Algebraic thinking with and without algebraic representation: a threeyear longitudinal study
 661 Downloads
 8 Citations
Abstract
An Algebraic Thinking Test was given to 116 students aged 12–14, at the end of each of three years. This age span crosses two levels of school in New Zealand. This test assessed their ability to represent compensation in the four arithmetic operations both numerically and with letters for variables. The analyses of these results, together with the results from separate interviews designed to report individual progress of students in the New Zealand Numeracy Project, showed that students who had developed advanced mental strategies for dealing with additive, multiplicative and proportional operations, were the students who were capable of making full use of the alphanumeric symbols of algebra. These results, taken together with earlier studies by the authors, led to a proposal for a “pathway for algebraic thinking” accessible to all students.
Keywords
Operational Strategy Professional Development Program Lead Teacher Algebraic Thinking Decimal Fraction1 Introduction
In a recent plenary address to the North American Chapter of the International Group for the Psychology of Mathematics Education, Radford (2006) presented a compelling argument in support of a generalization approach to algebra in which he noted that, “the algebraic generalization of a pattern rests on the noticing of a local commonality that is then generalized to all terms of the sequence and serves as a warrant to build expressions of elements of the sequence that remains beyond the perceptual field” (p. 5). As well, Radford adds a third element, using the commonality to provide a direct expression or rule to specify any term of the sequence. But while it is customary to require that learners use the symbols of algebra to express such rules, Radford acknowledges that rulemaking proceeds through various layers of awareness articulated through different semiotic systems; words, gestures, pictures, graphs and symbols. Irwin & Britt (2005a) illustrate such a layer of awareness in a study that contributed to an evaluation of The New Zealand Numeracy Project, a national project in which students throughout New Zealand are encouraged to devise and experiment with a range of mental operational strategies in arithmetic. In that study, we argued that students who could apply an operational strategy to sensibly solve different numerical problems were disclosing an awareness of the relationships of the numbers involved as well as the underlying structure of the strategy. We claimed that successful application of such operational strategies demanded an awareness of the generality of the operational strategy thus illustrating algebraic thinking. Students’ explanation of their thinking revealed that they were treating the numbers as if they were variables. Fujii & Stephens (2001) refer to numbers used in this way as quasivariables (see also Warren & Cooper, 2002). The results of our study led to a view of algebra, particularly as it pertains to the pedagogy of introductory algebra, in which we do not see algebra as following arithmetic so that arithmetic has an ending that coincides with the beginning of algebra. Instead our view is consonant with those of Hewitt (1998, p. 20), who argues that algebra enables arithmetic to be carried out, and of Steffe (2001, p. 563), who argues that children’s knowledge of number together with numerical operational knowledge that is effective and reliable is essentially algebraic in nature.
Much has been written about the difficulties encountered during the transition from arithmetic to algebra (see for example, Filloy and Rojano, 1989; Herscovics & Linchevski, 1994; Kieran, 2001). But as Carraher, Schliemann, Brizuela and Earnest argue, acceptance of such a transition arises from an impoverished view of elementary mathematics in which mathematical generalization is postponed until the onset of algebra instruction (2006, p. 89). We argue similarly. The notion of algebra in arithmetic, in which generalization provides the basis for successful numerical operational thinking, dismisses the claim for transition and offers algebra for all through algebraic thinking with and without the symbols of algebra. Also, whereas Carraher et al (2006) argue for the early inclusion of algebraic symbols as a valuable tool for early algebraic thinking, our view is that young children need to work with several layers of awareness of generality prior to their introduction to the semiotics of algebra.
In this paper we begin our discussion of algebraic generality by illustrating how the development of algebraic thinking can evolve, without recourse to the symbols of algebra, from students developing an ongoing awareness of the underlying structure of operational strategies in arithmetic. We first draw on the New Zealand Numeracy Project (Ministry of Education, 2007), to illustrate the development of operational strategies in arithmetic and thus algebraic thinking for students at different ages, and then focus on a recent longitudinal study in which we attempt to ascertain what level of strategy development is likely to be necessary for students to extend their expressions of generality from using numbers themselves as quasivariables to a semiotic layer of awareness that embraces the literal symbols of algebra.
2 Algebra within arithmetic: algebraic thinking
2.1 The New Zealand Numeracy Project and algebraic thinking
In 1999, the New Zealand Ministry of Education introduced a professional development program in mathematics known as the Numeracy Project, motivated by the need to improve the number sense and operation sense of students by introducing a flexible approach to solving problems in numerical situations (see for example, McIntosh, Reys & Reys, 1992; Slavit 1999; Wright 1994). A primary purpose of the professional development project was to get teachers to focus on the strategies and knowledge that each student brought to a problem. At first the project was intended for year 1–3 students (aged 5–7 years). In 2001, the project was extended to year 4–6 students (aged 8–10 years). In 2002, following a pilot study involving over 4,000 year 7–10 students (aged 11–14), the project was expanded into the Secondary Numeracy Project (Ministry of Education, 2007), and offered to some 13,600 year 7–10 students. By 2007, nearly all elementary schools were using at least some aspects of the Numeracy Project and an increasing number of secondary schools were becoming involved.
