Students' Proof Schemes for Proving and Disproving of

Propositions

KoSze Lee

University of North Florida

Corresponding author:
KoSze Lee. Department of Foundations and Secondary Education, University of North Florida. 1

UNF Drive, Building 57, Room 2326. Jacksonville, FL, 32224, USA. 1-904-620-1775 (Office).

Email: [email protected]
 2

Abstract

Students’ mathematical proving remains crucial for classrooms practices that adopt reformed

curriculum approaches or common core standards (CCSSO, 2010). Teachers often face

challenges in teaching mathematical proving meaningfully to students. This may be due to their

limited knowledge of interpreting students’ proficiencies for constructing proofs as connected

deductive inferences (Stylianides, 2007). Classification of students' proof schemes provides

teachers with the necessary knowledge to obtain insights into students’ proving (Balacheff, 1988;

Harel & Sowder, 1998). However, little attention was given to the cognitive nuances in students’

inferences and proving attempts, students’ more sophisticated use of a random selection of

examples in the “Empirical” proofs (Balacheff, 1988; Harel & Sowder, 1998). Moreover, these

classifications were developed for proving mathematically true propositions (Balacheff, 1988;

Harel & Sowder, 1998); students’ schemes of disproving mathematically false propositions using

inferences and counterexamples were hardly investigated. This study presents two classifications

of proof schemes, each for students’ proving and disproving of mathematical propositions. The

refined scheme elaborated the transition phase between proofs based on examples and deductive

proofs, and further suggested the role played by students’ mathematical knowledge and use of

examples. Implications for research of students’ proof schemes are discussed.

Keywords: Proof and Proving, Proof Schemes, Mathematical Reasoning

 3

Students’ mathematical proving and deductive reasoning remain crucial for classrooms

practices that adopt reformed curriculum approaches or Common Core State Standards (CCSSO,

2010; NCTM, 2000). Its emphases on students’ conjecturing of mathematical propositions and

subsequent processes of proving by deductive inferences or disproving by counterexamples are

complementary to each other. A number of high school standards in the Geometry, Functions,

and Algebra strands have also provided clear expectations about students’ proficiencies to prove

or disprove mathematical theorems, properties, equations, and formulae. However, teachers often

face challenges in interpreting students’ proficiencies in mathematical proving from their written

proofs, and specifically the degree of soundness in students’ deductive reasoning (Knuth, 2002).

This may be due to teachers’ limited knowledge of mathematical proof or interpreting students’

proficiencies for constructing proofs as connected deductive inferences (Stylianides, 2007).

Classification of students' proof schemes provides teachers with the necessary knowledge to

obtain insights into students’ proving (Balacheff, 1988; Coe & Ruthven, 1994; Harel & Sowder,

1998; Healy & Hoyles, 2000).

Past research paid little attention to the cognitive nuances in students’ inferences and

proving attempts. For instance, students might prove a mathematical proposition unsuccessfully

based on the evidence of a single example or a random selection of examples. Though both

approaches exhibited mathematical and deductive reasoning errors and were classified under the

same category of “empirical scheme” (Balacheff, 1988; Harel & Sowder, 1998), the latter

approach indicated more sophisticated mathematical reasoning and knowledge from a cognitive

and epistemological perspective.

Moreover, these classifications were developed for students proving of mathematically

true propositions (Balacheff, 1988; Harel & Sowder, 1998); how students used inferences and
 4

counterexamples to disprove mathematically false propositions were hardly investigated.

Consequently, little is known about students’ proof schemes underlying their disproving of

mathematical propositions.

This proposal aims to study students’ mathematical proving and disproving of

mathematical propositions and how they use inferences and counterexamples in mathematical

proving. Specifically, the study aims to establish a classification of proof schemes, each for

students’ proving of mathematically true and false propositions using their written proofs.

Theoretical Framework

Mathematical Proving

Mathematical proving is defined as the search for examples and counterexamples to a

proposition in order to determine whether the proposition is true (Durand-Guerrier, 2003). The

extent to which students engage in deductive inferences during this process, either to consider

the proposition or the chain of inferences, indicates the proficiency levels of their proving. In this

regard, inductive and deductive inferential processes in mathematical proving are two ends of a

cognitive continuum between which students may traverse to make logical conclusions about the

proposition.

