Auckland University of Technology, New Zealand
Auckland University of Technology, New Zealand
Auckland University of Technology, New Zealand
Auckland University of Technology, New Zealand
Uppsala University, Sweden
University of Turku, Finland
University of Technology Sydney, Australia
Clear, T., Whalley, J., Robbins, P., Philpott, A., Eckerdal, A., Laakso, M., & Lister, R. (2011). Report on the final BRACElet workshop. Journal of Applied Computing and Information Technology, 15(1). Retrieved October 28, 2020 from http://www.citrenz.ac.nz/jacit/JACIT1501/2011Clear_BRACElet.html
This paper reports on the thirteenth and final BRACElet workshop. In this paper we provide a brief retrospective review of the workshops and the findings that have resulted from this multi-institutional multinational investigation into the teaching and learning of novice programmers. Subsequently we report on the work undertaken during the final workshop and then discuss future avenues for research that have evolved as a result of the BRACElet project.
novice programmers, SOLO, Bloom, assessment
The workshop reported in this paper is the final workshop in a series of workshops. A brief outline of the outcomes of the previous workshops, partly quoting Clear, Philpott, and Robbins (2009), is presented below. A fuller discussion of the BRACElet projects findings can be found in Section 2.
December 2004: The inaugural workshop began at Auckland University of Technology (AUT) with a review of the findings of the Leeds working group which had recently been published (Lister et al., 2004). The participants decided to replicate and further the Leeds Group study by basing the question set design on an accepted educational model. The participants selected the revised Bloom's taxonomy (Anderson et al., 2001) and then devised a question set, including some of the original Leeds questions, which mapped to the "understand" cognitive process dimension. The questions were designed to test the students’ ability to reason about code and associated abstractions. After the workshop, as part of a pilot study, these questions were included in the final exam of the introductory programming (CS1) course in a number of participating institutions.
July 2005: During the second workshop held at the 18th Annual NACCQ conference in Tauranga, participants analysed, evaluated and published the results of the pilot study (Whalley et al., 2006). The results were also employed to further hone the research instrument. A second educational model, SOLO taxonomy (Biggs & Collis, 1982) was introduced to analyze and classify student responses to questions that required them to explain the purpose of a piece of code. By the end of the second workshop a toolkit had been developed that allowed the participants to undertake a fully-fledged study at their respective institutions. The results from this work were disseminated at ACE 2006 (Whalley et al., 2006) and further papers were authored over the next few months by subgroups.
March 2006: A third workshop took place at AUT during which the group developed a prototype ‘common framework' (Lister, Whalley, & Clear, 2006) that allowed researchers to compare and contrast studies undertaken within BRACElet, but that also gave them the flexibility to tailor research to their particular interests.
July 2006: A fourth workshop was held at the 19th Annual NACCQ conference in Wellington. The common framework was reviewed and subsequently refined for the next phase of the project, an investigation into the code writing skills of novice programmers.
July 2008: A fifth workshop was held at the 20th Annual NACCQ conference in Nelson. It was at this workshop that a new type of question was introduced to the common framework pool, namely Parsons’ Problems (Parsons & Haden, 2006). This type of problem was devised by Parsons to help students acquire competence with the structural syntax of programming. Parsons’ problems, in essence, provide all the code required to solve a given problem but require the students to order (and possibly select then order) the lines of code to form a correct solution. A helpful paper on the use of these questions in assessment is provided by Denny, Luxton-Reilly and Simon (2008). The BRACElet group initially introduced the problem as an assessment question that we believed was assessing an intermediate stage between code writing and code reading. This assumption was proven incorrect in later work detailed in Section 3 when we investigated a possible hierarchy of programming skills using the common framework.
December 2007: A sixth workshop was held at AUT University, sponsored by ACM & SIGCSE. Contributed data was analysed. New research instruments were developed to allow for the evaluation of novice programmers' program-writing skills and to enable comparisons to be made with their program reading skills. This toolkit was to be subsequently implemented at the participants' institutions in Australasia, the data analysed and joint publication(s) produced.