The Number Framework for Stage 1–4 operational strategies that involve counting
Operational domains  

Global stage  Addition and subtraction  Multiplication and division  Proportions and ratio 
Counting  
Stage 0  Emergent  
Emergent  Unable to count a given or form a set of up to ten objects  
Stage 1  Onetoone counting  Onetoone counting  Unequal sharing 
One to one counting  Able to count a set of objects  Able to count a set of objects  Unable to divide a region or set into two or four equal parts 
Unable to form a set of objects to solve simple addition and subtraction problems  Unable to form a set of objects to solve simple multiplication and division problems  
Stage 2  Counting from one  Counting from one  Equal sharing 
Counting from one on materials  Counts objects to solve simple addition and subtraction problems  Solves simple multiplication and division problems by counting onetoone with the aid of materials  Able to divide a region or set into a given number of equal parts using materials 
Needs to use materials such as counters or fingers  
Stage 3  Counting from one  Counting from one  Equal sharing 
Counting from one by imaging  Counts objects by visualizing or imaging  Counts all the objects in simple multiplication and division problems by imaging the objects  Able to share a region or set into a given number of equal parts using materials or by imaging 
Unaware of 10 as a counting unit  Uses materials to solve multiplication and division problems with larger numbers  
Solves multidigit problems by counting all the objects  
Stage 4  Counting on  Skip counting  
Advanced counting  Counts on or back to solve simple addition and subtraction problems  Skip counts to solve simple multiplication and division problems using materials or imaging 
The Number Framework for Stage 5–8 operational strategies that involve partwhole thinking
Operational domains  

Global stage  Addition and subtraction  Multiplication and division  Proportions and ratio 
Partwhole  
Stage 5  Early addition and subtraction  Multiplication by repeated addition  Fraction of a number by addition 
Early additive  Uses a limited range of mental strategies to solve addition and subtraction problems. E.g., 8 + 7 is 8 + 8−1 (doubles) and 39 + 26 = 40 + 25 = 65 (partitioning and compensation)  Uses a combination of known multiplication facts and repeated addition. E.g., 4 × 6 is (6 + 6) + (6 + 6) = 12 + 12 = 24  Uses addition facts to find the fraction of a number. E.g., \( \tfrac{{\text{1}}} {{\text{3}}} \) of 12 is 4 since 4 + 4 + 4 = 12 
Uses known multiplication facts with repeated addition, to anticipate the result of division. E.g., 20 ÷ 4 = 5 since 5 + 5 = 10 and 10 + 10 = 20  Solves division problems mentally using halving or deriving from known addition facts. E.g., when seven pies are shared among four children each gets one pie plus \( \tfrac{{\text{1}}} {{\text{2}}} \) of a pie plus \( \tfrac{{\text{1}}} {{\text{4}}} \) of a pie  
Stage 6  Advanced addition and subtraction of whole numbers  Derived multiplication  Fraction of a number by addition and multiplication 
Advanced additive–early multiplicative  Can estimate answers and solve mentally addition and subtraction problems that involve whole numbers by choosing appropriately from a broad range of advanced mental strategies. E.g., 324 − 86 = 324−100 + 14 and 1242−986 = 1242 + 14 − (986 + 14)  Uses a combination of known multiplication facts and mental strategies to derive answers to multiplication and division problems. E.g., 4 × 8 = 2 × 16 = 32 (doubling and halving) and 9 × 6 is (10 × 6) − 6 = 54  Uses repeated halving or known multiplication and division facts to solve problems that involve finding fractions of a set or region, and division with remainders. E.g., \( \tfrac{1} {3} \) of 36 = 12 since 3 × 10 = 30, 6 ÷ 3 = 2, and 10 + 2 = 12. 
Stage 7  Addition and subtraction of decimals and integers  Advanced multiplication and division  Fractions, ratios, and proportions by multiplication 
Advanced multiplicative–early proportional  Can estimate answers and solve mentally addition and subtraction problems that involve decimals, integers and related fractions by choosing appropriately from a broad range of advanced mental strategies. E.g., 3.2 + 1.95 = 3.2 + 2 − 0.05 = 5.2 − 0.05 = 5.15  Chooses appropriately from a broad range of mental strategies to estimate answers and solve multiplication and division problems. E.g., 24 x 6 is (20 × 6) + (4 × 6) or 25 × 6 − 6; 81 ÷ 9 = 9 so 81 ÷ 3 = 3 × 9; and 4 × 25 = 100, so 92÷4 = 25 − 2 = 23  Uses a range of multiplication and division strategies to estimate answers and solve problems with fractions, proportions, and ratios. E.g., \( 13 \div 5 = (10 \div 5) + (3 \div 5) = 2\frac{3} {5}; \) 3:5 is equivalent to 24:40 since 8 × 3 = 24 and 8 × 5 = 40 
Stage 8  Addition and subtraction of fractions  Multiplication and division of decimals, multiplication of fractions  Fractions, ratios and proportions by reunitising 
Advanced proportional  Uses a broad range of mental partitioning strategies to estimate answers and solve problems that involve adding and subtracting fractions including decimals  Chooses appropriately from a broad range of mental strategies to estimate answers and solve problems that involve the multiplication and division of decimals and the multiplication of fractions. E.g., 4.2 ÷ 0.25 = (4.2 × 4) ÷ (0.25 × 4) = 16.8 ÷ 1  Chooses appropriately from a broad range of mental strategies to estimate answers and solve problems that involve fractions proportions and ratios. E.g., 6:9 is equivalent to 16:24 since \( 6 \times 1\tfrac{1} {2} = 9 \) and \( 16 \times 1\tfrac{1} {2} = 24 \) or \( 9 \times 2\tfrac{2} {3} = 24 \) and \( 6 \times 2\tfrac{2} {3} = 16 \) 
Combines ratios and proportions. E.g., 20 counters in ratio of 2:3 combined with 60 counters in ratio of 8:7 gives a combined ratio of 1:1 
The strategies illustrated in Table 2 involve partwhole thinking in which students recognize that numbers are abstract units that can be treated simultaneously or can be partitioned and recombined.

Sandra has 394 stamps. She gets another 79 stamps from her brother. How many stamps does she have then? One sensible strategy for this is to add 6 to 394 and subtract 6 from 79 so that the task becomes 400 + 73 or 473. Another strategy may be to split 79 into 6 + 73 and then combine 394 and 6 before adding 73 to the result, 400. Either strategy illustrates Stage 6, Advanced Additive partwhole thinking.

At the car factory, they need 4 wheels to make each car. How many cars could they make with 72 wheels? A sensible strategy for this task may be to see that 80 wheels are needed for 20 cars since 4 × 20 = 80, so that 20 − 2 = 18 cars need 72 wheels. Another strategy might be to note that 10 cars need 40 wheels since 10 × 40 = 40, and the remaining 32 wheels are required for a further 8 cars. So altogether 72 wheels are needed for 18 cars. Either strategy illustrates Stage 7 Advanced multiplicative partwhole thinking.

It takes 10 balls of wool to make 15 beanies (hats). How may balls of wool does it take to make 6 beanies? One sensible strategy for this task may be to se that two balls of wool will make three beanies so that four balls of wool will make six beanies. A second strategy may be to see that since 10 balls of wool make 15 beanies, 1 ball of wool will make \( 1\frac{1} {2} \) beanies so that 4 balls of wool will make 6 beanies. Either strategy illustrates Stage 8 Advanced Proportional thinking.