Proof Construction in Mathematical Proving and Proof Scheme

Incorporating Harel & Sowder’s (1998) notion, proof construction is defined as the

process of removing one’s doubt about a mathematical proposition through their search for and

reasoning of possible examples and counterexamples. Students use their mathematical

knowledge to construct arguments that connect the given mathematical premises to the

conjectured mathematical conclusions (Stylianides, 2007). It focuses on students’ ability to use

their mathematical knowledge to construct arguments that connect the given mathematical
 5

premises to the conjectured mathematical conclusions (Stylianides, 2007) and is different from

evaluating the logical validity of a proof (Selden & Selden, 2003; Weber & Alcock, 2004).

Depending on the mathematical proposition, there are two types of proof constructions:

Deductive-proof Construction for true propositions and Proof-by-counterexample Construction

for false propositions. Accordingly, a student’s proof scheme is the cognitive scheme underlying

one’s proof construction.

The research questions pursued in this proposal are:

1) What are students’ proof schemes underlying their Deductive-proof Construction?

2) What are students’ proof schemes underlying their Proof-by-counterexample

Construction?

Method

The data of this study came from proofs constructed by 60 Singapore Secondary Three

(equivalent to U.S. 9th Grade) student participants in a study about logical reasoning and

mathematical proving. Each student attempted four Proof Construction tasks (two Deductive-

proof construction and two Proof-by-counterexample) on two separate sessions during after-

school hours. During each session, students were allowed to use calculators and writing paper.

To minimize the impact due to a lack of relevant mathematical knowledge, they were

encouraged to clarify the mathematical terms and definitions used in the tasks with the

researcher.

The four tasks consisted of one true proposition and one false proposition in Elementary

Number theory, and one true and one false proposition in Quadratics. Students had learnt in class

how to apply these sets of knowledge to solve typical mathematics problems but not for Proof

Construction tasks. Each task described a mathematical situation and posed a mathematical
 6

implication, posed in the form of “If P then Q” since student reasoning and proving of

mathematical implications is central to the learning of mathematics and provide productive

opportunities for using counterexamples (for a disciplinary-based narrative and empirical

findings, see Durand-Guerrier, 2003). Students were asked to decide whether it was true or false

and to justify their conclusions using the most convincing argument by either proving

mathematically true implications for Deductive-proof Construction or disproving mathematically

false implications for Proof-by-counterexample Construction.

Following are a sample task for Proof-by-counterexample Construction:

(Proof-by-counterexample Construction) “A prime number is a whole number that has

exactly two factors, 1 and the number itself (Note that 1 is not a prime number since it

has only one factor). Two positive whole numbers, which may or may not be prime

numbers, are added together. Decide whether the following rule is true or false: If two

prime numbers are added together, then the sum is an even number.

Justify why your conclusion must be true or false using the most convincing argument.”

To elicit a wide collection of student proof schemes, two different sets of tasks containing

similar implications were phrased in a similar style and given to each student (see Table 1

below). Half of the students attempted the first set in the first occasion and the second set in the

second occasion while the other half attempted the sets in reverse order.

<Insert Table 1 here>

Coding and Analysis

A total of 480 students’ written proofs were analyzed: 120 true and false implications

each related to elementary number theory and quadratics. Two coding classifications were

developed for analyzing students’ productive use of inferences in Deductive-proof Construction

 7

and Proof-by-counterexample Construction. Both sets of coding schemes were initially

developed as broad categories based on past research (e.g., Balacheff, 1988; Harel & Sowder,

1998). Thus classification of external conviction schemes, empirical-based schemes, and analytic

schemes were used for coding students’ works of Deductive-proof Construction, while only

some of the classifications, such as empirical-based schemes, were adapted for coding students’

works of Proof-by-counterexample Construction wherever applicable.

