - DF(AB)(DF ) = (DE)(AC )
AABC = ADEF BA ED

AB DE (AB + AC ) (DE + EF )

AB II DE AC - EF AC EF

Represented within this set of expressions were most of the concepts from the Task

A list, but in a more abstract, symbolic form. For Task B, the student was given a

pair of the expressions above and asked to compare them using the same scale as

for the first task. A typical student evaluation of the degree of relationship between

AABC ~ AEDF and AC/BA = DF/ED was the number 3 because the relationship

is limited to the fact that the left-hand ratio is equivalent to the ratio of the lengths

of the sides of the triangle. Each student was given all combinations of pairs from

the 10 expressions.

To judge the relationship of the pairs on Task B, the student would presumably

have to be familiar with the mathematical symbols, theorems, postulates, and

definitions that correspond to the given expression and to the mathematical, struc-

tural relationship between expressions. For example, to recognize that ABIDE =

ACIDF relates highly with AABC ~ ADEF, the student must recognize that AB,

AC, DE, and DF are corresponding sides of the similar triangles. At the same time

the student must recognize that ABIAC = DE/EF would be an inappropriate

correspondence because the sides given are not corresponding sides.

Pilot tests of Tasks A and B were administered to 23 students on two separate

occasions, 18 days apart, in a school district and class comparable to those in the

major study. The students' similarity judgments were converted to similarity

matrices and subjected to multidimensional scaling analysis using MDSCALE

(Rohlf, Kishpaugh, & Kirk, 1977). The rank order correlations obtained were then

converted to z scores using Fisher's r to z transformation. The resulting z scores

were correlated, yielding reliability coefficients of .71 for Task A and .77 for Task B.

Procedure

The 40 students in the sample were selected on the basis of their combined per-

formance on the Longeot Test of Formal Operations and the TOLT. Raw scores

on the two tests were standardized and averaged to form a mean z score for each

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 81

student. These scores were ranked and used to identify students as concrete op-

erational and formal operational. Since only the lowest 20 standard scores could

be clearly classified as concrete operational on both measures of cognitive devel-

opment, only the top 20 formal scores were selected to represent the formal group.

Following an instructional unit on ratio, proportion, and similarity, all 161 stu-

dents were administered Tasks A and B and were instructed to make their judg-

ments in the context of the similarity unit. All students easily completed both tasks

within a 40-minute class period. The five content experts were administered Tasks

A and B in order to derive the content structure for each task. The experts initially

made their own judgments of the concepts independently. After each expert's

matrix of similarity judgments had been determined, the experts met and discussed

their interpretations, eventually forming a consensus matrix. The consensus ma-

trix for each task, A and B, was transformed using multidimensional scaling into

an expert map for subsequent analysis. Correlations of the experts' individual

similarity matrices for Tasks A and B with the consensus expert maps ranged from

.86 to .95.

Data Analysis

The similarity matrices from each of the two tasks were analyzed for each stu-

dent in the sample. The similarity matrices were then standardized, and distance

coefficients were determined for each concept pair using the Euclidean distance

algorithm (Shepard, 1980). The resulting distance matrices were subjected to

nonmetric multidimensional scaling analysis (Kruskal, 1964a) using MDSCALE.

This analysis yields a measure of stress, or goodness of fit, for the distances in the

configuration space to a monotone function of the original distances. The rank order

correlations between each student's maps and the expert map were converted to z

scores and tested for significance using an analysis of variance. In this way, each

student map was statistically compared to the expert map. Prototypical matrices

were formed for the concrete and formal groups using aggregate means for each

set of scores. These prototypical maps were then correlated with the expert maps

for both tasks and with each of the individual student maps.

RESULTS

Task A: Expert Map

Multidimensional scaling analyses were performed on the expert consensus

matrix in several dimensions to determine the number of dimensions through which

to perform the analyses. The best fit for Task A was a four-dimensional model.

Although the statistical analyses were performed using a four-dimensional model,

the two-dimensional model shown in Figure 1 is sufficient to describe both the

dimensionality and the clustering of the concepts of the map.