January 2008: The seventh workshop was held at Wollongong at the ACE 2008 conference. This was the focal meeting for gaining Australian input related to the Carrick Institute Joint Associate Fellowship of Raymond Lister and Professor Jenny Edwards. An action research cycle was initiated at this meeting with the goal that this BRACElet iteration would: consolidate the "explain in plain English question"; generate new questions; examine any gender effects; examine any differences between international and local students (will have to be careful to differentiate between local " native speakers" and local ESL); relate answers to SOLO levels; examine differences between Undergraduate and Postgraduate; help develop an ideal exam paper.
July 2008: The eighth workshop was held at AUT University concurrent with the NACCQ Conference. The purpose of the workshop was to analyse assessment data from novice programmers, and thus further the inquiries into how novice programmers comprehend and write computer programs. Results from the workshop were suggested SOLO categories applicable to program writing tasks, and a modified SOLO classification scheme that combined reading and writing.
September 2008: The ninth workshop was held as a two day affair in connection with the ICER 2008 conference, in Sydney, and was supported by the Australian Learning and Teaching Council (formerly Carrick Institute). It was here that we started to talk about the value of replication studies to verify earlier findings. This workshop focused on the discussion of a relationship, if any, between reading/explaining code and writing code, and took its first step away from a folk-pedagogic debate, toward an evidence-based debate. It was at this juncture that the focus in BRACElet began to move beyond whether there was any relationship between reading/explaining code and writing code, to a discussion of the nature of the relationship – was it a causal relationship, or was it an indication of an underlying, yet to be identified, deeper skill?
January 2009: a half-day event, held at the end of the ACE2009 conference, in Wellington. This workshop was largely a reflection and discussion of progress to date. Because the data collection and analysis methods were well established it was at this point we decided to publish a paper to specify the next phase of data collection and analysis (Whalley & Lister, 2009). We also reviewed the draft of the group's first replication study paper (Lister, Fidge & Teague, 2009). This paper replicated work by Lopez et al. (2008), using data collected from a different exam paper, conducted at a different educational institution. This study was later published (Lister, Fidge & Teague, 2009), but used a different statistical approach from Lopez et al. (2008). The results presented at the workshop were consistent with Lopez et al. (2008). Furthermore, the similarities and differences between this paper and the Lopez et al. paper led to a good discussion on what is essential, and what is not essential, to such a BRACElet study.
June 2009: A working group at ITiCSE in Paris provided the forum for the eleventh BRACElet workshop. The group replicated earlier analyses using a far broader pool of naturally occurring data, refined the SOLO taxonomy in code-explaining questions, extended the taxonomy to code-writing questions, furthered some earlier studies on students’ ‘doodling’ while answering exam questions, and explored a further theoretical basis for the work that until this point had been primarily empirical (Lister et al., 2009).
January 2010: The twelfth workshop was held in Brisbane as a half day event at the end of the ACE 2010 conference. BRACElet activities (now known as "threads”) for the coming year were identified. The threads were as follows:
The key findings of the BRACElet project are founded on the analysis of naturally occurring data (see: Lister et al., 2010). First we present a list of the core findings related to novices learning to program.
The BRACElet project has also contributed to research methodology in the field. They have published guidelines and toolkits for managing longitudinal multi-institutional multi-national projects (Whalley, Clear & Lister, 2007), in addition to providing methods that allow for the analysis of assessments and students code.
The purpose of the final workshop was to extend the work on theory initiated at the Paris ITiCSE Working group in 2009, and to launch an ongoing programme of research through the Software engineering Research Laboratory (SERL) at AUT into the novice to expert programmer continuum, the “N2Expert programmer continuum”. This project will be titled the “NExpertise” project. The scope would extend from ‘programming in the small’ to ‘programming in the large’, and the work of professional programmers. Thus the work has the goal of contributing to broader and deeper understandings and impacting practice for both educators and software practitioners.
This two day workshop was held at Auckland University of Technology in September 2010, and was attended by twelve participants from New Zealand, Australia, Finland and Sweden.
