Sie sind auf Seite 1von 15

This article was downloaded by: [de Villiers, Michael]

On: 29 July 2009


Access details: Access Details: [subscription number 913522939]
Publisher Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Canadian Journal of Science, Mathematics and Technology Education


Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t794150240

Calculus Students' Use and Interpretation of Variables: Algebraic vs. Arithmetic


Thinking
Susan S. Gray a; Barbara J. Loud b; Carole P. Sokolowski c
a
University of New England, Biddeford, Maine b Regis College, Weston, Massachusetts c Merrimack College,
North Andover, Massachusetts

Online Publication Date: 01 April 2009

To cite this Article Gray, Susan S., Loud, Barbara J. and Sokolowski, Carole P.(2009)'Calculus Students' Use and Interpretation of
Variables: Algebraic vs. Arithmetic Thinking',Canadian Journal of Science, Mathematics and Technology Education,9:2,59 — 72
To link to this Article: DOI: 10.1080/14926150902873434
URL: http://dx.doi.org/10.1080/14926150902873434

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
CANADIAN JOURNAL OF SCIENCE, MATHEMATICS
AND TECHNOLOGY EDUCATION, 9(2), 59–72, 2009
Copyright 
C OISE
ISSN: 1492-6156 print / 1942-4051 online
DOI: 10.1080/14926150902873434

Calculus Students’ Use and Interpretation of Variables:


Algebraic vs. Arithmetic Thinking
Susan S. Gray
University of New England, Biddeford, Maine

Barbara J. Loud
Regis College, Weston, Massachusetts
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

Carole P. Sokolowski
Merrimack College, North Andover, Massachusetts

Abstract: The ability to use and interpret algebraic variables as generalized numbers and changing
quantities is fundamental to the learning of calculus. This study considers the use of variables in
these advanced ways as a component of algebraic thinking. College introductory calculus students’
(n = 174) written responses to algebra problems requiring the use and interpretation of variables as
changing quantities were examined for evidence of algebraic and arithmetic thinking. A framework
was developed to describe and categorize examples of algebraic, transitional, and arithmetic thinking
reflected in these students’ uses of variables. The extent to which students’ responses showed evidence
of algebraic or arithmetic thinking was quantified and related to their course grades. Only one third
of the responses of these entering calculus students were identified as representative of algebraic
thinking. This study extends previous research by showing that evidence of algebraic thinking in
students’ work was positively related to successful performance in calculus.

Résumé: La capacité d’utiliser et d’interpréter les variables algébriques comme des nombres
généralisés ou des quantités variables est fondamentale pour l’apprentissage du calcul différentiel
et intégral. La présente étude considère une telle utilisation avancée des variables comme une com-
posante de la pensée algébrique. Nous avons analysé les réponses écrites de 174 étudiants dans un
cours universitaire d’introduction au calcul différentiel et intégral, à des problèmes d’algèbre qui
demandaient d’utiliser et d’interpréter certaines variables comme des quantités variables, afin de
cerner les indices d’une pensée algébrique ou arithmétique. Un cadre a été mis au point pour décrire
et catégoriser les exemples de pensée algébrique, transitionnelle et arithmétique, tels que reflétés par
l’utilisation des variables chez ces étudiants. Nous avons analysé dans quelle mesure les réponses des
étudiants pouvaient être considérées comme exemples de pensée algébrique ou arithmétique; cette
mesure a ensuite été quantifiée et mise en relation avec les notes obtenues dans les cours. Seulement
un tiers des réponses des étudiants de première année en calcul se sont avérées des exemples de
pensée algébrique. Cette étude élargit le champ de certaines recherches précédentes, car elle montre

Address correspondence to Carole P. Sokolowski, Mathematics Department, Merrimack College, 315 Turnpike Street,
North Andover, MA 01845. E-mail: carole.sokolowski@merrimack.edu
60 GRAY ET AL.

que les indices de pensée algébrique dans le travail des étudiants sont effectivement liés à une bonne
performance en calcul.

INTRODUCTION

Algebraic variables can be used and interpreted in many ways. Variable is not a single, static
concept that is easily understood. Although this concept is usually introduced to students by
the latter part of middle school, its many and complex uses are revealed only after years of
study. Ideally, the years of algebra instruction and practice in secondary school help students
progress from using variables as representatives of specific unknowns toward using variables in
more advanced ways. The study of calculus, with its fundamental concepts of limit, derivative,
and integral, requires an ability to understand algebraic variables as generalized numbers and as
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

functionally related varying quantities.