Since 2000, when it began as a modest experiment, the Numeracy Project has been subjected to a number of evaluations (see http://www.nzmaths.co.nz/numeracy/References/Reports.aspx). These evaluations have provided ongoing national data for the project over a fiveyear period. Evidence, including measures of effect sizes related to the performance of students, including those from different ethnic groups, on the Number Framework and of issues related to the validity of the Framework and reliability of tests administered by classroom teachers (r = 0.81), have supported the continuing development of the project (see Irwin & Niederer, 2002; Irwin, 2004; Tagg & Thomas, 2006a, 2006b; Thomas, Tagg & Ward, 2006; YoungLoveridge, 2004, 2006), and have led to the establishment of standards for students’ learning in number for each of the first four levels of the national curriculum (for ages 5 through 12), which is currently in its final stage of redevelopment (see Ministry of Education, 2006b).
2.2 Algebraic thinking in a Numeracy Project primary classroom
In order to clarify the algebraic nature of operational strategy teaching and learning, we illustrate a typical teaching sequence from the Numeracy Project. We describe the actions and responses of Mary, a student aged 8, as she engages with a series of tasks designed to help her advance from Stage 4 (Advanced Counting) to Stage 5 (Early Additive) strategy thinking.
Mary’s teacher began by asking her to put counters on a tensframe to show nine and then a further four counters on another tensframe to show four (Fig. 1a). Figure 1b shows the outcome of Mary’s actions that transform 9 + 4 into 10 + 3 so leading to her recognition of 13 as the answer.
Mary then went on to use the tensframes to figure out the solution to several similar tasks, 8 + 5, 7 + 6, and 9 + 7 before she was challenged by her teacher to see if she could work out 19 + 4, 27 + 6, and 38 + 7 without recourse to the tensframes. She could revert to using the tensframes if she was unsure what to do. She was asked by her teacher to explain her thinking after each response. For 19 + 4, she said she took one from the 4 and put it on the 19. So in her mind she could see 20 and 3 making a total of 23. She similarly transformed 27 + 6 into 30 + 3 = 33 and 38 + 7 into 40 + 5 = 45 each time describing an additive compensation strategy for which the underlying structure or generalization may be represented algebraically as a + b = (a + c) + (b − c). Such descriptions of her thinking represent a layer of awareness of generality expressed in words.
There are several aspects associated with Mary’s thinking that warrant further analysis. Firstly, the role assigned to Mary’s use of the tensframes is one of “ImageMaking”, an early level of understanding drawn from the Pirie–Kieren Model for the Growth of Mathematical Understanding (Pirie & Kieren, 1989, 1994), that forms the basis of a teaching model recommended for use with the Numeracy Project materials (Ministry of Education, 2007). In the Numeracy Project Teaching Model, students begin with a “Using Materials” phase designed to build concrete images that reflect her thinking and in so doing reflect the ImageMaking level proposed by Pirie and Kieren. As she manipulates the counters on the tensframes, she notices the space/s to be filled on one frame to make it complete and also where she can get the counter/s to do the filling of the space/s. She begins to develop an awareness of the consequence of this compensation action (Pirie–Kieren Image Having level), and so begins to visualize the structure of the transformation that she uses subsequently to solve the more challenging tasks, 19 + 4, 27 + 6, and 38 + 7. In the Teaching Model these tasks are included within a “Using Imaging” phase and are designed to challenge Mary’s thinking towards an awareness of the transformations involved in the tasks but in the absence of the tensframes. And as this happens, she is likely to have begun to isolate the features that are common to each task. This Pirie–Kieren “Property Noticing” level of understanding—“Using Number Properties” phase in the Project’s Teaching Model—in which she is able to explain the roles of the various elements in the transformation, leads to the conclusion that Mary has generalized the transformation, that is, she has engaged in algebraic thinking. As with the Pirie–Kieren model, the Project’s Teaching Model incorporates opportunities for foldingback. For example, if Mary had not succeeded with the tasks at the Using Imaging phase she would fold back to work with more tasks in the Using Materials phase. She would then advance again to the Using Imaging phase when she had demonstrated success with these tasks.
2.3 Algebraic thinking in the numeracy project: a research study with students aged 12
In 2002, we carried out an evaluation of one aspect of the Numeracy Project (Irwin & Britt, 2005a). The goal of the evaluation was to gauge the extent to which students aged 12, who participated in the Numeracy Project, used operational strategies, deemed algebraic in nature, more successfully than students of a similar age who did not participate in the project. We wanted to test our conjecture that project students would show a greater awareness of the algebraic structure of problems in arithmetic. The 21item test comprised six sections: compensation in addition, x + y = (x + a) + (y − a); compensation in subtraction, x − y = (x + a) + (y + a); compensation in the distributive law of multiplication over addition, xy = x (y + a) − xa; equivalence with sums and differences in which one of four structures is, “If x + a = b, then x = b − a”; compensation in multiplication, \( xy = (ax){\left( {\frac{y} {a}} \right)}; \) and equivalence with fractional values, again in which one of four structures is, “If \( \frac{a} {b} = \frac{{an}} {x}, \) then x = b × n”.
Altogether the study involved 899 students aged 12, from four schools of which 431 participated in the Numeracy Project in 2002 and 468 did not. Students of this age were chosen for the study because they had not been taught formal algebra prior to the study. The test was subjected to a Rasch analysis with reliability 0.88 as estimated by KuderRichardson’s formula 20. An analysis of the scores of the students who had participated in the Numeracy Project was significantly higher than that attained by those who had not participated in the project F(1,895) = 47.44, P < 0.01. The results suggested strongly that students’ involvement in the Numeracy Project was likely to lead, not only to improved outcomes for the arithmetic involved, but with respect to current and future algebraic activity, improved algebraic thinking skills. These results are in accord with the views of a number of researchers such as MacGregor and Stacey (1999) and Orton and Orton (1994) who have argued that the understanding of and skill in using arithmetical relations are a necessary prerequisite for learning algebra. While we do not disagree with this we feel that we are able to explain why this is likely to be the case. In our view, it is not merely that students involved in such arithmetic activity develop greater skill in this area but rather it is because they have already begun their algebraic development as a consequence of their participation in the Numeracy Project where a considerable amount of their time had been given to working with the strategies as described in the Project Framework (see Tables 1, 2). That work, as we have previously argued, demands the development of awareness of generality representing algebraic thinking.