The workshop began with a keynote given by Dr. Anna Eckerdal: “The delicate art of connecting theory and practice: A variation theory study in programming education”. The second keynote of the workshop was given by Dr. Raymond Lister. He presented recent work, which will appear as two papers in the Australian Computer Education Conference in January 2011. His talk included a discussion of novice programmer reasoning and Concrete and other Neo-Piagetian theories (Lister, 2011) and work arising from BRACElet work on early relational reasoning in the novice programmer (Corney, Lister & Teague, 2011). The two keynotes were followed by a presentation on the preliminary data patterns and potential research designs for studying the development of expertise in novice programmers (BRACElet → NExpertise), arising from the data to be utilised in this workshop, given by Mike Lopez.
Following the presentations, five potential research questions the workshop leaders had prepared relating to previous BRACElet work were discussed. Subsequently the workshop participants formed three groups that focused on three of these questions.
Data provided, from courses spanning CS0, CS1, CS2 and C3, by the participants was then inspected in order to identify appropriate data and a preliminary analysis was undertaken to establish the feasibility of further each avenue of research.
Group 1: The group intended to investigate results from a previously performed phenomenographic study (Marton & Booth, 1997) on novice students’ understanding of the concept object and class (Eckerdal & Thuné, 2005). The results of the phenomenographic analysis gave an outcome space with three categories which illustrate qualitatively different understandings among the students in the study. Students demonstrated that they focused on one or more of the following critical aspects (Marton & Tsui, 2004) related to the two concepts. First, a text representation of the concepts, objects and classes are pieces of text with a certain structure. Next, an understanding that objects are active when a program is executed, and the actions are described in the methods of the class. Finally a few students also focused on the modelling aspect of the concepts. The phenomenographic study named these critical aspects TEXT, ACTION, and MODEL respectively.
The aim of this group's investigation was to discover if it is possible to statistically refute the existence of a hierarchy in the development of novice students’ understanding with respect to the critical aspects TEXT, ACTION, and MODEL. The members in the group hypothesized that there exists an order between the three aspects in terms of how students learn them. The hierarchy starts with the TEXT aspect, then the ACTION and finally the MODEL aspect. The members further assumed that there is a common belief among educators that students learn to program according to this order. This implies e.g. that the ACTION aspect cannot be understood before the students understand the TEXT aspect. Such a belief can influence the way educators teach programming, and if the data refutes this belief, it is important information to the community of educators.
Group 2: There was a discussion among all participants of the workshop on research on misconception. Some researchers claim that, at least in Computer Science, there exist misconceptions that are so common that they are worth studying. Group 2 was interested in investigating students misconceptions of assignments (=) in Java, and how this relates to the representation of equality in mathematics. Are there any patterns in the errors related to assignment in students’ exams that can be understood in relation to the mathematical equality symbol and its conception? Closely related to assignment is state which is suggested by Shinners-Kennedy (2008) to be a threshold concept (Meyer & Land, 2005).
Group 3: The group was interested in performing a longitudinal study on individual students. The hypothesis was that students who perform well in introductory courses also perform well on subsequent courses. The group began to investigate the hypothesis by looking at students’ results in later courses, and correlating it to results for the same students’ in earlier courses. The research question is interesting in relation to common understandings among educators that students are “good” or “bad” and that students remain “good” or “bad” throughout their education. Students who perform well in later courses have thus probably performed well in earlier courses. If the analyses showed the opposite, or no such pattern, the hypothesis is refuted.
Group 1: The group reported that they had tried several strategies on several data sets before one of the group members suggested a functioning strategy. The strategy the group used was to limit the analysis to try to refute a hierarchy between the first two aspects, TEXT and ACTION. The reason for this decision was mainly that the group found no exam question in the CS1 exams that reflected the MODEL aspect. If, however, the statistical analysis of the data refutes a hierarchy between the TEXT and the ACTION aspect, the hypothesis of a hierarchy between all three aspects is refuted.
The exam question that was chosen to illustrate the TEXT aspect was a Parsons’ puzzle question, and the question that illustrates the ACTION aspect was a question where a class Bus was given and the students had to predict return values of method calls after several other methods were called. The first question focused on code and the structure of the code, while the second question focused on the program execution.