Subsequently, the researcher made finer distinctions within each type of schemes based

on the cognitive nuances shown in students’ proofs. For instance, some students’ proofs

considered cases of the implication for which deductive reasoning was incomplete and this was

classified as a new category of proof schemes different from an empirical-based or analytic

scheme. The classification of schemes assumed a hierarchical ordering of the extent of students'

proficiencies in their use of deductive inferences that echoes contemporary views of

mathematical proving (Balacheff, 1988; Coe & Ruthven, 1994; Durand-Guerrier, 2008;

Stylianides & Stylianides, 2009). In the event when students’ works demonstrated two or more

different types of approaches, the approach that corresponded to higher-level of sophisticated use

of inferences would be coded and classified as that level. A second coder, who was a high school

mathematics teacher, then coded each proof using the refined classification. The inter-rater

agreement ratings for coding both Deductive-proof Construction and Proof-by-counterexample

Construction proofs were 85.4%.

Findings

Deductive-proof Construction Schemes

Students' Deductive-proof Construction were distinguished into seven levels (labeled

level 0 to 6 in an ascending order). In general, the levels can be described as a progression

 8

through four phases: attempts that failed to relate the antecedent with the consequent (level 0),

proofs that were based on examples or logical reasoning errors (levels 1 to 2), proofs that were

based on incomplete deductive inferences (levels 3 to 4), i.e., “missing a step” (Lin, 2005), and

proofs that were based on coherent deductive inferences (levels 5 to 6). Table 2 below showed

the seven levels in the coding scheme.

<Insert Table 2 here>

At level 0, some proof schemes resembled “External Conviction” (Harel & Sowder,

1998) while others cited irrelevant mathematical properties that did not explain the conditional

relationship of the implication. At both levels 1 and 2, proof schemes exhibited little or no

deductive inferences and used only examples. Typically, students generated a few examples that

verified the implication, which resembled the empirical-based schemes (Harel & Sowder, 1998).

Some students might relate the antecedent to the consequent via inference of irrelevant property

or erroneous logical reasoning (empirical-based inferences or invalid form of logical reasoning),

which only qualified their responses as level 1.

Proof schemes at level 2 was distinctively characterized by students’ more sophisticated

use of examples and productive inferences. Students might consider a mathematical example as a

“crucial experiment” for testing the implication (Balacheff, 1988; Chazan, 1993). Alternatively,

students generated examples to verify the implication by make strategic choices based on either

random or case-based sampling (see Fig. 1 below). Some students inferred mathematical

properties from a set of examples and use them to further justify the implication.

<Insert Figure 1 here>

Students’ proof schemes at both levels 3 and 4 exhibited productive but incomplete chain

of deductive inferences. The distinction between them was determined by whether one case of
 9

the antecedent or one or two deductive inferences were skipped in students’ proofs. For example,

students proved the implication for the case of even numbers but not the odd numbers were

classified as level 3 (see Fig. 2 below).

<Insert Figure 2 here>

At level 4, students constructed a chain of inferences that was almost entirely deductive

or complete. Distinction between levels 4 and 5 was based on whether the proof might have one

or two inferences absent, or could be interpreted as non-deductive or had minor errors. That is,

the proof is “missing a step” in making a chain of deductive inferences (Lin, 2005). For instance,

“3 times an odd number x will give an odd number” was not substantiated by “product of two

odd numbers is an odd number” and might be interpreted as an inductive inference. Minor errors

were either due to writing errors or lack of logical organization in sequencing the chain of

inferences. In the figure (Fig. 3) below, the student missed a step in justifying the conclusion

from the inferences of “the y-axis would touch on either part of the curve below the x-axis.”

<Insert Figure 3 here>

Proof Schemes at levels 5 and 6 exhibited logically valid and coherent chain of deductive

inferences, with the latter level constructing a concise proof using mathematical symbols and

notations (Boero, 1999; Miyazaki, 2000). In Figure 4 (level 6 scheme) below, the student

formulated the sum of the numbers as a product of two numbers and considered the case of the

number being odd and even. For each case, the student was able to justify why the product would

result in an even number and thus prove the implication.

<Insert Figure 4 here>

The distribution of students’ proof schemes across the levels by the type of tasks is

shown below (Table 3). More students were exhibiting higher levels of proof schemes for the
 10

Elementary Number Theory tasks than the Quadratics tasks. Majority of the students were

struggling with relating the antecedent and the consequent and using the examples in a way that

exhibited strategic choices or to generate productive inferences about the implication.