The first dimension, the vertical axis, represents a range of equality from the most

equal at the top (congruent and equal) to least equal at the bottom (proportion and

geometric mean). The triangles are ordered from equilateral triangle (with the most

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
82 Structuring Geometric Content

* Equal

Corresponding

Congruent

Angles

Equilateral

Triangle

Isosceles

Parallel

Triangle

Similar

Polygon*

Corresponding

Sides

Ratio

Right

Proportion

Triangle

Geometric

Mean

Figure 1. Task A : Expert consensus map.

congruent parts) to right triangle (with the fewest necessarily congruent parts). That

the term corresponding angles is near the top reflects the fact that corresponding

angles are congruent; that the term corresponding sides is nearer the bottom re-

flects that corresponding sides of similar figures are proportional. The location of

the term parallel is mathematically consistent because parallel lines are equidis-

tant and because many theorems involving parallels yield corresponding and con-

gruent angles.

The second dimension, the horizontal axis, can be considered as a whole/part

continuum, with whole figures such as polygons and triangles on the right and parts

of figures such as angles and sides on the left. To understand the third dimension,

imagine an oblique plane with the terms congruent and equal slightly above the

first quadrant and the terms ratio and proportion slightly below the third quadrant.

This dimension is predominantly composed of relational operators such as congru-

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 83

ent, equal, ratio, and proportion. The fourth dimension is defined solely by the

term parallel. Similar, the term central to the unit, is closest to the intersection of

all four axes.

The relationships of the concepts in the similarity unit are represented not only

by the dimensionality of the map but also by the clustering of terms such as equal

and congruent and the terms equilateral triangle and isosceles triangle. The clus-

tering of the terms corresponding sides, ratio, and proportion is also mathemati-

cally meaningful, as is the proximity of this cluster to the terms geometric mean

and right triangle. Each of these terms is highly related by the many right-triangle

similarity theorems that yield both corresponding proportional sides in the same

ratio and geometric means.

Task A: Prototypical Formal Map

The map representing the prototypical formal operational student is shown in

Figure 2. This map represents visually the moderately high correlation (r = .61)

between the prototypical formal student map and the expert map. The correlations

between the individual formal maps of all 13 terms and the prototypical map ranged

from .77 to .92, with a mean correlation of .86, indicating that the prototypical map

was representative of the entire group.

As in the expert map, the vertical dimension of the prototypical formal map

represents a range of equality from most to least equal. Likewise, the order of the

terms along the vertical dimension is generally consistent with the expert consen-

sus map. With the exception of the terms parallel and polygon, the horizontal

dimension is more compressed than that of the experts, but the order is consistent.

Unlike the experts' judgments, the judgments of the formal operational students

did not result in a dimension described by relational operators. Rather, the proto-

typical formal map is three-dimensional, with the third dimension being represented

solely by the term parallel.

The clusters of the terms equal and congruent and the terms proportion and ra-

tio are similar to those of the experts. Unlike the experts, however, the formal

operational students clustered the terms corresponding angles, similar, and corre-

sponding sides.

Task A: Prototypical Concrete Map

The prototypical concrete map (see Figure 3) represents visually the moderately

low correlation with the expert map (r = .35). Yet the prototypical map correlated

quite highly with the individual concrete maps, with correlations from .65 to .82

and a mean correlation of .77.

The vertical dimension of the prototypical concrete map represents a dimension

of equality, but the ordering of the terms differs from both the expert map and the

prototypical formal map. In particular, the terms corresponding sides and corre-

sponding angles have been reversed in order, with corresponding sides occupying

a position much nearer to the top of the axis and positioned between congruent and

equal.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
84 Structuring Geometric Content

Equal *

Congruent *

Equilateral

Triangle

Corresponding

Angles

Isosceles

< Similar

Triangle

Corresponding

Sides

Parallel <

* Proportion

< Ratio

Polygon

Geometric

Mean

Right

Triangle

Figure 2. Task A: Prototypical formal student map.

The horizontal axis of the prototypical concrete map is somewhat more com-

pressed than that of the prototypical formal map. In fact, with the exception of the

terms parallel and polygon, the remaining terms can be described by a nearly lin-

ear model. Like the prototypical formal map, the prototypical concrete map does

not contain a separate dimension representing the relational operators. As with the

other two maps, parallel occupies its own dimension.

The clustering on the concrete map is also different from both the expert and

prototypical formal maps. On the concrete map, there is a large cluster in the second

quadrant of the terms congruent, corresponding sides, equal, corresponding angles,

and equilateral triangle. The terms ratio and proportion are more separate than in

the other two maps, as are the terms congruent and equal.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 85

Congruent

Corresponding

Sides

Equal

Corresponding *

Angles

Equilateral

Triangle

Parallel

SIsosceles

Triangle

Similar

Proportion

Right

Ratio
Triangle Polygon

Geometric

Mean

Figure 3. Task A : Prototypical concrete student map.