Past research indicates that college students who have difficulty using algebraic variables
as varying quantities often have difficulty in their performance in calculus (Carlson, Jacobs,
Coe, Larson, & Hsu, 2002; Gray, Loud, & Sokolowski, 2005; White & Mitchelmore, 1996). In
an investigation of students’ understandings of the concept of derivative, White and Mitchel-
more (1996) posited that their subjects’ difficulties with the derivative were attributable to an
“under-developed concept of variable” (p. 91). Carlson et al. (2002) examined calculus students’
covariational reasoning and concluded that such reasoning is fundamental to the understanding of
calculus. We previously reported that calculus students who had shown an ability to use variables
as varying quantities achieved a mean final course grade of B− (reasonably good performance);
by comparison, students who were categorized as having either a moderate or basic level of un-
derstanding of variables only achieved a mean course grade of D+ (marginal performance; Gray
et al., 2005). Mean grades were reported in that study. Marginal performance was not classified
as successful because, for many majors, a grade of D+ was not sufficient for students to take
the next sequential course. There were students who had used variables at an advanced level
but did not earn high grades in their calculus course. However, in that study, no students who
showed only the most basic level of understanding of variables achieved even marginal grades in
calculus. Although an advanced understanding of variables as varying quantities is not a suffi-
cient condition for student success, it may be a necessary condition. In an effort to analyze what
types of difficulties undergraduates exhibit in this regard, this article describes calculus students’
approaches to algebra problems that require the use or interpretation of variables as generalized
numbers or varying quantities.

THEORETICAL FRAMEWORK

For the present study, an operational-structural (process-object) framework was used to categorize
and describe students’ responses to a selection of five problems from an algebra test (Sokolowski,
1997) adapted from a large-scale British study (Hart, Brown, Kerslake, Küchemann, & Ruddock,
1985; Küchemann, 1981). Many researchers have written about algebraic concepts from an
operational-structural perspective (Dubinsky, 1991; Dubinsky & Harel, 1992; Jacobs, 2002;
Kieran, 1992; Sfard & Linchevski, 1994; Stacey & MacGregor, 2000; Tall et al., 2000; Trigueros
ALGEBRAIC VS. ARITHMETIC THINKING 61

& Ursini, 2003). Sfard and Linchevski (1994) explained the operational-structural framework by
describing mathematical operations, or processes, as series of actions that are ultimately reified
into a cognitive structure, or object, upon which new operations can be performed in order to
form even more complex structures.
In general, an operational viewpoint is characterized by an almost exclusive focus on ma-
nipulating expressions or equations in search of numerical values. Sfard and Linchevski (1994)
described an operational viewpoint as one which takes students from arithmetic to “algebra of a
fixed value” (p. 102), in which students solve an algebraic equation to find the unknown value
of the variable by performing a sequence of operations. Jacobs (2002) reported that secondary
school students studying advanced calculus used a calculational approach when they treated
the “variable as a tool for solving an equation or finding an unknown value” (p. 203). Stacey
and MacGregor (2000) referred to this process of performing operations on equations to obtain
numerical answers as illustrative of arithmetic thinking.
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

By contrast, the ability to work with algebraic expressions or equations more generally and
to interpret these as representations of mathematical relationships without needing to attach
numerical values to the variables is characteristic of a structural perspective. According to Sfard
and Linchevski (1994), a structural viewpoint would be manifested in “functional algebra”
(p. 108), in which students consider an algebraic function as a reified structure on which other
operations can be performed. Similarly, Stacey and MacGregor (2000) referred to algebraic
thinking as that which allows students to view variables, expressions, and equations as structures
of general representation. Students do not need to find numerical referents for the variables
in order to work with them as representations of generalized numbers or covarying quantities.
Several researchers have developed models for describing students’ approaches to working with
variables based on these various interpretations of operational-structural cognitive development
(Carlson et al., 2002; Jacobs, 2002; Trigueros & Ursini, 2003).
For this article, the terms algebraic thinking and arithmetic thinking most closely align with
Stacey and MacGregor’s (2000) descriptions given above. These terms are used to describe
the characteristics of students’ responses to the algebra problems analyzed for this study. To
clarify this terminology more specifically with regard to students’ uses of algebraic variables,
Küchemann’s (1981) catagorizations of the Chelsea Diagnostic Algebra Test items (Hart et al.,
1985) were utilized.
Küchemann (1981) developed four hierarchical levels of students’ interpretations of algebraic
variables. Arithmetic thinking, as defined for the present study, can be viewed as roughly equivalent
to Küchemann’s Levels 1 and 2, which are very basic uses and interpretations of variables.
According to Küchemann, these two levels of using variables are characterized by evaluating,
ignoring, or using variables as objects or labels as early learners of algebra tend to do. For
example, in order to solve a simple equation such as x + 3 = 5, a student could easily find that
the answer is x = 2 by replacing the x with values until the correct value is obtained; he/she could
simply evaluate the variable. This tendency to evaluate variables is quite persistent. Even more
experienced students sometimes use guess-and-check methods inappropriately to solve much
more difficult equations. An example of a problem that allows a student to ignore the variables is:
If q + r = 25, then q + r + 4 = ? Here, the student need only recognize that 4 must also be added
to the 25; the student need not actually work with the variables; they are essentially ignored. The
use of variables as labels is illustrated in an expression such as 5a, for which the interpretation of
“5 apples’ is common, rather than the intended interpretation of “5 times the number of apples.”
62 GRAY ET AL.