A pilot analysis on 15 randomly selected exams was performed where all participants in the group were actively involved. A classification scheme was agreed upon. The results from the pilot study seemed to refute the hypothesis of a hierarchy between the two aspects TEXT and ACTION.
The group decided to split the analysis of the rest of the exam between the members in the group. The group will tentatively continue the communication in the group via email.
Group 2: The group examined interesting findings from an initial analysis of an end of semester exam for pre-degree students. A three part question involved nothing more than assignment statements, as follows:
Students had to state the value in each variable once the code had been executed.
For i, only half the students correctly gave the answer 4, 4. 30% gave r = 2, s = 4, seemingly ignoring the last line of code. For ii, 60% gave p = 8, q = 1 with only 12% giving the correct answer. For iii, 58% gave x = 7, y = 5, z = 3 with only 14% giving the correct answer. For the last two parts, most students seemed to take the values of the variables from the initial assignment and not track their changes.
In addition, in a CS1 exam, 6 of 10 students did not answer correctly on the question “What are the values of variables a and b after the following code has been executed?” The code is similar to ii) in the table 1: a = 2, b = 3, a = b, b = a. The wrong answer many students gave was: a = 3 and b = 2; they understood that the purpose was to swap values of two variables. The group will report interesting findings from an initial analysis of CS0 and CS1 exams.
The group will continue to analyse the problem with assignment in programming, and to point out the existing problems to CSE community. The group plans to design multiple experiments to investigate underlying issues behind this phenomenon. For example, we will conduct an experiment with two groups of students, a control group in which the participants answer the questions with pen and paper, and a second group which will utilize the same exercises in a digital learning environment (ViLLe, see Kaila et al., 2010). The goal is to find out what is the effect of the media (paper vs. digital) behind this phenomenon.
Group 3: The analysis emanated from a previous study on CS3 students’ assignment responses (Philpott, Clear & Whalley, 2009). This analysis categorised the students according to the SOLO taxonomy as previously interpreted (Clear, Philpott, Robbins & Simon, 2009/2010). Answers from certain assignment responses were categorised either as Unistructural, Multistructural, or Relational. The group identified students that had participated in the previous study, and looked for the same students’ exams from CS1 and CS2. The initial findings from the analysis on these students’ results for previous courses show a group of students who performed better on the CS3 assessment than on the earlier exams: the students were categorised as Relational on the SOLO scale on CS3, and lower on the previous exams. They also found a group of students who performed worse on the CS3 assessment, Unistructural on the SOLO scale, than on the previous exams. Preliminary findings show no patterns in the data. This can be due to the fact that many of the students who were analysed during the workshop were not ‘traditional’ students, but came from a wide variety of backgrounds. The group concluded that they need to analyse more data before they had any findings, and note that it is hard to longitudinally track students’ learning. If the lack of pattern still applies when a larger data set is analysed, this will refute the hypothesis that it is possible to predict from exam results how students will perform later in their education. The group also reported that there was no data available to follow the students’ performance in detail.
As is typical with BRACElet workshops an agenda for follow-on work and publications has been charted (even for this “Final” workshop). All collaborations inevitably go through their phases and the parties to BRACElet have decided to formally bring the project to an end. This is timely with the 2010 publication of the results of the ALTC fellowship (Lister & Edwards, 2010), and reflects a diverging of interests apparent at the Jan 2010 workshop where new and separate threads have been spawned.
The formal cessation of the BRACElet project enables the Co-Principal Investigators to pursue their own lines of research, and begin some fresh enquiries. For the Auckland University of Technology Co-PI’s we see this direction spanning the full novice to expert continuum through the NExpertise project, in the hope that by better understanding the work of experienced professionals and intermediate programmers we may gain greater insights into how the fuller process of developing programming knowledge and skills develops and how these skills might best be taught and assessed. The differences between programming in the small and programming in the large are quite marked, so we will have many new questions, activities and contexts to consider. This will also raise new methodological challenges in determining how to study the more advanced programmer, in scoping research studies, in defining the artifacts required, and deciding how they may be collected and analysed.