Task A: Individual Maps Across Levels

An analysis of variance comparing the correlations with the expert map for each

of the concrete and formal operational students for Task A is summarized in Table

2. As the table shows, the maps of the formal operators correlated significantly

higher (p < .0001) with the expert map than the maps of the concrete operators did.

There were no significant teacher effects.

Discussion of Task A

The decrease in the number of dimensions and the compression of the range for

many terms along the axes reflected both the formal and concrete students' inabil-

ity to differentiate concepts as clearly as the experts did. For example, the rela-

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
86 Structuring Geometric Content

Table 2

Analysis of Variance of Correlations Between the Expert Map and Individual

Student Maps-Task A

Variance

Source SS df estimate F

Between 0.63 3

Cognitive Level 0.61 1 0.61 55.60*

Teacher 0.02 1 0.02 1.61

Cognitive Level x Teacher 0.00 1 0.00 0.26

Within 0.39 36 0.01

*p < .0001.

tional operators congruent, equal, ratio, and proportion were distinguished by the

experts as related to, but still somehow different from, the other terms (as reflected

by the oblique plane). Although the formal students' judgments did not yield a

separate dimension for these terms, they did result in much more distinct clusters

than are present in the concrete students' map. Structural relationships among the

triangles that are quite understandable on the expert map (Figure 1) are less clear

but still reasonable on the formal map (Figure 2). On the concrete map (Figure 3),

however, there is no apparent structural relationship among the triangles themselves

or between the triangles and the rest of the terms in the unit. Although the term

parallel described its own dimension on all three maps, its orientation with respect

to the other two axes is much more interpretable on the expert map than on either

of the two prototypical student maps.

A critical difference in the three maps involves the terms corresponding angles

and corresponding sides. In the expert map, the term corresponding angles is near

the top of the map, adjacent to the terms congruent and equal. This location is

consistent with the fact that corresponding angles of similar figures (and of paral-

lel lines) are congruent. On the other hand, the term corresponding sides lies in

the third quadrant of the expert map, clustered with the terms ratio and proportion.

This location is consistent with the fact that the corresponding sides of similar

figures are in the same ratio and thus are proportional. In the prototypical formal

map, the term corresponding sides has moved into the second quadrant, much closer

to the term corresponding angles. This arrangement may have resulted from the

formal operational students seeing the terms as similar because each starts with the

same word; the experts realized that the terms should be conceptually linked to other

terms rather than to each other. The prototypical concrete student also located these

two terms near each other, but the reversed ordering of the two terms on the verti-

cal axis represents an invalid relationship because the term corresponding sides is

closer to congruent, the term corresponding angles is closer to the term propor-

tion, and both terms are much too far from the terms proportion and ratio. The

clustering of the five concepts in the second quadrant of Figure 3 is mathematically

more reasonable if considered in the context of geometry outside of the similarity

unit. The concrete students may have based their judgments at least partially on

the relationships of congruent figures despite the instruction that their judgments

be based exclusively on the similarity unit.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 87

The loss of the clustering of the terms congruent and equal and the terms

proportion and ratio as one goes from Figure 1 to Figure 3 seems indicative of the

concrete students' inability to integrate these concept pairs within the unit as a

whole. Like the experts and the formal operational students, the concrete opera-

tional students judged the two sets of pairs as nearly the same, assigning each pair

a numerical value of 8, 9, or 10. The best fit model, however, could not maintain

this proximity in the map because the concrete operational students inconsistently

related the terms to other terms in the unit. That the concrete students were un-

aware of these inconsistencies in their reasoning is characteristic of the level of

concrete operations (Inhelder & Piaget, 1958).

Task B: Expert Map

The two-dimensional map representing the three-dimensional model formed as

a result of the consensus of the expert judgments on Task B is shown in Figure 4.

The vertical dimension appears to distinguish between expressions directly related

to AABC - ADEF (near the bottom of the axis) and expressions that are a re-

sult of AABC being similar to AEDF (near the top of the axis). Thus, the ex-

pression (AB + AC)/AC = (DE + EF)/EF is the furthest away from AABC~- ADEF

because (a) it is not derived from that particular pair of similar figures and (b) it is

not an immediate result of the similarity of triangles ABC and EDF.