This interpretation of variables as labels or objects, rather than as indicators of quantities, has
proved to be very resilient. When students model word problems with algebraic equations, they
often make errors based on an inappropriate use of variables as labels (Clement, Lochhead, &
Monk, 1981; Gray et al., 2005).
Algebraic thinking as defined in this article is evidenced by the use of variables as specific
unknowns, generalized numbers, and varying quantities, which are indicators of Küchemann’s
(1981) Levels 3 and 4, his more advanced levels of understanding. Generally, Level 3 refers to
using variables as specific unknowns. When one solves an equation such as 3x +5 = 4 − 2x,
one is seeking the specific unknown, represented by the variable, that will make the equation a
true statement. The ability to solve equations with the variable appearing on both sides of the
equal sign is an indication, according to some researchers, that students may have moved from
arithmetic to algebraic methods (Filloy & Rojano, 1989; Herscovics & Linchevski, 1994).
Most previous studies examining evidence of algebraic thinking in their subjects were con-
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

ducted with young students who were just beginning to learn algebra (Kieran, 1992; Stacey &
MacGregor, 2000). In the current study, we are concerned with undergraduate calculus students
who have had a number of years of experience with the rules and syntax of algebra. For these
students, merely being able to solve an equation like 3x + 5 = 4 − 2x is not a sufficient indicator
of algebraic thinking, as it may be for younger students. For this reason, we are focusing on the
use and interpretation of variables as generalized numbers and varying quantities as more robust
indicators of algebraic thinking. These two uses of variables are characteristics of Küchemann’s
(1981) Level 4. Variables are used as generalized numbers when they represent entire sets of
numbers. For example, 5 + x = x + 5 represents the commutative property of addition for
real numbers. Students who understand variables as generalized numbers would recognize that
the x in the equation is more than a place-holder for a specific unknown; rather, it implicitly
assumes infinitely many values. Finally, in a function such as y = 3x + 10, two variables, x and
y, are covarying with one other and assume a range of related values. The focus of the present
study is on these advanced uses of variables that are at once the most crucially foundational for
understanding the major concepts of calculus and the most difficult for students to grasp.

GOALS OF THE STUDY

There were three major goals of this study to examine entering calculus students’ responses
to problems that used variables as generalized numbers or varying quantities: first, to classify
students’ responses to these problems as indicators of arithmetic or algebraic thinking; second,
to determine their success rates on these problems with respect to algebraic thinking; and third,
to determine the relationship between students’ course grades and their performance on these
problems.

METHOD

An adaptation (Sokolowski, 1997) of the Chelsea Diagnostic Algebra Test (Hart et al., 1985)
was administered over a period of four semesters on the first day of class to 174 introductory
calculus students at two private liberal arts colleges in New England. The test consisted of 23
ALGEBRAIC VS. ARITHMETIC THINKING 63

TABLE 1
Problems Analyzed

Problem

3 Which is larger, 2n or n+2? Explain.


16 What can you say about c if c + d = 10 and c is less than d?
19a a = b + 3. What happens to a if b is increased by 2?
19b f = 3g + 1. What happens to f if g is increased by 2?
21 If this equation (x + 1)3 + x = 349 is true when x = 6, then what value of x
will make this equation (5x + 1)3 + 5x = 349 true?