2.4 Algebraic Thinking in the Numeracy Project: a teaching study with students aged 13
As previously indicated, the Numeracy Project was expanded in 2002 to include some 13,600 students in secondary school. The Project teaching program for these students focused not only on using a range of operational strategies in number that we describe as algebraic thinking but also provided increasing opportunities for students to express generality with the symbols of algebra. In the following, we show aspects of a lesson conducted in a Secondary Numeracy Project girlsonly secondary school by a student teacher with 30 students aged about 13, working mostly together in pairs.
The students, all of whom had been adjudged as being at least Stage 6 (Advanced Additive—Early Multiplicative) in the Number Framework (see Table 2), were asked to devise with the aid of colored counters, different representations of the 5by5 coaster and to illustrate each representation with an operational strategy for counting the total number of circles on the edge of a coaster with different numbers of circles, for example, 100, 47, and 139 on a side. They were then to attempt to write algebraic expressions, derived from their representations, for the total number of circles on the edge of a coaster with any number, n, of circles on a side.
A small number of students also produced the rule, 4 × n − 4, which they were able to explain as, “Each side has n circles so that makes four groups of n circles from the four sides. But each corner circle overlaps another corner circle, so four corner circles must be subtracted leaving a total of 4 × n − 4 circles”. Interestingly, none of these students represented this particular rule, arguably the simplest rule for the task, numerically prior to writing the rule algebraically. They said that they just saw the rule.
These students who were able to see generality within the figural representations and were able to express the generalization for a coaster with n circles on each side in a valid algebraic form may accordingly be classified as predominantly “Figural Generalizers” working at the “Representational Level of Generalizing”: which Rivera (2006, p. 4) notes, “is characterized by an ability to use symbols in expressing and communicating a generalization”. Comments made by a number of the students suggest that an important factor leading to successful rules or strategies for counting circles on the edge of coasters related to their use of colored counters to devise patterns of circles.
2.5 A research study in which generality is expressed by young children
Thus far we have focused our discussion of generality to students in primary or elementary schools and to beginning secondary classrooms. But it would be remiss if we were to ignore discussion of the algebraic thinking that children engage in prior to and during the beginning of their formal education. The earlier focus on additive compensation carried out by Mary (see Sect. 2.2) brings to the fore a study by Irwin (1996) who demonstrated that the concepts of compensation and covariation of the size of the whole with the size of the parts, without reference to numbers, was understood by children of 4 years of age. In her study, children aged 4 were asked if the amount of lollies (sweets) in two containers would be the same or different if some were moved from one container to another, if one was removed or added to one container, or if one was removed from one container and another added to the other container. Children of this age were certain that the total quantity would stay the same if an item was moved and would increase or decrease if one of the parts was altered. At age 5 and 6 they could explain these relationships, sometimes with a principle in their own language that showed they understood this as a generalization. For example, one child said the total number in a compensatory move would be the same, “The same, except (doll’s name) put one of the lollies from here to here”. Expressed algebraically, they understood that if x + y = W, then (x + a) + (y − a) = W, where x and y are the generalized unknown (uncounted) number of lollies in the separate containers, a stands for the number of lollies transferred from one container to the other and W stands for the number of lollies (uncounted) altogether.
However, if they were given a similar task with numbers only using doubles facts that they knew such as 5 + 5, and asked whether or not it would be the same as 4 + 6, they were unsuccessful until about age 7. As one child phrased it using a visual image for the equality of 10 + 10 and 11 + 9, “…because if you put one of the group of 11 over to the 9 group they would both be 10 and that means 20.” Since young children can understand this compensation concept when no numbers are attached, it may be that the complexity of learning to understand numbers distracts students from the knowledge that they had in a protoquantitative form (Resnick, 1992) before going to school.
2.6 The growth of algebraic thinking: from numbers to symbols. The 2004–2006 study
Following the study (Irwin & Britt, 2005a) in which we assessed the difference between students engaged in the Numeracy Project and those not in the Numeracy Project, we wanted to explore the growth in algebraic thinking across the important move for students from intermediate school (students aged 11 through 12), to secondary school. Our hope was that the algebraic thinking skills gained in intermediate school would not be lost when students proceeded to secondary school, where traditionally algebra has meant “doing arithmetic with letters”, which Mason with Graham and JohnstonWilder (2005, p. 309) contend, “has proved fruitless for countless generations”.
We reasoned that students, who had developed through their participation in the Numeracy Project, an awareness of the generality in a range of numerical operational strategies, would be able to extend their algebraic thinking to include the standard alphanumeric symbols of algebra. Students who could generalize in this way are at the “Representational Level of Generalizing” (Rivera, 2006). We argued that it was likely they would be able to capitalize on these generalizing skills as they progressed through secondary school.
This study was carried out as one of the regular evaluations of the Numeracy Project that is now used in almost all schools in New Zealand. We sampled students from eight schools in two cities, four from each. Specifically we focused on two questions: (1) What level of strategy development represented in the Numeracy Project Framework is likely to be necessary for students who have participated in the Numeracy Project to be able to extend their expressions of generality to include the literal symbols of algebra to represent compensation strategies in arithmetic that involve addition, subtraction, multiplication and division? (2) What are some of the critical professional characteristics displayed by teachers whose students are able to successfully use algebraic expressions to represent compensation strategies in arithmetic? For separate reports for each of the three years of this study see Irwin & Britt, (2005b, 2006, 2007).
3 Method
3.1 Participants
Characteristics of secondary schools that participated in the assessment of algebraic thinking in 2006
School pair number  Decile ranking of intermediate school^{a}  Number of students in the intermediate school^{a}  Decile ranking of secondary school^{a}  Number of students in secondary school^{a}  Number of Students for whom three results are available 

1  2  222  4  726  13 
2  3  280  4  1201  14 
3  5  621  6  1583  61 
4  7  530  8  1314  28 
Total  116 
Altogether, 116 students took the test on three occasions. The number of students in the different pairs of schools varied, as shown in Table 3. This reflects mobility of students, absences on one of the three occasions, and related factors. There is no reason to suggest that the sample was atypical.