There certainly seems to be scope for much fruitful work and we hope we can build another community of scholars working collegially on the NExpertise project, applying a model yet to emerge but similar to that we have promoted through the BRACElet studies.
In closing we wish to farewell the good ship BRACElet and thank all our colleagues who have sailed in her.
Anderson, L.W. & Krathwohl, D.R. (Eds.). (2001). A taxonomy for Learning, teaching, and assessing: A revision of Bloom's taxonomy of educational objectives. New York: Addison Wesley Longman.
Biggs, J. B., & Collis, K. F. (1982). Evaluating the quality of learning: The SOLO taxonomy (Structure of the Observed Learning Outcome), New York: Academic Press.
Clear, T., Philpott, A., Robbins, P. & Simon. (2009/2010, Dec/Jan). Report on the Eighth BRACElet Workshop: BRACElet Technical Report 01/08 AUT University, Auckland. Bulletin of Applied Computing and Information Technology, 7(1), Retrieved November 9, 2010 from http://www.naccq.ac.nz/bacit/0701/2009Clear_BRACElet_Report.htm
Clear, T., Whalley, J., Lister, R.F., Carbone, A., Hu, M., & Sheard, J., et al. (2008). Reliably Classifying Novice Programmer Exam Responses using the SOLO Taxonomy, National Advisory Committee on Computing Qualifications, Auckland, New Zealand, July 2008 in 21st Annual Conference of the National Advisory Committee on Computing Qualifications, Samuel Mann & Mike Lopez (Eds.), National Advisory Committee on Computing Qualifications, New Zealand, 23-30.
Denny, P., Luxton-Reilly, A. & Simon, B. (2008). Evaluating a new exam question: Parsons problems. In Proceedings of the Fourth international Workshop on Computing Education Research (ICER '08). ACM, New York, NY, USA, 113-124.
Eckerdal, A. & Thuné, M. (2005) Novice Java Programmers' conceptions of 'Object' and 'Class' and Variation Theory. In Proceedings of the 10th Annual SIGCSE Conference on Innovation and Technology in Computer Science Education, pp 89-93, Lisbon, Portugal
Kaila, E., Rajala, T., Laakso, M.-J. & Salakoski, T. (2010). Long-term Effects of Program Visualization. In Proceedings of the 12th Australasian Computing Education Conference (ACE 2010), January 18-22, 2010, Brisbane, Australia.
Lister R., Adams, E. S., Fitzgerald, S., Fone, W., Hamer, J., & Lindholm, M., et al. (2004). A multi-national study of reading and tracing skills in novice programmers. SIGCSE Bulletin 36(4), 119-150.
Lister, R., & Edwards, J. (2010). Teaching Novice Computer Programmers: bringing the scholarly approach to Australia. A report on the BRACElet project. Sydney: Australian Learning and Teaching Council, retrieved 24 Nov 2010 from http://www.altc.edu.au/resource-teaching-novice-computer-programmers-bringing-scholarly-approach-australia-report-bracelet-
Lister, R., Whalley, J. & Clear, T. (2006). For Discussion: A Framework for a Meta-Project on Students Programmers (BRACElet Technical Report No. 0106). Auckland: Auckland University of Technology.
Lister, R., Simon, B., Thompson, E., Whalley, J. & Prasad, C. (2006). Not seeing the forest for the trees: novice programmers and the SOLO taxonomy, SIGCSE Bulletin, 38: 3, 118-122.
Lister, R., Fidge, C & Teague, D. (2009). Further evidence of a relationship between explaining, tracing and writing skills in introductory programming. SIGCSE Bulletin. 41, 3 (July 2009), 161-165.
Lister, R, Clear, T., Simon, Bouvier, D. J. Carter, P, & Eckerdal, A., et al. (2010). Naturally occurring data as research instrument: analyzing examination responses to study the novice programmer. SIGCSE Bulletin. 41:4, 156-173.