questions, with 15 of those having multiple parts. Students were given a maximum of 50 minutes
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

to complete the test without the use of calculators. They received no directions other than the test
questions as stated, and all written work was recorded on the test paper.
For this study, a selection of five problems was analyzed because successful responses to these
problems were likely to illustrate algebraic thinking. Four of the five problems incorporated the
use of variables as generalized numbers or varying quantities. All five problems were chosen
because we expected that they could be easily answered from a structural perspective, thus
indicating algebraic thinking, by students who have advanced to the study of calculus, and also
because evidence of arithmetic thinking would be observable in answers indicative of operational
or calculational approaches. The five problems analyzed are presented in Table 1, numbered as
they were on the algebra test (Hart et al., 1985; Sokolowski, 1997).
All responses were recorded and placed into one of three categories: arithmetic, algebraic,
or transitional. Written characteristics of arithmetic responses included symbolic manipulation,
evaluation of the variable as a single value, or numeric substitution. These responses were
considered to be operational approaches to the problems. Responses were categorized as algebraic
when they showed evidence of using variables to represent multiple referents in generalized
expressions and functional relationships and/or because they reflected a structural perspective in
solving the problem. In addition, there were many responses that were deemed to be transitional.
Responses in the transitional category included evidence of thinking about variables as taking
on more than one value, but they were incomplete or not fully generalized. Consensus was
reached among the researchers regarding criteria and placement of responses into the appropriate
categories.
In addition to the qualitative analysis, responses were given a numeric score of 0, 1, or 2, based
on whether they were categorized as arithmetic, transitional, or algebraic, respectively. A score
of 0 was also used for blanks or for the relatively few responses that could not be categorized
because their meanings could not be interpreted in this framework. The relationship between
these calculus students’ mean final course grades and their numeric scores on these five problems
was also examined.

RESULTS AND DISCUSSION

In this section, the responses to each problem are described and categorized according to the
arithmetic/transitional/algebraic framework as previously defined. Percentages of responses in
64 GRAY ET AL.

each category and the relationship of numeric scores to course grades are also presented and
discussed.

Problem 3

Which is larger, 2n or n + 2? Explain.


Algebraic response: It depends. If n < 2, n + 2 is larger; if n = 2, 2n = n + 2; if n > 2, 2n is larger.

This problem involves the comparison of two expressions, both using the same variable. There
is a need to think of the variable algebraically as taking on a range of values while making this
comparison. Only 15% of the students responded to this problem by writing that it depends on
the value of n and included a good general explanation, either involving inequalities as shown in
the algebraic response above or clearly indicating an understanding of the structure of the answer
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

as pivoting about the number 2.


An additional 36% of the responses were categorized as transitional because they indicated
a recognition that the variable could represent more than one value. Although most of these
responses gave two or three numerical examples to support their conclusion that it depends on
the value of n, they did not identify n = 2 as the structural pivot point of the comparison. Finally,
a few other transitional responses stated that 2n is greater but then gave numerical exceptions.
Forty percent of the responses to Problem 3 were categorized as arithmetic. Almost all of these
stated that 2n is larger. These responses either gave one or more numerical examples to support
this conclusion or gave an explanation stating that “multiplication makes numbers larger.” This
persistent generalization that multiplication makes numbers larger than addition seems to indicate
that these students were thinking of natural numbers as the referents of the variable. They based
their conclusions on the results of testing only one or two values, thus suggesting an arithmetic
approach to this problem.
In general, the vast majority of students used natural numbers to test values to compare 2n and
n + 2. This almost exclusive use of natural numbers may have been prompted by the use of the
letter n as the variable in this problem.

Problem 16

What can you say about c if c + d = 10 and c is less than d?


Algebraic response: c < 5.

In this problem, two variables are covarying and must be compared in the context of one
function under the constraint of an inequality. Thirty-three percent of the students answered this
problem algebraically with c < 5. This problem requires respondents to think of a set of values
covarying with a second set of values. Therefore, the response c < 5 was the only one considered
to be indicative of algebraic thinking.
Twenty-four percent of the responses were classified as transitional. Approximately two thirds
of these were written in inequality form. For example, responses such as 0 ≤ c < 4 or 1 ≤ c ≤ 4
were considered to be insufficient as algebraic responses because of the use of 0 or 1 as the
lower bound and 4, rather than 5, as the upper bound of the inequality. Most of the remaining
responses consisted of a list of integers between 1 and 4, inclusive. Responses of these types were
ALGEBRAIC VS. ARITHMETIC THINKING 65

categorized as transitional because they indicated that students considered a range of values but
did not fully generalize their answers.
Twenty-eight percent of the responses were categorized as arithmetic. These included solving
for c (c = 10 − d or c < 10 − d) or giving a single number value for c. They reflected an
arithmetic view of the problem because students simply performed a symbolic manipulation or
gave a single numerical answer.

Problem 19a

a = b + 3. What happens to a if b is increased by 2?