The horizontal dimension is less well-defined but appears to represent parallels,

congruences, proportions, and similarity from left to right. The location of the

proportions between the similar statements and the congruent statements makes

sense because the proportions would hold whether the triangles were similar or

congruent. As in Task A, the expression relating to parallels in Task B (AB II DE)

determined its own dimension.

An examination of the clusters within this map reveals more about the relation-

ships than the dimensions do. The cluster in the first quadrant consists of the

proportions based on triangles ABC and EDF. The cluster in the third quadrant

shows AC/BA = DF/ED immediately adjacent to AABC- ~ADEF because it is a

direct proportion, and ABIDE = AC/DF and the product (AB)(DF) = (DE)(AC) are

also close because they are relationships derived from the proportions rather than

from the triangle similarity. Were the same triangles to be congruent rather than

similar, these same relationships would still be true, thus making sense of the

proximity of the triangle congruence statement AABC - ADEF. In each of these

cases, ZA would be congruent to ZD, accounting for its presence in the cluster as

well.

Task B: Prototypical Formal Map

The aggregate cognitive representation of Task B for the formal operational

students is seen in Figure 5. The prototypical map correlated quite highly with the

individual formal maps, with correlations ranging from .85 to .97 and a mean

correlation of .93. The high visual similarity of Figures 4 and 5 also reflects the

high statistical correlation of the expert and prototypical formal maps (r = .96).

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
88 Structuring Geometric Content

SAB//DE

(AB+AC)/AC=(DE+EF)/EF *

AB/AC=DE/EF *

A ABC ---AEDF *

<A=<D *

AAC/BA=DF/ED

A ABC - DEF

AB/DE=AC/DF? f A ABC-A DEF


(AB)(DF)= (DE)(AC)

Figure 4. Task B : Expert consensus map.

As in the expert map, all the expressions related to triangles ABC and DEF occupy

the third quadrant in Figure 5, and all the expressions related to triangles ABC and

EDF occupy the first quadrant. Thus, the vertical axis again appears to serve the

function of separating the two sets of triangles. The horizontal axis is much more

compressed than that of the expert map and is actually quite close to a two-dimen-

sional model, with the second dimension being defined solely by AB II DE.

The clustering of the terms in the prototypical formal map is similar to that of

the expert map except that the first quadrant cluster is somewhat more compressed

and the third quadrant cluster has all six expressions occupying the same location.

Task B: Prototypical Concrete Map

The map derived from the concrete subjects on Task B is shown in Figure 6.

The correlations between the prototypical concrete map and the individual concrete

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 89

(AB+AC)/AC=(DE+EF)/EF *

AB/AC=DE/EF *

A ABC L-/ EDF *

* AB/DE

AC/BA=DF/ED

AB/DE=AC/DF

A ABC' L DEF

(AB) (DF) = (DE) (AC)

A ABC- A DEF *

<A=<D

Figure 5. Task B : Prototypical formal student map.

maps ranged from .79 to .95, with a mean of .86. Because the greatest difference

between the expert map and the concrete map was in the spread of the clusters, the

prototypical concrete map correlated quite highly (r = .81) with the expert map.

As in Figures 4 and 5, the vertical axis in Figure 6 serves to separate the two pairs

of triangles. The locations of the expressions along the vertical dimension are

somewhat similar to the expert map, with the exception of ZA - ZD, ABIDE =

AC/DF and (AB)(DF) = (DE)(AC).

In general, the terms are located along the horizontal dimension as follows: those

related to parts (segments and angles) appear on the left, followed by wholes (the

triangle relationships), and then numerical relationships (derived proportions and

products) on the right. The proportions that are a direct result of the correspond-

ing sides of those triangles being proportional tend to be to the immediate right of

the corresponding triangle-similarity statement (e.g., AB/AC = DE/EF is to the right

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
90 Structuring Geometric Content

(AB+AC)/AC= (DE+EF)/EF

* AB//DE

AB/AC=DE/EF

A ABC--A EDF

(AB)(DF)=(DE)(AC) *

AB/DE=AC/DF *

* AC/BA=DF/ED

A ABC' ADEF *

<A <D n A ABC- A DEF

Figure 6. Task B : Prototypical concrete student map.

of AABC - AEDF), whereas the relationships involving products (e.g., (AB)(DF) =

(DE)(AC)) or proportions (e.g., (AB + AC)/AC = (DE + EF)/EF) based on the initial

proportions are even further to the right. There is, however, no mathematically clear

rationale to justify such an arrangement. As in the expert and prototypical formal

maps, the expression AB II DE defined its own dimension.