<Insert Table 3 here>

For the tasks related to Number Theory, majority of the proofs were classified as levels 1

and 2, which indicated prevalent use of examples to verify the implication or more productive

ways. There were also a number of proofs in which deductive inferences were used and were

thus classified as levels 3 and 4. For the tasks related to Quadratics, majority of the proofs were

classified as levels 0 to 1, suggesting that the tasks were challenging and students struggled to

find sophisticated ways to use examples productively. Across both topics, there were very few

proofs being classified as levels 5 and 6 because few students were able to prove the implications

in rigorously when making deductive inferences. Only two of the proofs related to Number

Theory were at level 6, which indicated that constructing formal deductive proofs using

mathematical symbols is challenging, even for students who were proficient in applying

formulas, graphical properties, and manipulating algebraic symbols related to Quadratics.

Proof-by- counterexample Construction Schemes

Students’ Proof-by-Counterexample Construction were distinguished into six levels

(labeled level 0 to 5 in an ascending order). In general, the levels can be described as a

progression through four phases, of which the first two phases resembled their counterparts in

Deductive-proof Construction schemes: attempts that failed to relate the antecedent with the

consequent (level 0), proofs that were based on examples or logical reasoning errors (levels 1 to

2), proofs that were based on deductive inferences but justify the implication incorrectly due to

misconceptions (level 3), and proofs that were based on coherent deductive inferences using
 11

constructed counterexamples. Table 4 below showed the six levels of the coding scheme in

hierarchical order. Apart from level 0, the hierarchical levels posited a general distinction

between proofs which considered the implications using examples (coded as levels 1 to 2) and

proofs which considered the implication using deductive inferences but not possible

counterexamples (levels 3), and proofs which indicate logically valid Proof-by-counterexamples

had been constructed by students (levels 4 to 5).

<Insert Table 4>

At level 0, the proof schemes were similar to those of Deductive-proof construction.

Proof schemes at both level 1 and 2 were also similar to their counterparts in Deductive-proof

Construction. At level 1, students used example or made logically invalid inferences to justify

the implication and exhibited little or no use of deductive inferences during the process. This is

akin to incorrect empirical proof attempts to prove a false implication (Ko & Knuth, 2009). At

level 2, students exhibited sophisticated use of the examples by inferring properties from the

examples to verify the implication.

At level 3, students' proof schemes exhibited productive use of deductive inferences but

limited consideration of possible counterexamples. For cases of implications or examples they

had identified, they were able to make deductive inferences to prove the implication. However,

students did not consider a subset of mathematical objects (e.g., asserting that all prime numbers

are odd numbers and did not consider that 2 is the only even prime number), which would

qualify as counterexamples to the implication. Figure 5 below shows such an instance of

students’ proof of deciding whether the implication “If two prime numbers are multiplied

together, then the product is an odd number” was true.

<insert Figure 5 here>

 12

The student started considering the set of prime numbers based on the end-digits of the

numbers but incorrectly asserted that all prime numbers were odd numbers. She did not consider

the prime number “2”. Next she deduced that when two prime numbers with odd end-digits were

multiplied together, the resulting end-digit was also an odd number. Her conclusion would have

been logically valid if not for the omission of the case of prime number 2.

For proof schemes at levels 4 and 5, students had successfully constructed one or more

counterexamples to falsify the implication. Proof schemes at level 4 exhibited student

falsification of the implication based on one specific counterexample. In contrast, proof schemes

at level 5 exhibited students’ consideration of a generic counterexample that characterized a

collection of counterexamples by a common property (Hoyles & Küchemann, 2002). In the

figure below (Fig. 6), the student concluded that “anything multiplying with 2” would constitute

a collection of counterexamples.

<Insert Figure 6 here>

The distribution of students’ proof schemes across the levels by the type of tasks is

shown below (Table 5). Note that more students were exhibiting higher levels of proof schemes

for the Elementary Number Theory tasks than the Quadratics tasks.