The intermediate schools were all participants in the Numeracy Project. The secondary schools were chosen by finding out which secondary school the majority of students from the intermediate school moved on to. We asked schools to give the test to all students in year 8 (age 12) at intermediate school in 2004, year 9 at secondary school in 2005, and year 10 in 2006. In addition, each secondary school was asked to give the test to all of their year 9 and 10 students in each of the three years. Our main focus was on the students who took the test for each of the three years, thus giving us three data points for each of the students.
3.2 The assessment instrument
The full test is available as Appendix K in Irwin & Britt (2007).
3.3 Procedure
Teachers in all schools gave the Algebraic Thinking Test to their students during the final term of the year in question. The tests were then returned to the authors for marking.
3.4 Method of analysis
All tests were marked either by the first author, or under the supervision of the authors, by preservice secondary student teachers familiar with the Numeracy Project. Markers were told to mark items as correct if they followed a method reliant on understanding of the underlying compensation principle. Scores for each participant were then entered on a spreadsheet that enabled us to see which items each student was successful on. Spreadsheets were compared across years to identify students tested on all three occasions. We then gained totals for each subject and each item. A random coefficients’ analysis of the data was undertaken by means of SPSS’s mixed linear model (Kreft & de Leeuw, 1998; Singer & Willett, 2003).
4 Results
Three analyses were carried out to evaluate the data gathered over 3 years from the tests of algebraic thinking. The first analysis determined the correlation of individual students’ scores on the Algebraic Thinking Test with their three strategy scores obtained from the Numeracy Project Diagnostic Interview at the end of year 9 (age 13). The second analysis looked at the Algebraic Thinking Test scores for students who took the test in years 8, 9 and 10 in 2004, 2005 and 2006 respectively. The third analysis took a closer look at what was happening in the one pair of schools where the mean scores on the Algebraic Thinking Test were significantly superior to the other schools.
4.1 Relationship between scores on the Algebraic Thinking Test and the Numeracy Project Diagnostic Interview
In 2005, the scores that year 9 students gained on the Algebraic Thinking Test and on the Numeracy Project Diagnostic Interview were correlated for the three secondary schools in the Secondary Numeracy Project. These students had been given the diagnostic interview described in Sect. 2.1, “The New Zealand Numeracy Project and Algebraic Thinking”. Scores were calculated by awarding a score of 6 for a strategy judgment of Stage 6, a score of 7 for a strategy judgment of Stage 7 and so. While scores for each of these operational domains are reported individually for diagnostic purposes, we added them together, reasoning that students needed to have a good ability in all skills to be a competent mathematics student in this age group.
For 557 students for whom both scores were available, the correlation between scores for these strategy stages on the interview and scores on the Algebraic Thinking Test was 0.47 with significance <0.01. Thus we were assured that individual interviews demonstrated that students who were flexible in their thinking about numerical problems as fostered by the Numeracy Project were the same ones who could transfer from using numbers flexibly to using letters to express the generalizations.
A further analysis of the diagnostic test strategy scores shows that students with a total score above the median score of 19 across the three operational domains on the Numeracy Project Diagnostic Test were those who demonstrated an awareness of the underlying algebraic structure of the operation and could express the generalization algebraically on our Algebraic Thinking test. This means that students who were operating at or above the advanced additive stage on all three operational domains were the students most likely to transfer this flexible numerical thinking to algebraic thinking in which the generalization is expressed algebraically.
4.2 Development of algebraic thinking across three years
This analysis was of the scores of the 116 students who took the Algebraic Thinking Test on three occasions.
Correlation coefficients among the students’ scores over three years
Year 9  Year 10  

Year 8  Pearson correlation  0.640**  0.639** 
Year 9  Pearson correlation  0.714** 
The other model took into consideration the effect of attending different pairs of schools. This model provided a marginally better fit to the data than the simpler one. The goodnessoffit quantity, −2 log likelihood, was 2028.3 for the simpler model and 2021.1 for the more complex one, a difference equivalent to chisquare of 7.2 with 3 df, and P = 0.066. If this is taken to show a worthwhile improvement in the model’s fit, then the resulting estimated initial score and improvement rate for students from each school can be shown in Fig. 6 by the lines labeled “1”, “2”, “3”, and “4”. For both models, the rate of change (1.3 points per year) represented a significant improvement.
Scores from intermediate school 3 differed significantly from schools 2 and 4 (P < 0.05) but not from intermediate school 1 (P = 0.081) because of the large variance in the small number of student scores at that school, (t test). The difference between schools when all three years were taken into consideration approached significance F(3,112) = 2.620, (P = 0.054).
A still more complex model, which allowed for different average rates of improvement for students from different schools, did not improve the fit; it therefore is not discussed further.
4.3 What was happening in one pair of schools
Teachers in all schools were interviewed but we chose to focus on the interviews from the two Pair 3 schools in order that we might identify factors that contributed to the success of their students. We interviewed the assistant principal who was the lead teacher for mathematics at the contributing intermediate school in 2004 and in 2006 and the teachers and head of the mathematics department at the secondary school in 2005 and in 2006. The important characteristics of the numeracy lead teacher in the intermediate school appeared to include a firm knowledge of relevant mathematics, of the Numeracy Project, and of the connections between number and algebra. Other teachers had enabled her to recognize when firmer supervision of her teachers was needed. She had the full support of her principal, who was informed about the Numeracy Project. He told us, “all credit went to the woman in charge of mathematics”. This lead teacher gave credit to the man who had led their school through the professional development aspect of the project. She wrote in an email to K. Irwin, (1/5/2006), “During the 2 years we were on the contract, our teaching was very focused on number and I would say it took about 80% of the (mathematics) teaching time. Also the facilitator may have contributed to our brighter students being encouraged to move from the strategies into algebraic thinking. He gave me a lot of good ideas for doing this”. In describing her own teaching, she said that in 2004 she had already taken her top group into algebra. When we asked if this was traditional algebra, she appeared almost shocked. Of course not, she took the students from their use of strategies for numerical calculation into the generalities that could be expressed algebraically. This was a natural growth for her. In the email quoted above, she reported that the school had chosen algebra as one of their target areas for year 7 for 2006. “A curriculum combining algebra and number can only be a help. Very sensible.” She thought that most of the teachers were managing to teach the Numeracy Project well but indicated that there were two groups of teachers that needed her help: the less confident teachers and those who did not want to change their practice.