Although the ordering of the terms along the axes is quite consistent with the

expert and prototypical formal maps, the clustering of the concepts is not. In the

third quadrant, there is a relatively tight cluster of AABC- ~ADEF and AABC

ADEF, but the remaining terms are relatively unclustered.

Task B: Individual Maps Across Levels

An analysis of variance comparing the correlations with the expert map for each

of the concrete and formal operational students for Task B is summarized in Table

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 91

3. As with Task A, the maps of the formal operators correlated significantly higher

(p < .001) with the expert map than the maps of concrete operators did. There were

also no significant teacher effects.

Table 3

Analysis of Variance of Correlations Between the Expert Map and Individual

Student Maps-Task B

Variance

Source SS df estimate F

Between 2.18 3

Cognitive Level 1.51 1 1.54 7.36*

Teacher 0.04 1 0.04 0.18

Cognitive Level x Teacher 0.60 1 0.60 2.87

Within 7.55 36

*p < .001.

Discussion of Task B

The major difference between the expert and the formal maps (Figures 4 and 5)

was in the tight clustering of the concepts in the third quadrant. The fact that the

prototypical formal students' judgments resulted in all the terms being placed at

the same location reflects the formal students' inability to recognize the subtle

distinctions between expressions that were highly related mathematically. The

concrete students, on the other hand, failed to see these expressions as highly related

(Figure 6). The experts and the formal students recognized that the products and

proportions that held for the first set of triangles would not hold for the second set

of triangles and therefore separated the two. Also, the extreme placement of ZA _

ZD by the concrete students may reflect their knowledge that ZA must be congruent

to ZD for AABC and ADEF, whereas the placement by the experts acknowledges

the possibility that ZA might also be congruent to ZD for AABC and AEDF as

well.

Despite these differences, the concrete students' map was closer to the map of

the experts than anticipated. Although the task involved abstract mathematical

symbolism, it is a routine instructional procedure to instruct students in an algorithm

for dealing with the symbols. It is possible that the concrete operational students

were using a rote procedure, one that was not available to them for Task A. Such

a step-by-step algorithm is well within the capabilities of students operating at the

concrete level (Farrell, 1987; Inhelder & Piaget, 1958). The high correlation

between the expert and concrete maps may also have been a consequence of the

multidimensional scaling approach, which emphasizes rank order over cluster.

CONCLUSIONS

Despite idiosyncracies in individual student maps, clear prototypical maps could

be derived for both the formal and concrete operational groups. The prototypical

maps reflected significant differences in the ways in which the students organized

subject matter content. These results are particularly significant because the

prototypes were drawn from students exposed to different texts and different

teachers.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
92 Structuring Geometric Content

The concrete operational students' organization of the content indicates that

confusion with terms may lead to confusion with concepts. On one level, they

apparently assumed that because two terms begin with the same word

(corresponding sides and corresponding angles) they should be highly related

mathematically. On another level, they apparently assumed that relationships

among terms persist regardless of context. For instance, they seemed to assume

that the relationship between corresponding sides and congruent that exists in

congruence units is maintained in the similarity unit.

Although the concrete operational students were capable of making reasonably

accurate judgments of the relationships of some given pairs of terms, their

judgments of the relationships of those same terms with the remaining terms from

the unit were inconsistent. Apparently, they either accepted, failed to see, or could

not resolve inconsistent judgments across multiple terms.

The formal operational students organized the subject matter significantly more

like the content experts than did the concrete operational students. Neither student

group, however, was able to judge the relational terms as sufficiently different from

the other terms to generate their own plane. It may be that the ability to make this

distinction reflects a level of understanding that identifies the experts as expert.

These results suggest that teachers of geometry need to attend carefully to the

role of language in students' understanding of the relationships between concepts

and, in particular, of how those relationships change from unit to unit. Teachers

must also beware of assuming that a knowledge of theorems means an

understanding of the relationships given in and across the theorems. Thus, teachers

need to provide students with problems and experiences that will help them build

and organize a framework for the concepts and relationships (such as in Crowley,

1987, and Farrell, 1987).

