Department of Mathematics Kansas State University Manhattan, KS, USA bennett@math.ksu.edu
Abstract— Clinical interviews were held with students in senior
level electrical engineering and methods of teaching secondary II. METHODOLOGY mathematics classes. These students had all completed a 4- semester calculus sequence including differential equations. A. Subjects Engineering students had then taken advanced classes applying This study builds on earlier research about the level of mathematical ideas in real-world contexts while mathematics conceptual understanding exhibited by students at the end of a education students had taken a similar amount of coursework in 4-semester math sequence, Calculus 1, 2, 3, and Differential advanced mathematics courses. Students were interviewed to Equations [3]. The participants consisted of 20 senior-level determine how their conceptual understanding of function and engineering majors and nine senior-level mathematics accumulation (integration) had grown or regressed during work after calculus. Back-transfer, where later learning improves education majors. Each group was selected from their classes in conceptual understanding of earlier material, was observed. the Spring semester of 2011. The engineering majors were recruited from ECE 512: Linear Systems in the College of Keywords-component; back transfer; APOS; function; Engineering. The mathematics education majors were acculumation; understanding recruited from MATH 570: History of Mathematics in the Mathematics Department. These groups were selected because students in both groups had completed the same 4-course I. INTRODUCTION sequence, typically during their first two years of college, and There have been numerous studies looking at the concept of had both taken additional courses which used that material transfer, where material learned in one course is available (or afterward. However, the engineering students had seen the not) for application in later courses (for a variety of examples calculus ideas applied in more real-world settings than the of such studies, see chapter 3 of [1]). The idea of transfer education majors, few of whom took any additional science underlies much of the structure of the modern engineering courses and instead primarily looked at more abstract curriculum, where students begin in calculus and physics and applications in advanced math courses. By comparing the then apply that knowledge in later work. Much less studied is conceptual understanding displayed by the seniors to that back-transfer, where seeing material in a later course deepens displayed by sophomores, back transfer could be assessed. By understandings of concepts that could have been developed in comparing engineering and education majors, the role of an earlier course [2]. A proper understanding of how engineers’ different contexts of application could also be assessed. conceptual understanding develops throughout their training requires an understanding of how both transfer and back- B. Interviews transfer work. The research questions addressed in this study are: Each student participated in a (roughly) 60 minute clinical interview. Interviews were conversation-style following a • Does back transfer occur? protocol with predetermined questions along with spontaneous, related follow-up questions determined by the course of the • If so, does the application of math in different contexts interview to get at understanding. Questions were taken mostly lead to different levels of back transfer? from the level of Calculus I and II with two questions coming This study focuses on the concepts of function and from Differential Equations. The questions were chosen based accumulation (integration) as developed in calculus. However, on whether each would demonstrate a student’s understanding a better understanding of back transfer in mathematics should of various calculus concepts including function, differentiation, provide a guide to the study of transfer and back-transfer in and integration. For example, the interview questions included: other aspects of the engineering curriculum. Graph x(t) = cos t, y(t) = sin t, 0 ≤ t ≤ 2π Graph x(t) = cos 2t, y(t) = sin 2t, 0 ≤ t ≤ π Are these the same functions?
The researchers were not specifically interested in the These codes were then used by the researchers to rate each “correctness” of the student responses. However these interview with an average level of confidence and willingness questions lead easily to a discussion of ideas about what it from 1 to 3 using half steps so that they would be more means to be a function, from which the students’ understanding comparable to the rating of the student’s level of of the concept could be assessed. understanding on the modified APOS scale.