<Insert Table 5 here>

For the tasks related to Number Theory, majority of the proofs were classified as level 4

or 5, which indicated success in using counterexamples for their Proof-by-counterexample

construction and that the task was less challenging for the students. Moreover, around one third

or more of the proofs were using generic counterexamples. For the tasks related to Quadratics,

majority of the proofs were classified as level 0 to 1 in both occasions. This suggested that the

tasks were probably too challenging for many students. They were struggling with relating the
 13

antecedent and the consequent concerning property of the quadratic functions though they were

able to relate graphing and factoring of quadratic equations. Interestingly, only two written

proofs were classified as level 2 for the topic of Quadratics.

Discussion

This study suggested that a classification of students’ proof schemes, based on the extent

of their search and reasoning of examples and counterexamples and their use of deductive

inferences, revealed finer cognitive nuances in proof construction. In this study, students’ proof

schemes underlying Deductive-proof Construction not only bear resemblance to past findings

about proof schemes which were characterized as “External conviction,” “Naive empiricism,”

“Crucial experiment,” and “Analytical” (Balacheff, 1988; Harel & Sowder, 1998) but also

refined the hierarchical classification. In comparison to college students, this study suggested

that high school students exhibited proof schemes of an intermediate phase between empirical-

based and analytical-based schemes. They consisted of two finer levels (levels 3 and 4) that

characterized a mix of deductive inferences with reasoning errors or inductive inferences.

Cognitive nuances underlying students’ use of examples were also evident. They constructed

examples based on strategic choices or make inferences from examples to justify a proposition.

More significantly, this study proposed a classification of students’ proof schemes for

understanding their Proof-by-counterexample Construction. The schemes revealed additional

cognitive nuances in students’ disproving of a false proposition beyond a mere dichotomy of

whether a proposition can be falsified by a counterexample. From a logical perspective, it

seemed asymmetrical to have a single instance of counterexample to disprove a proposition but

require every single instance to satisfy a proposition using deductive inferences. However, the

cognitive nuances underlying students’ proof schemes revealed an intermediate phase (level 3) of
 14

productive inferential processes in between the failure and success in disproving a proposition.

Specifically, students were able to use deductive inferences effectively to prove a proposition for

a smaller subset of mathematical objects, though they might have overlooked some cases for

counterexamples.

In addition, the distinction between levels 4 and 5 in students’ proof schemes was

meaningful from cognitive and epistemological perspectives. Disproving a proposition by one

counterexample (level 4) or more counterexamples (level 5) does not make any logical

difference but does demonstrate that proofs present epistemic values as well as logical values for

students (Duval, 1993). In the course of disproving by a larger set of counterexamples, students

not only justified their conviction about the mathematical propositions but also developed

sophisticated mathematical knowledge about the proposition. The distinction also corroborated

Durand-Guerrier’s (2003) view that proof construction is better characterized as an inferential

process that searches for examples and counterexamples to a proposition in order to determine

the maximal set on which the proposition is satisfied. Such views are also echoed in the

mathematical discipline and philosophy (Lakatos, 1976). However, the cognitive and

epistemological factors that drive students to explore more counterexamples remains an

interesting question. Nonetheless, the proposed classifications of students’ proof schemes

underlying both Deductive-Proof Construction and Proof-by-counterexample Construction

highlighted the transition phases between proofs based on examples and deductive proofs.

The classification of proof schemes further suggested the role of students’ mathematical

knowledge in proof construction. As seen from the distribution of proof schemes across levels

(see Tables 2 and 4), a number of students consider cases of a proposition when they were

constructing examples or deductive inferences, for instances, the cases of odd and even numbers
 15

for the set of whole numbers. Using an appropriate ways of distinguishing different cases (or

subsets) of the mathematical objects in the proposition, students were able to prove the

proposition. Such an approach required students’ mathematical knowledge about the properties

about mathematical objects as well as the strategic leverage of partitioning the set into these

cases versus alternative ways. Students’ mathematical knowledge help facilitated their proof

constructions in addition to knowledge about proofs and logical statements. The interplay

between students’ mathematical knowledge and inferential processes in during proof

construction thus generate further questions of when and how students’ mathematical knowledge

facilitate their proof construction. Inquiry into the interplay deserves equal attention in

understanding students’ proof constructions.

 16

References

Balacheff, Nicholas. (1988). Aspects of Proof in pupils' practice of school mathematics. In D.