Although the results indicate that formal operational students produce different

structures from those of concrete operational students, further research is needed

to identify the implications of these differences and whether or not they are

important. It may be that although concrete operational students' maps differ from

those of experts, they nevertheless have an adequate, functioning structure of the

content. On the other hand, the differences in the maps may indicate meaningful

differences in knowledge of the content. In-depth interviews that reveal students'

rationales for their judgments may help clarify this issue. The significance of

structural differences may also be revealed by studies that relate structure to

achievement, stability, and retention over time.

Further research might also employ other concept-mapping techniques (such as

Novak, 1983) in order to determine relational aspects that can only be inferred by

a multidimensional scaling approach. Such models might also be used to examine

the relationships between cognitive structure and different levels of geometric

thought within the van Hiele model, possibly resulting in instructional strategies

tailored to the student's cognitive level. Such investigations might also approach

structural differences from additional developmental perspectives, such as skill

theory (Fischer, 1980).

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
Janet L. McDonald 93

REFERENCES

Branca, N. A. (1980). Communication of mathematical structure an its relationship to achievement.

Journal for Research in Mathematics Education, 11, 37-49.

Burger, W. F., & Shaughnessy, J. M. (1986). Characterizing the van Hiele levels of development in

geometry. Journal for Research in Mathematics Education, 17, 31-48.

Champagne, A. B., & Klopfer, L. E. (1981). Structuring process skills and the solution of verbal

problems involving science concepts. Science Education, 65, 493-511.

Champagne, A. B., Klopfer, L. E., DeSena, A. T., & Squires, D. A. (1981). Structural representation

of students' knowledge before and after science instruction. Journal of Research in Science Teach-

ing, 18, 97-11.

Collins, A. M., & Loftus, E. F. (1975). A spreading activation theory of semantic processing. Psy-

chological Review, 82, 407-428.

Collis, K. F. (1975). A study of concrete and formal operations inschool mathematics: A Piagetian

viewpoint. Melbourne: Australian Council for Educational Research.

Crowley, M. L. (1987). The van Hiele model of the development of geometric thought. In M. M.

Lindquist & A. P. Shulte (Eds.), Learning and Teaching Geometry, K-12 (1987 Yearbook of the

National Council of Teachers of Mathematics, pp. 1-16). Reston, VA: NCTM.

Eignor, D. R. (1978, April). Statistical suggestions for the study of cognitive structures. Paper pre-

sented at the annual meeting of the American Educational Research Association, Toronto.

Farmer, W. A., Farrell, M. A., Clark, R. M., & McDonald, J. L. (1982). A validity study of two paper-

and-pencil tests of concrete and formal operations. Journal of Research in Science Teaching, 19,

457-485.

Farrell, M. A. (1987). Geometry for secondary school teachers. In M. M. Lindquist & A. P. Shulte

(Eds.), Learning and TeachingGeometry, K-12 (1987 Yearbook of the National Council of Teach-

ers of Mathematics, pp.236-250). Reston, VA: NCTM.

Fischer, K. W. (1980). A theory of cognitive development: The control and construction of hierar-

chies of skills. Psychological Review, 87, 477-531.

Fuys, D., Geddes, D., & Tischler, R. (Eds.). (1985). An investigation of the van Hiele model of think-

ing in geometry among adolescents (Final report of the Investigation of the van Hiele Model of

Thinking in Geometry Among Adolescents Project). Brooklyn, NY: Brooklyn College, School of

Education.

Geeslin, W. E., & Shavelson, R. J. (1975). An exploratory analysis of the representation of a mathe-

matical structure in students' cognitive structures. American Educational Research Journal, 62,

21-39.

Hayes-Roth, B. (1977). Evolution of cognitive structures and processes. Psychological Review, 84,

260-278.

Hirsch, C. R., Roberts, M. A., Coblentz, D. O., Samide, A. J., & Schoen, H. L. (1979). Geometry.

Glenview, IL: Scott, Foresman.

Hoffer, A. (1983). Van Hiele-based research. In R. Lesh & M. Landau (Eds.), Acquisition of Mathe-

matical Concepts and Processes (pp. 205-227). New York: Academic Press.

Humphreys, L. G., & Parsons, C. R. (1979). Piagetian tasks measure intelligence and intelligence

tests assess cognitive development. Intelligence, 3, 369-382.

Humphreys, L. G., Rich, S. A., & Dewey, T. C. (1985). A Piagetian test of general intelligence.