Algebraic response: a increases by 2.

Two variables are covarying in an additive functional relationship in this problem. Sixty-seven
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

percent of the responses to this problem were categorized as algebraic. An algebraic response
indicates that a quantifiable change in a is the result of a change in b in the functional relationship.
Although this problem has an algebraic response rate that is much higher than the others, it is so
closely aligned with Problem 19b that its results help to inform the interpretation of the results
of the subsequent problem.
Eight percent of the responses stated that a increases. These responses were categorized as
transitional, indicating that students recognized that there was a connection between increasing
the value of b and increasing the value of a; however, they did not specify the magnitude of the
resulting increase. An alternative interpretation of this response is that students simply answered
mechanically, thinking that an increase in one variable automatically would produce an increase
in the other.
Finally, there were nineteen percent of the responses that were reflective of arithmetic thinking.
Typical responses in this category were a = b + 5 or a = (b + 2) + 3. Students were not increasing
the value of b by 2 but were representing their interpretation of the procedure of adding 2 to b.
These responses and others, such as 2b + 3 or a = 5, suggest that students were using the 2
to perform a computation instead of recognizing that the increase by 2 in one variable causes a
corresponding change in the other variable.

Problem 19b

f = 3g + 1. What happens to f if g is increased by 2?


Algebraic response: f increases by 6.

Here, two variables are covarying in a multiplicative and additive functional relationship.
Although this problem is similar to the previous problem, the results were quite different. Only
30% of the students gave an algebraic response to this problem. Most had only the answer and
nothing else written. However, the written work accompanying 12 of these responses showed that
students tested a few values for g in the given function and then concluded that f increases by 6.
The systematic testing of values indicates that students recognized the functional relationship
of the variables in this problem. The act of generalizing the results to produce the answer was
considered to be indicative of algebraic thinking.
66 GRAY ET AL.

Fourteen percent of the responses to this problem stated that f increases. As in Problem 19a,
these responses were included in the transitional category because they illustrated a sense of
covariation between the two increasing variables but did not specify the amount of the increase.
Another 40% of the responses to Problem 19b were categorized as arithmetic. Most of the
responses in this category stated that f doubles, increases by 2, or increases by some other number.
Some students appeared to interpret the expression “increased by 2” as doubling. However, 13 of
the 18 students who responded in this way did not state that a doubles in the previous problem.
The multiplicative nature of Problem 19b may have changed their interpretation of “increased by
2” into a doubling effect. There were other computational responses that indicated an arithmetic
view, such as f = 7, f = 5g +1, or f = 6g + 1. Here, as in Problem 19a, students performed a
calculation with the 2 or substituted the 2 for g. The relationship between two variables in which
one variable changes with respect to changes in the other variable was not apparent in these
responses.
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

Problem 21

If this equation (x + 1)3 + x = 349 is true when x = 6, then what value of x will make this equation
(5x + 1)3 + 5x = 349 true?
Algebraic response: 6/5 or 1.2.

Only one variable is used in this problem. It is used as a specific unknown in both equations,
but its value is different in the context of the second equation. Thus, it is necessary to see the
structural similarity between the two equations and to compare the value of the variable in one
equation with that in the other in order to answer this question from an algebraic perspective.
Only 27% of the students responded algebraically to this problem. Although a few students
wrote 5x = 6 on their papers, the majority showed just the answer, 6/5 or 1.2. To arrive at this
conclusion, it would be necessary to identify the structural similarity between the two equations,
deduce that 5x = 6, and solve this equation for the new value of x. This approach is an indication of
algebraic thinking. It is unlikely that students would arrive at this correct answer by an arithmetic
method of testing values.
Thirty-one percent of the responses showed attempts to expand the cubic expression or to
perform numeric substitutions or solve for x. Some other typical incorrect numerical answers
were: 30, 1, 2, 3, 4, or 6. Because none of these had anything else written on the test paper, they
may have been indications of attempts at numeric substitution or guess and check methods. All
of these indicate arithmetic or calculational approaches to the problem.
There were no responses to this problem that were categorized as transitional, but 42% of the
students did not give a response. Although this problem was near the end of the test, most of these
students completed the following few problems, so the blanks do not necessarily signal a lack of
time. It is more likely that students saw this as a complex problem and decided to skip over it.