Pimm (Ed.), Mathematics, teachers and children (pp. 216-235). London: Hodder &
Stoughton.
Boero, P. (1999). Argumentation and mathematical proof: A complex, productive, unavoidable
relationship in mathematics and mathematics education. International Newsletter on the
Teaching and Learning of Mathematical Proof, 7/8.
CCSSO. (2010). Common Core State Standards for Mathematics.
http://www.corestandards.org/assets/CCSSI_Math%20Standards.pdf
Chazan, Daniel. (1993). High School Geometry Students' justification for their views of
empirical evidence and mathematical proofs. Educational Studies in Mathematics, 24,
359-387.
Coe, Robert, & Ruthven, Kenneth. (1994). Proof Practices and Constructs of Advanced
Mathematics Students. British Educational Research Journal, 20(1), 41-53.
Durand-Guerrier, V. (2003). Which notion of implication is the right one? From logical
considerations to a didactic perspective. Educational Studies in Mathematics, 53(1), 5-34.
Duval, R. (1993). Argumenter, Démontrer, Expliquer: Continuité ou Rupture Cognitive. Petit,
10(31), 37–61.
Harel, Guershon, & Sowder, Larry. (1998). Students’ Proof Schemes: Results from exploratory
studies. In A. H. Schoenfeld, J. Kaput & E. Dubinsky (Eds.), Research in College
Mathematics Education III (pp. 234 - 283). Providence, R.I.: American Mathematical
Society.
Healy, Lulu, & Hoyles, Celia. (2000). A study of proof conceptions in algebra. Journal for
Research in Mathematics Education, 31(4), 396-428.
Hoyles, Celia, & Küchemann, Dietmar. (2002). Students' understanding of logical implication.
Educational Studies in Mathematics, 51, 193-223.
Knuth, Eric J. (2002). Secondary School Mathematics Teachers' Conceptions of Proof. Journal
for Research in Mathematics Education, 33(5), 379-405.
Ko, Yi-Yin, & Knuth, Eric J. (2009). Undergraduate mathematics majors’ writing performance
producing proofs and counterexamples about continuous functions. Journal of
Mathematical Behavior, 28, 68-77.
Lin, F. (2005). Modeling Students' Learning on Mathematical Proof and Refutation. Paper
presented at the the 29th Conference of the International Group for the Psychology of
Mathematics Education, Melbourne.
Miyazaki, Mikio. (2000). Levels of Proof in Lower Secondary School Mathematics. Educational
Studies in Mathematics, 41(1), 47.
NCTM. (2000). Principles and standards for school mathematics. Reston, VA: National Council
of Teachers of Mathematics.
 17

Selden, Annie, & Selden, John. (2003). Validations of Proofs Considered as Texts: Can
Undergraduates Tell Whether an Argument Proves a Theorem? Journal for Research in
Mathematics Education, 34(1), 4-36.
Stylianides, Andreas. (2007). Proof and Proving in School Mathematics. Journal for Research in
Mathematics Education, 38(3), 289-321.
Weber, Keith, & Alcock, Lara. (2004). Semantic and Syntactic Proof Productions.(Author
abstract). Educational Studies in Mathematics, 56(3), 209(226).
 18

Table 1

The types of Proof Construction Tasks in two sets

Task First set Second set Types of Proof Topic

construction
1 [x is a whole number] If 3x If a whole number x is Deductive- Elementary
is added to x2, then the sum added to its square, x2, proof Number
is an even number. then the result is an even Construction Theory
number.

2 If two prime numbers are If two prime numbers are Proof-by- Elementary
multiplied together, then added together, then the counterexample Number
the product is an odd sum is an even number. Construction Theory
number.
3 [For graphs of If x 2  Mx  N can be Deductive- Quadratics
y=ax2+bx+c] If a is factorized, then proof
positive and c is negative, M  N 1. Construction
then the x-intercepts of the
graphs are one positive
number and one negative
number.
4 If x 2  Mx  N can be [For graphs of Proof-by- Quadratics
factorized, then M  N  1 . y=ax2+bx+c] If the x - counterexample
intercepts of the graphs Construction
are both positive
numbers, then both a
and c are positive.