Developmental Psychology, 21, 872-877.

Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from childhood to adolescence. New

York: Basic Books.

Johnson, P. E. (1967). Some psychological aspects of subject matter structure. Journal of Educational

Psychology, 58, 75-83.

Johnson, P. E. (1969). On the communication of concepts in science. Journal of Educational Psy-

chology, 60, 32-40.

Jurgensen, R. C., Donnelly, A. J., Maier, J. E., & Rising, G. R. (1975).Geometry. Boston: Houghton

Mifflin.

Kruskal, J. B. (1964a). Multidimensional scaling by optimizing goodness of fit to a nonmetric

hypothesis. Psychometrica, 29, 1-27.

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions
94 Structuring Geometric Content

Kruskal, J. B. (1964b). Nonmetric multidimensional scaling: A numerical method. Psychometrica,

29, 115-129.

Lawson, A. E., & Renner, J. E. (1975). Relationship of science subject matter and developmental

levels of learners. Journal of Research in Science Teaching, 12, 347-358.

McDonald, J. L., & Sheehan, D. L. (1983, June). Use of the Longeot test to assess formal opera-

tions: A research summary. Paper presented at the annual symposium of the Jean Piaget Society,

Philadelphia.

Novak, J. D. (1983). The use of concept mapping and knowledge vee mapping with junior high school

students. Science Education, 67, 625-645.

Piaget, J. (1970). Science of education and the psychology of the child. New York: Orion Press.

Piaget, J. (1972). Intellectual evolution from adolescence to adulthood. Human Development, 15, 1-12.

Preece, P. F. W. (1978). Associative structure and the schema of proportionality. Journal of Research

in Science Teaching, 15, 395-399.

Reif, F., & Heller, J. I. (1982). Knowledge structure and problem solving in physics. Educational

Psychologist, 17, 102-127.

Renner, J. W., Stafford, D. G., Lawson, A. E., McKinnon, J. W., Friot, F. E., & Kellogg, D. H. (1976).

Research, teaching, and learning with the Piaget model. Norman: University of Oklahoma Press.

Rohlf, F. J., Kishpaugh, J., & Kirk, D. (1977). Numerical taxonomy system of multivariate statistical

programs [Computer programs and manuals]. Stony Brook: State University of New York at Stony

Brook.

Schoenfeld, A. H., & Herrmann, D. J. (1982). Problem perception and knowledge structure in expert

and novice mathematical problem solvers. Journal of Experimental Psychology: Learning, Memory

& Cognition, 8, 484-494.

Schwab, J. J. (1962). The concept of the structure of a discipline. Educational Record, 43, 197-205.

Shavelson, R. J. (1974). Methods for examining representations of a subject matter structure in a

student's memory. Journal of Research in Science Teaching, 11, 231-249.

Shayer, M., & Adey, P. (1981). Towards a science of science teaching. London: Heinemann.

Sheehan, D. (1970). The effectiveness of concrete and formal instructional procedures with concrete

and formal operational students (Doctoral dissertation, State University of New York at Albany).

Dissertation Abstracts International, 31, 2748A.

Shepard, R. N. (1980). Multidimensional scaling, tree-fitting, and clustering. Science, 210, 390-398.

Silver, E. A. (1982). Problem perception, problem schemata and problem solving. Journal of Mathe-

matical Behavior, 3, 169-181.

Simon, H. A. (1979). Information processing models of cognition. Annual Review of Psychology,

30, 363-396.

Stasz, C., Shavelson, R. J., Cox, D. L., & Moore, C. A. (1976). Field independence and the structur-

ing of knowledge in a social studies minicourse. Journal of Educational Psychology, 68, 550-558.

Strauss, S., & Kroy, M. (1977). The child as logician or methodologist? A critique of formal operations.

Human Development, 20, 102-117.

Tobin, K. G., & Capie, W. (1981). The development and validation of a group test of logical think-

ing. Educational and Psychological Measurement, 41, 413-423.

Thro, M. P. (1978). Relationships between associative structure and content structure of physics con-

cepts. Journal of Educational Psychology, 70, 971-978.

AUTHOR

JANET L. McDONALD, Director of Educational Studies, Associate Professor of Education, Union

College, Schenectady, NY 12308

This content downloaded from 147.8.31.43 on Mon, 29 Feb 2016 21:25:46 UTC
All use subject to JSTOR Terms and Conditions

You might also like