Summary of Response Categories

Table 2 summarizes the most common types of responses that were identified as indicators of
arithmetic, transitional, or algebraic thinking according to the categorization framework used in
this study. Examples of responses that reflected arithmetic thinking are those that showed evidence
of numeric substitution or rote symbolic manipulation. Transitional responses in general showed
ALGEBRAIC VS. ARITHMETIC THINKING 67

TABLE 2
Examples of Responses Indicative of Algebraic, Transitional, or Arithmetic Thinking
Problem Algebraic Transitional Arithmetic
Problem 3 It depends. It depends (with 2 or 2n (with numerical
Which is larger, 2n or If n < 2, n + 2 > 2n; 3 numerical examples examples)
n + 2? Explain. if n = 2, n + 2 = 2n; for justification). 2n, multiplication
if n > 2, n + 2 < 2n 2n, with numerical makes it larger.
exceptions or n+2
qualifications
Problem 16 c<5 0<c<4 c = 10 - d
What can you say 0≤c<5 c < 10 - d
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

about c if c + d = 10 c≤4 c equals a single


and c < d? c = 1, 2, 3, 4 integer.
Problem 19a a increases by 2. a increases. b+5
a = b + 3. What (b + 2) + 3
happens to a if b is 2b + 3
increased by 2? a=5
a+2
Problem 19b f increases by 6. f increases. Increases by 2, 7, or
f = 3g + 1. What some other number.
happens to f if g is f doubles.
increased by 2? f=7
f = 5g + 1
f = 6g + 1
Problem 21 x = 6/5 or 1.2 No examples were Attempts to solve
If (x + 1) + x = 349,
3
apparent in this for x or substitute
x = 6; category. numbers
If (5x + 1) + 5x = 349,
3

x =?

some indication of considering more than one value for the variables. Algebraic responses showed
evidence of interpreting the variables as representative of covarying quantities or of taking a
structural approach to solving the problem.
The percentages of responses in each category are given in Table 3. Responses identified in
the table as “Uncategorized” were incomplete answers or those whose intent or meaning could
68 GRAY ET AL.

TABLE 3
Percentages of Responses in Each Category

Problem Number Algebraic Transitional Arithmetic Uncategorized Blank

3 15 36 40 9 1
16 33 24 28 5 10
19a 67 8 19 3 2
19b 30 14 40 7 9
21 27 0 31 0 42

Note. Percentages may not sum to 100% horizontally due to round-off error.

not be determined within the context of the problem (e.g., for Problem 3, “n can be anything”;
for Problem 16, “c does not equal 5”; for Problem 19a, “a becomes a − 2”; for Problem 19b,
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

“subtracted by 2”).
Percentages of responses in each category varied among the five problems and were highly
contextual according to problem type. Of primary concern is the evidence that less than one third
of these calculus students’ responses to four out of five of these relatively simple problems were
indicative of algebraic thinking as defined above.

Comparison of Mean Grades and Numeric Scores

Numeric scores ranged from 0 to 10 and were determined by assigning values of 0, 1, or 2


to arithmetic, transitional, or algebraic responses, respectively. Course grades were represented
with a scale ranging from 0 to 4, indicating the following performance descriptors: 4.0, superior;
3.0, good; 2.0, fair; 1.0, minimally passing; 0, fail. Students’ numeric scores were compared
to their mean final course grades in calculus. Figure 1 graphically illustrates the relationship,

FIGURE 1 Numeric Scores vs. Mean Course Grades in Calculus


ALGEBRAIC VS. ARITHMETIC THINKING 69

TABLE 4
Distribution of Numeric Scores and Mean Grades

Score 0, 1 2, 3 4, 5 6, 7 8, 9, 10

Mean grade 1.5 1.6 2.5 2.8 2.9


n 27 41 54 30 22

showing that the group of students that was most successful with respect to algebraic think-
ing (scores of 8, 9, or 10) earned more than a full grade point higher than those who were
least successful (scores of 0, 1, 2, or 3). Moreover, these low scores represent, almost ex-
clusively, responses that were categorized as arithmetic. Scores of 4 or greater were achieved
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

primarily by students whose responses to at least two of the problems were categorized as alge-
braic. This break between scores of 2, 3 and 4, 5 is where the largest increase in mean grades
occurs.
To make the comparison of numeric scores with mean grades, the scores 0 through 10 were
placed into five groupings of two score points in each group, starting with zero. Because there
were only four scores of 10, these were combined with the group of scores including 8 and 9.
The resulting frequencies formed an approximately normal distribution. These frequencies, along
with the mean grades for each group, are recorded in Table 4. It is noteworthy that fewer than
13% of these calculus students answered four or five problems algebraically (scores of 8, 9, or
10), and 39% answered only one or no problems algebraically (scores of 0, 1, 2, or 3).

