THE INFLUENCE OF TEACHING PROGRAMMING ON LEARNING MATHEMATICS Geoff Wright, Peter Rich and Robert Lee Brigham Young University United States ge.wright@gmail.com peter_rich@byu.edu robert.leefam@gmail.com ABSTRACT There are several correlations between mathematics and programming, logistical thinking, problem solving, functions, variables, coordinates, and so forth. Many have argued the benefit of learning one having a positive influence on learning the other. Strangely, despite this belief, and supporting evidence, curriculum having a focus explicitly tied to mathematics and programming is not common in today’s K-12 classrooms. This project investigates a curriculum, titled: Bootstrap, that was explicitly developed to make mathematics connections to programming. The curriculum was implemented at three schools. A pre and post mathematics inventory was used to measure the impact teaching this course (which used a video game development programming pedagogy) had on student understanding of mathematics. The findings from the study suggest that student understanding of functions and variables did increase after participating in the course. KEYWORDS Programming Mathematics Technology Transfer Convergent Cognition BACKGROUND There has long existed a highly correlated relationship between academic success in mathematics and computer programming (Wright, Rich, & Leatham, 2010; Rich et al., 2012). Studies on the use of Logo in education demonstrate increased pattern recognition, fraction comprehension (Clements, 1995), problem-solving (Subhi, 1999), geometry, understanding of variable (McCoy, 1996), and spatial and symbolic mental representations (Hoyles & Noss, 1992), among many other cognitive factors (Clements, 2002). Despite years of research, teaching math through computer programming has not been generally adopted by secondary education (Johnson, 2000). Schanzer (2011) argues that much of the research curricula use imperative programming processes from languages such as Logo and BASIC. Imperative programming uses constructs such as loops, mutable variables, procedures, and GOTO statements that are not related to algebra. Students learning these curricula may be learning syntax and programming structures rather than algebraic concepts (p. 9). In 2005, Schanzer developed a curriculum called Bootstrap. The curriculum teaches algebraic concepts through programming computer games. Bootstrap uses Racket, a pure functional language, which initiates execution through algebraic expressions (Felleisen, 2009; Schanzer, 2011). The purpose of this study was to explore how learning the functional language Racket through the Bootstrap curriculum teaches the algebraic concepts of variables and functions to middle school students. Bootstrap was started by the TeachScheme! program team with Emmanuel Schanzer. Bootstrap has been implemented as an afterschool program. Felleisen (2010) states that “the program currently works with students in some ten underserved neighborhoods across the U.S.” (web retrieval) Felleisen asserts that the Bootstrap program “provides the strongest evidence yet that teaching functional programming directly affects the mathematics skills and interests of K-12 students” (web retrieval ). The TeachScheme! program, started in 1995, is an outreach project hosted by six universities. The program addresses the problems that stem from teaching objectoriented languages in beginning computer science classes. Part of the TeachScheme! program is to help all students better learn math through programming. Through TeachScheme!, Scheme has become the primary language researched in the math education through programming paradigm. Two studies presenting evidence of the value of Scheme/Racket to students’ math education have recently been document. One study was conducted by a teacher (Ms. North) at Westside High School in the Houston Independent School District in Texas. The other study was conducted by Paz et al. in northern Israel. North claims on her website that 100% of her 9th grades students enrolled in a combination programming/algebra class passed both the school district snapshot test and the math TAKS (Texas Assessment of Knowledge and Skills) Test. The result was compared to only 60% of students at the same high school passed the tests. Paz studied 11th grade students at various schools in Northern Israel. The study was a qualitative empirical study that began with “mathematical issues that came up serendipitously during a functional programming course.” The result of the study showed that a programming course in Scheme did help learning about functions, but also led to confusion about the formal function concept. There are not enough studies to back the claims of teaching Scheme helps students with algebraic skills. Instead, the idea is presented as a natural outcome of the functional nature of Scheme (See Felleisen et al. 1999). There is a need to investigate in a more disciplined manner the effect learning mathematics through Scheme has on students’ actual mathematic ability. RESEARCH In 2011 researchers from LPU, a large private university (an acronym) taught secondary students at two middle schools and local high school to program using the Bootstrap curriculum. Middle school participants (N = 17) opted to participate in an after-school programming curriculum to learn to make computer games. Math assessment comparison scores with a control group indicated that these students were representative of 7-8 graders in their schools. High school students (N = 7) represented 10-12th grade students who traditionally struggled with mathematics and were place in the course during school hours. Though there are many who focus on creating educational games for learning, Kafai (2006) proposes that we “turn the tables: by making games for learning instead of playing games for learning” [emphasis added] (p. 36). Bootstrap is a 9-lesson course that purports to teaches kids to program a computer games to introduce mathematics such as Cartesian Coordinates, order of operations, and the Pythagorean Theorem. By using a pure functional programming language, students gain practice at using variables and functions. The curriculum teaches that variables can be various types of objects, including integers, strings, and images. The advanced algebraic concepts of composition of functions and piecewise functions are used as part of development. We used a pre-test/post-test repeated measures design