Table 2

Descriptors for coding students’ Deductive-proof Constructions

Level Characteristics Description of students' proofs

0 Irrelevant or minimal  Do not know how to prove or appeal to conviction in
engagement in external knowledge sources
inferences  Unable to relate the antecedent and the consequent with
examples
1 Novice use of  Conclude the implication is true (or false incorrectly) using
examples or logical one or more examples or an incorrect example or
reasoning counterexample
 Falsify the implication "if P then Q" through erroneous
logical reasoning, e.g., “if not P then not Q”
 19

 Derive a mathematical property unrelated to the implication

2 Strategic use of  Choose examples based on different cases of the antecedent
examples for or extreme examples
reasoning  Uses mathematical properties inferred from generated
examples to make conclusions
3 Use of deductive  Deduce relevant mathematical properties but missing one or
inferences with two deductive inferences for proving the implication
major flaws in  Deduce the implication to be true for some cases of the
logical coherence antecedent but omit other cases
and validity

4 Use of deductive  Generate a chain of deductive inferences to justify

inferences with conclusions but one or two inferences may be interpreted as
minor flaws in inductive due to insufficient substantiation or as writing
logical coherence errors
and validity  Displays disorganization in the chain of deductive
inferences
5 Construction of  Generate a chain of deductive inferences to justify the
informal deductive implication using informal justifications
proofs
6 Construction of  Generate a chain of deductive inferences to justify the
deductive proofs implication using formal representations
using formal
representations

Table 3

Frequency distribution of students’ proof schemes by level across topics

1st Session 2nd Session

Levels Number theory Quadratics Number theory Quadratic Total
Level 0 2 22 2 10 36
Level 1 15 22 14 33 84
Level 2 19 11 16 6 52
Level 3 8 1 7 4 20
Level 4 15 1 14 3 33
Level 5 1 3 5 4 13
Level 6 0 0 2 0 240

Table 4
 20

Descriptors for coding students’ Proof-by-counterexample Constructions

Level Characteristics Description of students' proofs

0 Irrelevant or minimal  Do not know how to prove or appeal to conviction in
engagement in external knowledge sources
inferences  Unable to relate the antecedent with the consequent
with examples
1 Novice use of  Conclude the implication is false (or true incorrectly)
examples for reasoning using one or more examples or an incorrect example or
or logical reasoning counterexample
 Derive a property which is not related to the conclusion
 Falsify the implication "if P then Q" through erroneous
logical reasoning, e.g., “if not P then not Q”
2 Strategic use of  Choose examples based on different cases of the
examples for reasoning antecedent
 Use mathematical properties inferred from generated
examples to make conclusions
3 Deductive inferences  Deduce the implication to be true incorrectly for some
with major flaws in cases of the antecedent but omit other falsifying cases
logical coherence and due to misconceptions
validity
4 Construction of proof-  Deduce the implication to be false by constructing one
by-counterexample or few specific counterexamples
5 Construction of proof-  Deduce the implication to be false by constructing a
by-general- general set of counterexamples and identifying the
counterexample property of the set that falsifies the implication

Table 5

Frequency distribution of students’ proof schemes by levels across topics and sessions

1st Session 2nd Session

Levels Number theory Quadratics Number theory Quadratic Total
Level 0 3 26 0 17 46
Level 1 8 22 4 24 58
Level 2 0 0 0 2 2
Level 3 3 1 5 3 12
Level 4 27 5 26 11 69
Level 5 19 6 25 3 53
 Page 21

1 Figure Captions

2 Figure 1. Level 2 – considering examples for cases of the antecedent (odd or even)

3 Figure 2. Level 3 – Incomplete deductive inferences but omitting a case of the antecedent

4 Figure 3. Level 4 – Proof that misses a step in making connected deductive inferences

5 Figure 4. Level 6 – use of formal representations in logically valid proof

6 Figure 5. Level 3 - Proof-by-counterexample Construction with inadequate consideration

7 Figure 6. Level 5 - Proof-by-counterexample Construction using a generic counterexample

9 Figure 1

10

11 Figure 2
 Page 22

2 Figure 3

4 Figure 4
 Page 23

2 Figure 5

4 Figure 6