CONCLUSIONS AND IMPLICATIONS

The present study created a framework to describe some of the qualitative characteristics of
algebraic and arithmetic thinking reflected in college students’ uses of variables. In addition, the
extent to which students’ responses showed evidence of algebraic or arithmetic thinking was
quantified and related to their introductory calculus course grades.
The majority of students in this study had difficulty using variables as generalized numbers
and varying quantities to answer these five problems algebraically. Although we thought these
problems should have been relatively simple for entering calculus students, they proved to be
quite challenging. Based on the arithmetic/transitional/algebraic framework of using variables
employed in this article, only 34% of all the responses were categorized as representative of
algebraic thinking, compared to the 31% that were indicative of arithmetic ways of thinking. In
fact, only 4 of the 174 students answered all five problems algebraically. Many of the arithmetic
responses to Problems 16 and 21 showed the tendency to manipulate one of the variables in order
to “solve” for the other. Correctly performing operations with algebraic variables and symbols is
not in itself a reflection of algebraic thinking. Rather, algebraic thinking, as defined for this study,
requires a structural perspective wherein variables are interpreted as representative of changing
quantities.
The results of the present study align with the growing body of evidence documenting college
students’ difficulties using variables as varying quantities. Trigueros and Ursini (2003) found that
very few first-year undergraduates who had not taken calculus were able to interpret variables
70 GRAY ET AL.

as varying quantities in relational situations. Although it may be assumed that calculus students
would be able to interpret variables as varying quantities, previous studies report evidence to the
contrary and suggest that students’ abilities to interpret variables have an impact on their success
in calculus. For example, Jacobs (2002) found that secondary school advanced calculus students
who had difficulty interpreting variables as covarying quantities were also likely to have difficulty
understanding the calculus concepts of limit and derivative. Moreover, completion of a semester
of calculus may not be sufficient to fully develop students’ understandings of variables as varying
quantities. Carlson et al. (2002) reported that undergraduates studying second-semester calculus
had considerable difficulty with tasks that required reasoning with covarying quantities in relation
to rates of change.
The present study extends previous research by documenting a relationship between under-
graduate calculus students’ uses of variables as reflective of algebraic or arithmetic thinking and
their performance in calculus. Students’ numeric scores on these five problems were positively
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

related to their mean final grades in introductory calculus. There are a myriad of factors that
contribute to course grades. The results of this study imply that facility in using variables alge-
braically as varying quantities may be one of those factors for calculus. These findings, along
with those of other researchers, indicate that there is a need to maintain an ongoing instructional
goal in calculus and precalculus to develop an understanding of the concept of variable in all its
complexities.
In addition to students’ strong tendencies to approach problems arithmetically, this study re-
vealed numerous examples of questionable use of mathematical symbolism, particularly related to
inequality notation. Many students gave responses suggesting that they were using the inequality
symbol to refer to positive integers rather than to sets of real numbers. Evidence from responses
to Problems 3 and 16 indicate that, when working with situations of inequality, these students
rarely appeared to consider real number domains for the variables. Students’ seemingly default
thinking about referents of variables as being integers or natural numbers could be detrimental
to their understanding of the major concepts in calculus, in particular, the concepts of limit and
continuity. However, it is possible that students were influenced by the use of the variable n
in Problem 3 because the letter n is often used in mathematics to indicate an integer. It is also
possible that students did not spontaneously think of using the real number domain in the context
of these problems without being prompted to do so.
The primary goals of this research were to document the ways that students responded to
problems requiring the use and interpretation of variables as changing quantities; to record the
frequencies and types of responses indicative of arithmetic, transitional, and algebraic thinking;
and to determine whether success rates on these problems with respect to algebraic thinking were
related to students’ grade performance in calculus. We decided to collect data from as large a
sample as was feasible in order to obtain a broad base for the range of possible responses from
undergraduates at the beginning of their study of calculus. Although the present study indicates
that there was a strong tendency to use an arithmetic approach to answer these five problems, the
results are based on observations of students’ written responses and do not provide evidence
that students were unable to reason structurally. They may have been using numerical methods
that were more familiar, intuitive, or convincing to them than were more generalized algebraic
approaches. The observations and conclusions reported in this study are based upon the authors’
interpretations and could be further confirmed, extended, or modified by expanding the variety
of problems and/or conducting interviews with respondents.
ALGEBRAIC VS. ARITHMETIC THINKING 71

The findings of this study support the viewpoint that instruction in courses prior to calculus
should include explicit attention to the many different uses of variables as a way to foster
development of algebraic thinking. Furthermore, the results of this study suggest that calculus
instruction should continue to emphasize the differing uses of variables in various contexts and
strive to develop students’ conceptions of variables as changing and covarying quantities. When
students are attempting to learn the concepts of limit, derivative, integral, and the connections
among these concepts crystallized in the fundamental theorem of calculus, they must possess a
robust and flexible view of all aspects of the changing quantities that form the foundation of their
mathematical studies.