to better understand the link between programming and learning math, administering algebraic assessments both before and after the course. In each case, we matched participants with a control group of their peers in the same grade, administering the pre and post assessments in either an algebra or communications course. A multiple regression analysis was conducted on the scores from the assessments to determine if the Bootstrap students gained a statistically better understanding of the expected concepts. To understand student thinking about algebraic and programmatic concepts, we conducted structured interviews with participants from each Bootstrap group. Structured interviews revolved around asking students to solve several questions and to talk through their thought process out loud. Written assessments covered: variables, functions, Order of Operations, and the transfer of concepts to algebraic notation. To understand students’ concept of variable, we used items directly from the Knuth et al. studies. The questions on functions, operations, and transfer were designed by the math education members of the team. Using the pre-assessment score as a coefficient, a multiple regression analysis was calculated on the difference between the post-assessment scores and the pre-assessment scores. An Effect Size was also calculated. Each category of questions was analyzed independently. Where the assessments where analyzed for five different categories, a Bonferroni adjustment on the p-value to 1% was used to determine statistical significance. Of the three courses taught, statistical analysis was done with two of the courses, one middle school course and the high school course. The middle school course consisted of 9 students from the Bootstrap course, and 17 students in the control group. The high school course had 7 Bootstrap students, and 9 students in the control group. Participants’ strongest scores came in their increased understanding of variables. The middle school course showed a p-value of .0009 with a confidence interval of 1 to 3.38, with an effect size of the scores was .91. The high school course showed a p-value of .002, confidence interval from .46 to 1.58, with an effect size of .94. The assessments showed that students’ understanding of functions was suggestive, but inconclusive. The second course showed a pvalue < .0001 with a confidence interval of 3.25 to 6.8. But p-value for the third course was .09, confidence interval of -.7 to .7. The Effect Size for the middle school course was .54, and the high school course was .52. The assessments did not show any improvement on the understanding of Order of Operations. The results on the additional questions not related to the course were mixed. The weakest showing was in the category of transfer to algebraic notation. Most students left these questions blank. Students from all three courses were interviewed concerning the transfer of concepts to algebra. Because the students had worked for several weeks using Racket notation, students readily understood functions written in Racket. Further, with some coaching, most students interviewed noted that a function written in Racket could also be written in the algebraic notation (i.e., f(x) ). Yet, when presented with questions on composition and piecewise functions, students struggled with questions written either in Racket or algebraic notation. From this study, we can conclude that these students did gain a better understanding of variables and how they work in math, a traditionally difficult concept for students to master (Knuth, Alibali, McNeil, Weinberg, & Stephens, 2005). Students also showed some improvement in their understanding in functions, though that understanding did not transfer to algebraic notation. From the interviews and assessments, it became clear that the biggest inhibitor of learning functions was practice time. Composition and piecewise functions were introduced at the end of the course, and students appeared to have not experienced enough practice to fully develop the concept. Expanding the course could help resolve this issue. An aspect of the course that we failed to measure but that appears very was students’ level of enthusiasm. While some students indicated that the course was fair, and preferred a regular math class; many students expressed excitement over their accomplishments. One student’s comment captured this sentiment well. He noted that that, although he had to create equations in both math and in programming, that in programming, “the equation comes to life.” The observation was that a project-based approach to math was more engaging than the standard drill and practice. The implications of such motivation are potentially far-reaching. In our presentation, we aim to present the full analysis of this study. In addition, we will present ideas of different types of games that students could program that would utilize the mathematics throughout the new Common Core that most states have adopted. As students engage with mathematics in ways that “bring the equation to life,” we believe they may gain an even greater understanding of the nature of such math in ways they cannot get in traditional math courses alone. REFERENCES Clements, D. H., & Sarama, J. (1997). Research on logo. Computers in the Schools, 14(1-2), 9-46. doi: Felleisen, M. (2010). TeachScheme!: A checkpoint. SIGPLAN Not., 45(9), 129-130. Hoyles, C., & Noss, R. (1992). Learning mathematics and logo. Cambridge: MIT Press. Johnson, D. C. (2000). Algorithmics and programming in the school mathematics curriculum: Support is waning - is there still a case to be made? Education and Information Technologies, 5(3), 201-201214. Kafai, Y. B. (2006). Playing and making games for learning. Games and Culture, 1(1), 36-40. Knuth, E. J., Alibali, M. W., McNeil, N. M., Weinberg, A., & Stephens, A. C. (2005). Middle school students' understanding of core algebraic concepts: Equivalence variable. ZDM.Zentralblatt Für Didaktik Der Mathematik.Articles, 37(1), 68-76. McCoy, L. (1996). Computer-Based mathematics learning. Journal of Research on Computing in Education, 28, 438–460. Subhi, T. (1999). The impact of LOGO on gifted children's achievement and creativity. Journal of Computer Assisted Learning, 15(2), 98-108. Wright, G., Rich, P., & Leatham, K. (2010). AC 2010-2130: Increasing student and school interest in engineering education by using a hands-on inquiry based programming curriculum.