She described the ways in which she worked with these less confident teachers as well as those who did not want to change their teaching practices. We formed the impression that she had a warm but professional relationship with the teachers that added to her effectiveness as a mathematics leader. Her overall view was that students were enjoying mathematics more now. Their attitude was positive, and they were not afraid of using letters in place of numbers. Their understanding of the generality of the numerical operational strategies and measurement formulae for perimeter and area deepened as they explored the use of different numbers in response to “Will it work all the time?”
The staff of this school had regular contact with the secondary schools that their students attended. They passed on results of students’ progress in the Numeracy strategies.
The head of the mathematics department at the secondary school in Pair 3 did not participate in the senior version of the Numeracy Project but knew about the project in some depth. The mathematics department had one teacher who taught only numeracy to those students who needed it. Others integrated the concepts into their traditional curriculum where they thought appropriate. In 2005, the teachers told us that they had put more emphasis on algebra than before. In 2006, the teachers in the department had taught number and algebra over terms 1 and 2 of the fourterm year and then integrated algebra into the other topics that they taught during the rest of the year. At the interview in 2006, we told the mathematics staff that their students outperformed the three other secondary schools and asked why they thought that might be. The teachers’ response was that they did not think their students were very good. As the discussion developed, we learned that the head of the mathematics department rewrote the departmental scheme for mathematics every year. For 2006, he had incorporated some of the ideas from the Algebraic Thinking Test. In 2006, he had judged that students from intermediate schools who had had the Numeracy Project were noticeably different from those who had not participated in the project. The teachers rejected the suggestion that their students had done well on this test as the result of any of their own efforts, saying that all credit must go to the intermediate school. Senior teachers at the school said that algebra was the basis of all high school mathematics and had to be brought in whenever possible. “We concentrate on algebra because they are doing badly.” “Number underpins everything.” They accepted different methods if the students could justify them: “Cannot have one size fits all.”
5 Discussion
5.1 Algebra within arithmetic
In an earlier study (Irwin & Britt, 2005a), we showed that students who had participated in the New Zealand Numeracy Project were more able than students who had not participated in the Numeracy Project to demonstrate an awareness of the underlying algebraic structure of the operational strategies they used to solve problems in arithmetic. We noted that this awareness of structure amounted to an awareness of generality and argued that such students were therefore engaging in algebraic thinking. We also claimed that, rather than using the literal symbols of algebra, these students who had no prior experience with such symbols, were thinking of the numbers themselves as variables. We referred to these as quasivariables (Fujii & Stephens, 2001) and subsequently argued that the arithmetic students used to solve problems was rooted in algebra. That is, without algebra there could be no arithmetic (Hewitt, 1998; Mason with Graham & JohnstonWilder, 2005). In this study, we have extended this relationship between algebra and arithmetic by showing that students, who had participated successfully in the Numeracy Project at the Intermediate School for years 7 and 8, were capable of algebraic thinking that involved representing numerical generalities with the special symbols of algebra. We also showed that these algebraic thinking skills for such students continually improved as they progressed through the first 2 years of secondary school.
This claim arises from the analyses of the data collected over the threeyear period of the study. We noted that despite individual differences in performance, average scores increased significantly over the 3 years from year 8 to year 10. Also, the correlation between the scores on the Numeracy Project Diagnostic test and the Algebraic Thinking Test, showed a significant relationship. These analyses taken together suggest strong support for a positive relationship between success in the Numeracy Project and subsequent algebraic thinking in which students use the literal symbols of algebra to express algebraic generalizations.
5.2 Teachers who effect improvement in algebraic thinking
We interviewed teachers in all of the schools and drew some conclusions from the interviews of School Pair 3. In particular, the roles of the lead teachers as key in providing support for the development of algebraic thinking warrant attention. The students in this pair of schools, designated as School 3 in the analyses, performed creditably on the algebraic thinking test in each of the three years of the study (see Fig. 6). In addition, Fig. 7 shows that in the secondary school of the School 3 pairing, students in year 10 performed better than the year 9 students in each of the three years they were tested with the Algebraic Thinking Test. School 3 also consistently outperformed the other pairs of schools (see Fig. 6). Much of the difference in performance in these schools was likely to have been attributable to the teachers directly responsible for the classroom mathematics teaching and overseeing that teaching.
There are several factors that seem to have been important contributors to the successful leadership roles of the assistant principal in the intermediate school who was in charge of teaching mathematics and of the head of the mathematics department at the secondary school.
The teacher responsible for leading mathematics in the intermediate school had a welldeveloped mathematical knowledge for teaching (Ball, 2002). She had taken university courses in mathematics designed specifically for teachers of intermediate and early secondary school students, she had acquired a deep understanding of the intentions of, and activities recommended in, the Numeracy Project, and she had grasped the notion of “algebra within arithmetic” as a result of the support she had received from the Numeracy Project’s School Facilitator who had been attached to work with her and her colleagues in their classrooms. In her leadership role, she had learnt how to successfully manage and encourage colleagues who struggled with change in their classroom environment. In addition she had leadership qualities that enabled her to recognize teachers who still worked from an impoverished view of arithmetic and found ways of helping them.
The secondary school teacher, who seemed to have made an important impact at his school, had been head of the school’s mathematics department for a number of years. He knew the mathematics he was expected to teach at all levels in the secondary school and seemed to be open to experimenting with alternative classroom teaching approaches. Although he had no direct classroom experience with the Numeracy Project, he knew of it and supported its overall goals and classroom intentions. Interestingly, although the students at his school had performed creditably on the Algebraic Thinking Test, he, together with some of his colleagues, felt that the students could do better. And in order to effect improvement, he had in late 2005, rewritten the school’s mathematics teaching scheme for 2006 to incorporate some of the ideas from the Algebraic Thinking Test. As part of this schoolbased curriculum development process he also seems to have convinced his colleagues of the efficacy of his teaching scheme modifications. Both of these teachers stood out as quality professionals who were respected by their colleagues and their respective school Principals.