Lopez, M., Whalley, J., Robbins, P. & Lister, R. (2008). Relationships between reading, tracing and writing skills in introductory programming. In Proceedings of the Fourth international Workshop on Computing Education Research (ICER '08). ACM, New York, NY, USA, 101-112.
Marton, F. & Booth, S. (1997). Learning and awareness. NJ: Lawrence Erlbaum Ass.
Marton, F. & Tsui, A. (2004), editors. Classroom Discourse and the Space of Learning. NJ: Lawrence Erlbaum Ass.
Meyer, J. H. & Land, R. (2005) Threshold concepts and troublesome knowledge (2): Epistemological considerations and a conceptual framework for teaching and learning. Higher Education, 49(3):373-388
Parsons, D. & Haden, P. (2006) Parsons' programming puzzles: a fun and effective learning tool for first programming courses. InProceedings of the Eighth Australasian Computing Education Conference, Hobart, Australia, 157-163.
Philpott, A., Robbins P, & Whalley J.L. (2007). Assessing the steps on the road to relational thinking, In Proceedings of the 20th Annual Conference of the National Advisory Committee on Computing Qualifications, Nelson, New Zealand. Mann, S. & Bridgeman, N. (Eds.), 286.
Philpott, A., Clear, T., & Whalley, J. (2009). Understanding Student Performance on an Algorithm Simulation Task: Implications for Guided Learning. In M. Guzdial & S. Fitzgerald (Eds.), In Proceedings of the 40th ACM Technical Symposium on Computer Science Education, Chattannooga, Tennessee: ACM, 408-412.
Sheard, J., Carbone, A., Lister, R.F., Simon, B., Thompson, E., & Whalley, J. (2008). Going SOLO to Assess Novice Programmers, ACM Special Interest Group on Computer Science Education Conference, Madrid, Spain, June 2008 In Proceedings of the 13th Annual SIGCSE Conference on innovation and Technology in Computer Science Education, June Amillo, Cary Laxer, Ernestina Menasalvas, & Alison Young (Eds.), Association of Computing Machinery, New York, NY, USA, 209-213.
Shinners-Kennedy, D. (2008) The Everydayness of Threshold Concepts: State as an Example from Computer Science, in: Threshold Concepts within the Disciplines, Land, R., Meyer, J.H.F. and Smith, J., (eds), Sense Publishers, Rotterdam, 119−128
Thompson, E., Luxton-Reilly, R., Whalley, J., Hu, M. & Robbins, P. (2008). Bloom's taxonomy for CS assessment. In Proceedings of the tenth conference on Australasian computing education>, Simon & Margaret Hamilton (Eds.), 78. Australian Computer Society, Inc., Darlinghurst, Australia, 155-161.
Venables, A., Tan, G. & Lister, R. (2009). A Closer Look at Tracing, Explaining and Code Writing Skills in the Novice Programmer, Fifth International Computing Education Research Workshop, Berkeley, CA, USA, August 2009 In Proceedings of the Fifth International Computing Education Research Workshop, Clancy, M., Caspersen, M.,& Lister, R. (Eds)., Association for Computing Machinery, New York, NY, USA, 117-128.
Whalley, J. L., Lister, R., Thompson, E., Clear, T., Robbins, P., & Kumar, P. K. A., et al. (2006). An Australasian Study of Reading and Comprehension Skills in Novice Programmers, using the Bloom and SOLO Taxonomies. Proceedings of the Eighth Australasian Computing Education Conference (ACE2006), Hobart, Australia CRPIT, 52, 243-252.
Whalley, J., Clear, T. & Lister, R. (2007). The Many Ways of the BRACElet Project.
Bulletin of Applied Computing and Information Technology, 5(1). Retrieved December 13, 2010 from http://www.naccq.ac.nz/bacit/0501/2007Whalley_BRACELET_Ways.htm
Whalley, J., & Lister, R. (2009). The BRACElet 2009.1 (Wellington) Specification, Eleventh Australasian Computing Education Conference, Wellington, New Zealand, January 2009 in Eleventh Australasian Computing Education Conference, Margaret Hamilton & Tony Clear (Eds.), Australian Computer Society Inc., Australia, pp. 9-18.