C. Analysis III. PRELIMINARY RESULTS
Interviews were transcribed and analyzed using both Senior students had, on average, progressed along the qualitative and quantitative techniques. APOS scale farther than students at the end of the calculus 1) First pass (qualitative) sequence. However, not all students exhibited growth. Indeed, After each interview, students were rated on their level of some students never advanced past the initial Action level conceptual understanding of function and accumulation. These during their entire college career. So in answer to research ratings were based on a modified Action-Process-Object- question 1, back transfer can occur, but is not guaranteed. Schema (APOS) scale that was refined to admit intermediate While engineers overall displayed higher levels of levels between each of the normal Action, Process, Object and conceptual understanding than education majors, we are unable Schema levels (see [4] for details about the APOS scale). A to conclude that exposure to more concrete applications was student was rated by the researchers by analyzing how they better at supporting back transfer than the more abstract answered the series of questions in the interview and deciding applications of advanced mathematics classes. Variations to what level a student could think about a particular topic. A within the majors were larger than variations between the student would be rated at a particular level if the researchers majors, which suggests that the nature of the student rather than felt that the student could think about a topic and operate the application is the most important variable in supporting within a problem at that level. The rating was not an indication back transfer. Of course, self-selection of students into majors to which level a student would naturally operate when complicates this analysis. answering a question but rather an indication of the highest level a student could reasonably operate when answering a Finally, the coding for confidence demonstrated that question. engineering majors were in general more willing to try problems when they didn’t know how to work them. Again, The overall rating for each student was assigned based on a self-selection is an issue here. But it appears that in comparing composite of the student’s individual ratings on each question. practical vs. abstract extensions of mathematical ideas, the Once all students had been rated, the percent of engineering difference lies less in terms of supporting back transfer and majors and the percent of mathematics education majors at more in terms of developing confidence to persist in problem each level were compared to results after Differential Equations solving in the face of uncertainty. to see if seniors demonstrated deeper understanding (showing back transfer) or lesser understanding (demonstrating forgetting IV. FUTURE WORK of basic concepts). Additional research will need to be done to identify what 2) Second pass (quantitative) conditions support development of conceptual understanding As a check to prevent the preconceptions of the researchers in general, and back transfer in particular. It seems clear from from blinding them to ideas expressed by the students, we also the preliminary results that these conditions depend not just on used a quantitative data-mining approach to analyzing the the type of application but also on the student. This involves interview data. The transcriptions were used to create a matrix not just their academic abilities, but also their attitudes, where each entry was the total number of times each student said each word. So the sixth entry in the eighth row would be especially toward working with problems where the solution the number of times the eighth student said the sixth most technique is not immediately obvious. The connections commonly used word from the interviews. A non-negative between transfer, back transfer, self-efficacy, and conceptual matrix factorization was then used on the resulting “word bags” understanding are not currently well understood. to identify different vocabularies used by different students. The researchers were somewhat surprised when the REFERENCES vocabularies did not indicate differences between different [1] J. Bransford, A. Brown, and R. Cocking (eds.), How People Learn: Brain, Mind, Experience, and School, National Academy Press, majors, but did identify groups based on words linked to Washington, DC, 2000, pp. 51-78. confidence or willingness to participate in answering questions [2] C. Hohensee, Backwards transfer: How mathematical understanding when they weren’t sure of the answer such as “don’t” and changes as one builds upon it, Doctoral Dissertation, San Diego State “know”, “understand”, etc. University, 2011. [3] A. Bennett, X. Nguyen, and T. Moore, “A longitudinal study on students 3) Third pass (qualitative) development and transfer of the concept of integration,” Proc. Amer. Once the quantitative analysis suggested student Soc. Eng. Ed., Vancouver, BC, 2011. confidence in dealing with questions where they were unsure [4] E. Dubinsky and M. McDonald, “APOS: A constructivist theory of of the answer was important, the interviews were divided into learning in undergraduate matheamtics education,” in D. Holton (ed.), segments of information (sentences or phrases as appropriate) The Teaching and Learning of Mathematics at University Level: An and coded for topic, level of understanding, current ICMI Study. Series: New ICMI Study Series, vol. 7 (pp. 273-280), Dordrechet: Kluwer. representation, confidence, and willingness to participate.