5.3 Teaching algebraic thinking
When the analyses discussed above are taken together we believe they contribute to a very strong case to be made for the inclusion in mathematics curricula of activity that is based on the notion of “algebra within arithmetic”. We contend that, where the focus for arithmetic in classroom curricula is on developing an awareness of the underlying structure of the operations rather than on merely getting answers through procedures and algorithms, it is likely there will be important positive consequences for the development of algebraic thinking. Thus, students who reach a level where they can readily access mental operational strategies, such as those identified in the Numeracy Project, to sensibly and flexibly solve numerical problems, are likely to be able to grasp and work with algebra where full use is made of algebraic symbols to express algebraic generality and which is central to most secondary school mathematics curricula.
We argue that the students in this study who can express numerical relationships algebraically are working at the Representational Level of Generalizing as proposed by Rivera (2006) in his Theory of Generalizing Types. In our earlier study (Irwin & Britt, 2005a), which also focused on students generalizing, we claimed that students who were able to use quasivariables (numbers for letters) to generalize numerical operational strategies were engaged in algebraic thinking. Furthermore, the study by Irwin (1996) revealed an awareness of generality among children aged 4 as they proposed additive compensation principles without needing to establish cardinality. These children had acquired this knowledge in a protoquantitative form (Resnick 1992) prior to their going to school. However, because teachers at the beginning school level often get caught up in helping their children get answers to simple numerical calculations, this early ability to generalize goes unnoticed.
In order to consider the educational implications arising from this study, we have combined its finding with those from the studies by Irwin & Britt (2005a) and Irwin (1996). As a result of having this wider perspective, we propose a pathway for algebraic thinking that we believe is accessible to all students motivated to develop a relational understanding of number and operations with number. The pathway firstly involves young children engaging in activities of a protoquantitative nature where they explore and describe in their own words, generalities arising from simple operational strategies that do not require extensive counting knowledge. It then involves primary school students deepening their understanding of a range of operational strategies in arithmetic and finally it involves middle school and beginning secondary school students expressing algebraically, generalizations drawn from numerical situations and figural representations. In proposing this pathway, we note that it incorporates many of the characteristics of both protorepresentational and representational generalizing that form the theoretical framework in the Theory of Generalizing Types proposed by Rivera (2006).
New Zealand teachers and their students have had a growing involvement in the Numeracy Project since 1999. The experimental nature of the project together with its strong evaluative component and accompanying teacher professional development, has led to improvements in skill level and confidence in teaching numeracy. Increasingly, teachers are developing awareness that their work entails considerably more than helping students get answers to problems involving numerical situations. They are supported by the Numeracy Project resource material and a curriculum that encourages the concept of algebra within arithmetic. This national “experiment” has led to professional development for most primary school teachers and an increasing number of secondary school teachers. The challenge now is to enable all teachers to see the benefit of an approach that includes seeing algebra within arithmetic.
Notes
Acknowledgments
We are grateful to the Ministry of Education for funding this research project. We also express our thanks to R. J. Irwin for some of the statistical analysis.
References
 Ball, D.L. (2002). Knowing mathematics for teaching: Relations between research and practice. Mathematics and Education Reform Newsletter, 14(3), 1–5.Google Scholar
 Carraher, D.W., Schliemann, A.D., Brizuela, B.M., Earnest, D. (2006). Arithmetic and algebra in early mathematics education. Journal for Research in Mathematics Education, 37, 2, 87–115.CrossRefGoogle Scholar
 Filloy, E., & Rojano, T. (1989). Solving equations: the transition from arithmetic to algebra. For the Learning of Mathematics, 9(2), 19–25.Google Scholar
 Fujii, T., & Stephens, M. (2001). Fostering an understanding of algebraic generalizations through numerical expressions. In: Proceedings of the 12th conference of the international commission on mathematical instruction: the future of the teaching and learning of algebra. Vol. 1, (pp. 258–264). Melbourne: The University of Melbourne.Google Scholar
 Herscovics, N., & Linchevski, L. (1994). A cognitive gap between arithmetic and algebra. Educational Studies in Mathematics, 27, 59–78.CrossRefGoogle Scholar
 Hewitt, D. (1998). Approaching arithmetic algebraically. Mathematics Teaching, 163, 19–29.Google Scholar
 Irwin, K.C. (1996). Children’s understanding of the principles of compensation and covariation in partwhole relationships. Journal for Research in Mathematics Education, 27, 25–40.CrossRefGoogle Scholar
 Irwin, K. C. (2004). An evaluation of the Numeracy Project for Years 7–9 2003. Final report to the Ministry of Education. Wellington: Learning Media.Google Scholar
 Irwin, K. C., & Britt, M. S. (2004). Operating with decimals as a partwhole concept. In: Proceedings of the 27th Annual Conference of the Mathematics Education Group of Australasia. Vol. 2, (pp. 312–319). Townsville/Sydney: MERGA.Google Scholar
 Irwin, K.C., & Britt, M.S. (2005a). The algebraic nature of students’ numerical manipulation in the New Zealand Numeracy Project. Educational Studies in Mathematics, 58, 169–188.CrossRefGoogle Scholar
 Irwin, K. C., & Britt, M. S. (2005b). Algebraic thinking in the Numeracy Project: year one of a threeyear study. In: J. Higgins, K. C. Irwin, G. Thomas, T. Trinick, & J. YoungLoveridge (Eds.), Findings from the New Zealand Numeracy Development Projects 2004 (pp. 47–55). Wellington: Learning Media.Google Scholar