ACKNOWLEDGEMENTS
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

Portions of this research were presented at the 2007 annual conference of Research in Undergradu-
ate Mathematics Education (RUME) in Phoenix, Arizona. The authors express their appreciation
to Professor Annie Selden for her helpful and insightful comments on an earlier draft of this
article.

REFERENCES

Carlson, M., Jacobs, S., Coe, E., Larson, S., & Hsu, E. (2002). Applying covariational reasoning while modeling dynamic
events: A framework and a study. Journal for Research in Mathematics Education, 33(2), 352–378.
Clement, J., Lochhead, J., & Monk, G. (1981). Translation difficulties in learning mathematics. American Mathematical
Monthly, 88(4), 286–290.
Dubinsky, E. (1991). Reflective abstraction in advanced mathematical thinking. In D. O. Tall (Ed.), Advanced mathemat-
ical thinking (pp. 95–123). Dordrecht: Kluwer.
Dubinsky, E., & Harel, G. (1992). The nature of the process conception of function. In G. Harel & E. Dubinsky (Eds.),
The concept of function: Aspects of epistemology and pedagogy, MAA Notes (Vol. 25, pp. 85–106). Washington, DC:
Mathematical Association of America.
Filloy, E., & Rojano, T. (1989). Solving equations: The transition from arithmetic to algebra. For the Learning of
Mathematics, 9(2), 19–25.
Gray, S. S., Loud, B. J., & Sokolowski, C. P. (2005). Undergraduates’ errors in using and interpreting algebraic variables:
A comparative study. In G. M. Lloyd, M. R. Wilson, J. L. Wilkins, & S. L. Behm (Eds.), Proceedings of the 27th
Annual Meeting of the North American Chapter of the International Group for the Psychology of Mathematics
Education [CD-ROM]. Eugene, OR: All Academic.
Hart, K., Brown, M., Kerslake, D., Küchemann, D., & Ruddock, G. (1985). Chelsea diagnostic mathematics tests:
Teacher’s guide. Windsor, UK: NFER-NELSON.
Herscovics, N., & Linchevski, L. (1994). A cognitive gap between arithmetic and algebra. Educational Studies in
Mathematics, 27, 59–78.
Jacobs, S. (2002). Advanced placement BC calculus students’ ways of thinking about variable. Dissertation Abstracts
International, 63 (02A), 484. (UMI No. AAI3043823)
Kieran, C. (1992). The learning and teaching of school algebra. In D. Grouws (Ed.), Handbook of research on mathematics
teaching and learning (pp. 398–419). New York: Macmillan.
Küchemann, D. (1981). Algebra. In K. M. Hart (Ed.), Children’s understanding of mathematics: 11–16 (pp. 102–119).
London: John Murray.
Sfard, A., & Linchevski, L. (1994). The gains and pitfalls of reification: The case of algebra. Educational Studies in
Mathematics, 26, 191–228.
72 GRAY ET AL.

Sokolowski, C. P. (1997). An investigation of college students’ conceptions of variable in linear inequality. Dissertation
Abstracts International, 58 (02A), 0407. (UMI No. AAG9723724)
Stacey, K., & MacGregor, M. (2000). Learning the algebraic method of solving problems. Journal of Mathematical
Behavior, 18(2), 149–167.
Tall, D., Gray, E., Bin Ali, M., Crowley, L., DeMarois, P., McGowan, M., et al. (2000). Symbols and the bifurcation
between procedural and conceptual thinking. Canadian Journal of Mathematics, Science, and Technology Education,
1(1), 81–104.
Trigueros, M., & Ursini, S. (2003). First-year undergraduates’ difficulties in working with different uses of variable. In A.
Selden, E. Dubinsky, G. Harel, & F. Hitt (Eds.), Research in collegiate mathematics education V (Vol. 12, pp. 1–29).
Providence, RI: American Mathematical Society.
White, P., & Mitchelmore, M. (1996). Conceptual knowledge in introductory calculus. Journal for Research in Mathe-
matics Education, 27(1), 79–95.
Downloaded By: [de Villiers, Michael] At: 21:45 29 July 2009

Das könnte Ihnen auch gefallen