 Irwin, K. C., & Britt, M. S. (2006). Algebraic thinking in the Numeracy Project: year two of a threeyear study. In: F. Ell, J. Higgins, K. C. Irwin, G. Thomas, T. Trinick, & J. YoungLoveridge (Eds.), Findings from the New Zealand Numeracy Development Projects 2005 (pp. 46–54). Wellington: Learning Media.Google Scholar
 Irwin, K. C., & Britt, M. S. (2007). The development of algebraic thinking: results of a threeyear study. In: B. Annan, F. Ell, J. Fisher, N. Hawera, J. Higgins, K. C. Irwin, A. Tagg, G. Thomas, T. Trinick, J. Ward, & J. YoungLoveridge (Eds.), Findings from the New Zealand Numeracy Development Projects 2006 (pp. 33–43). Wellington: Learning Media.Google Scholar
 Irwin, K. C., & Niederer, K. (2002). An evaluation of the Numeracy Exploratory Study (NEST) and the associated Numeracy Exploratory Study Assessment (NESTA) Years 7–10, 2001. Wellington: Learning Media.Google Scholar
 Kieran, C. (2001). Looking at the role of technology in facilitating the transition from arithmetic to algebraic thinking through the lens of a model of algebraic activity. In: Proceedings of the 12th conference of the international commission on mathematical instruction: the future of the teaching and learning of algebra. Vol. 3, (pp. 713–719). Melbourne: The University of Melbourne.Google Scholar
 Kreft, I., & de Leeuw, J. (1998). Introducing multilevel modeling. London: Sage Publications Ltd.Google Scholar
 McIntosh, A., Reys, B.J., & Reys, R.E. (1992). A proposed framework for examining basic number sense. For the Learning of Mathematics 12(3), 2–8.Google Scholar
 MacGregor, M., & Stacey, K. (1999). A flying start to algebra. Teaching Children Mathematics, 6(2), 78–85.Google Scholar
 Mason, J., Graham, A., & JohnstonWilder, S. (2005). Developing thinking in algebra. London: The Open University.Google Scholar
 Ministry of Education. (1992). Mathematics in the New Zealand curriculum. Wellington: Learning Media.Google Scholar
 Ministry of Education. (2006a). Book 1, The Number Framework. Numeracy professional development projects. Wellington: Learning Media.Google Scholar
 Ministry of Education. (2006b). The New Zealand curriculum. Draft for consultation. Wellington: Learning Media.Google Scholar
 Ministry of Education. (2007). New Zealand Numeracy Projects, Retrieved February 15, 2007, from http://www.nzmaths.co.nz/numeracy/index.htm.
 Orton, A., & Orton, J. (1994). Students’ perception and use of pattern and generalization. In: Proceedings of the 18th Annual Conference for the Psychology of Mathematics Education, Vol. 4, (pp. 407–114). Lisbon: Portugal.Google Scholar
 Pirie, S., & Kieren, T. (1989). A recursive theory of mathematical understanding. For the Learning of Mathematics, 13(1), 7–11.Google Scholar
 Pirie, S.E.B., & Kieren, T.E. (1994). Beyond metaphor: formalising in mathematical understanding within constructivist environments. For the Learning of Mathematics 14(1), 39–43.Google Scholar
 Radford, L. (2006). Algebraic thinking and the generalization of pattern: a semiotic perspective. In: Proceedings of the 28th annual meeting of the North American Chapter of the International group for the Psychology of Mathematics Education, Universidad Pedagógica Nacional Vol. 1 (pp. 2–21). Mérida, Mexico.Google Scholar
 Resnick, L.B. (1992). From protoquantitative to operators: Building mathematical competence on a foundation of everyday knowledge. In: L. G. Leinhardt, R. Putnam, & R. A. Hatrup (Eds.), Analysis of arithmetic for mathematics teaching (pp. 373–430). Hillsdale: Earlbaum.Google Scholar
 Rivera, F. (2006). Structures of elementary algebraic thinking: evidence from sixth grade students and implications for classroom practice. Intersection, 7(2), 2–13.Google Scholar
 Slavit, D. (1999). The role of operation sense in transitions from arithmetic to algebraic thought. Educational Studies in Mathematics, 37, 251–274.CrossRefGoogle Scholar
 Steffe, L. P. (2001). What is algebraic about children’s numerical operating? In: Proceedings of the 12th conference of the international commission on mathematical instruction: the future of the teaching and learning of algebra. The University of Melbourne, Vol. 2, (pp. 556–563). Melbourne: Australia.Google Scholar
 Tagg, A., & Thomas, G. (2006a). Secondary Numeracy Project knowledge test analysis. In: R. Harvey, J. Higgins, A, Tagg, & G. Thomas (Eds.), Evaluations of the 2005 Secondary Numeracy Pilot Project and the CAS Pilot Project. (pp. 5–11). Wellington: Learning Media.Google Scholar
 Tagg, A., & Thomas, G. (2006b). Performance on the Number Framework. In: R. Harvey, J. Higgins, A. Tagg, & G. Thomas (Eds.), Evaluations of the 2005 Secondary Numeracy Pilot Project and the CAS Pilot Project (pp. 12–35). Wellington: Learning Media.Google Scholar
 Thomas, G., Tagg, A., & Ward, J. (2006). Numeracy assessment: how reliable are teachers’ judgments? In: F. Ell, J. Higgins, K. C. Irwin, G. Thomas, T. Trinick, & J. YoungLoveridge (Eds.), Findings from the New Zealand Numeracy Development Projects, 2005 (pp. 91–102). Wellington: Learning Media.Google Scholar
 Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analyses: modeling change and event occurrence. Oxford: Oxford University Press.Google Scholar
 Warren, E., & Cooper, T. J. (2002). Arithmetic and quasivariables: a year 2 lesson to introduce algebra in the early years. In: Proceedings of the 25th annual conference of the mathematics education research group of Australasia, The University of Auckland, Vol. 2, (pp. 673–681). Sydney: MERGA.Google Scholar
 Wright, R. (1994). Mathematics in the lower primary years: a researchbased perspective on curricula and teaching practice. Mathematics Education Research Journal, 6, 23–36.Google Scholar
 YoungLoveridge, J. M. (2004). Patterns of performance and progress on Numeracy Projects 2001–2003: further analysis of Numeracy Data. Wellington: Learning Media.Google Scholar
 YoungLoveridge, J. M. (2006). Patterns of performance and progress on the Numeracy Development Project: Looking back from 2005. In: F. Ell, J. Higgins, K. C. Irwin, G. Thomas, T. Trinick, & J. YoungLoveridge (Eds.), Findings from the New Zealand Numeracy Development Projects 2005 (pp. 6–21). Wellington: Learning Media.Google Scholar