title: Structured Discussion in Computing teaching and learning author: Module 1 / 33347617 mainfont: Helvetica Neue linestretch: 1.5 papersize: a4paper geometry: margin=1.5in header-includes: \renewcommand{\abstractname}{Declaration}

abstract: '\noindent By submitting this, I declare it to be my own work. Any quotations from other sources have been credited to their authors. Approximate word count: 1800'

The structured discussion technique (Northedge, 1975) is a standard device for many reasons, with its advantages including group dynamics, a manageable load for the facilitator and grounding the learning process in the experience and skills of the learners (Gibbs, 1991). I have chosen to reflect on this technique specifically because over the course of about 16 years in Higher Education, as a student, researcher and lecturer, I have rarely if ever encountered it in the form described.

Structured discussion involves working on a problem or discussion point in four phases. The first phase involves working alone, and permits all students, whatever their level of experience or prior attainment, to engage with the problem. The second phase has students working in pairs, enabling them to compare and contrast in detail their approaches to the problem, and (by virtue of the pairs involving only one other person) forcing them to engage to avoid awkward silence. The third phase has pairs combine into slightly larger groups (fours, or sometimes sixes), introducing new ideas while aiming not to impose a new discussion dynamic. The final phase is a plenary reporting and discussion activity, where the groups report (a subset of) their findings and the students and facilitator attempt to produce a synthesis of the ideas generated through the activity.

This description of structured discussion is simplistic, and many variants of the technique exist. This essay is not the place for an exhaustive study of the variants and the research into its effectiveness. Instead, I will consider here the reasons for my not having consciously noticed it as a device previously, what aspects of my disciplines militate against its use, and the potential for modifiying its details for application in my own teaching.

Although I am now based in a department of Computing, my academic background was initially as a physicist: specifically, undergraduate studies in Natural Sciences, specializing in Physics and Maths from my second year; my doctoral studies were in a department of Applied Mathematics, and the material I worked on for my thesis regarded the observational consequences of exotic theories of matter on the night sky today. My move to Computing came after the doctorate, joining a research group into aspects of music and computing.

The disciplines have much in common as well as aspects of contrast. For the purposes of this essay, perhaps the aspect that they most have in common is the question of foundational material: in Physics, it is difficult (though not impossible) to graduate, let alone become an adept scientist, without mathematical fluency; it is even harder in Computing to graduate without the ability to programme to some degree, and the most obvious graduate destinations require programming ability, often testing it themselves at ‘assessment centres’ rather than taking a degree or degree transcript as evidence (on which much more could be said). One contrast between the disciplines is that there is a reasonable expectation at undergraduate entry that a student reading Physics will have reasonable mathematical fluency (a good grade at ‘A’-level Mathematics or equivalent) – though according to the Making Mathematics Count report of 2004, this expectation is weakening; there is no corresponding expectation of programming experience at undergraduate entry for Computing, and indeed the ‘A’-level qualifications in Information Technology and similar are often considered to be actively unhelpful for further study.

So, one reason why I have had limited exposure to structured discussion is that a significant amount of time in undergraduate teaching in both Physics and Computing is spent in delivering, exercising and extending foundational disciplinary skills: skills, furthermore, which are neither a matter of opinion nor open to substantial argument: they are simply necessary for that discipline, much as reading and writing are necessary for all degrees. That is not to say that the preponderance of the foundational material excludes aspects of opinion in the learning in the discipline: experimental or protocol design, communication and presentation are also of importance – but the depth of foundations needed to get there are sufficient to consume much of the attention of teachers and students. Indeed, one comparable undergraduate programme (Androutsopoulos et al., 2014) has recently altered its curriculum such that the first-year assessment consists entirely of determining which of 119 itemized observable behaviours (individual aspects of foundational skills) each student has demonstrated that they can perform.

Probably the setting in my previous experience in which the structured discussion technique could have been most directly applied was paper reading groups, where a research group would meet regularly (typically over lunch) and discuss a previously-specified published research article in the general research area. These sessions met most of the criteria for structured discussion to be useful, and were designed to have a single facilitator, responsible for leading the discussion; typically the number of people attending the groups is lower than the twelve or so suggested (Gibbs, 1991) for the technique’s effectiveness.

There are elements of undergraduate teaching, both assesssed and unassessed, where structued discussion is a natural fit; part of the final-year individual project involves group sessions where the students are ecnouraged to reflect on their own practice and their hopes and expectations for their future careers, and there is much space in that area for the structured discussion technique to be applied. To examine the applicability in the context of the more knowledge- and skills-based settings, however, we will have to consider individual aspects of structured discussion.

For example, one of the potential benefits of structured discussion is the development of transferrable skills by the very process of using it: skills such as active listening, group communication and effective presentation. These skills are presumably rarely listed as overall learning outcomes of programmes, though they might be listed at the module level and would contribute to overall programme learning outcomes; structured discussion would be one tool among many to address those learning outcomes, to be used where appropriate.

Another of the benefits of structured discussion is the elicitation of a group synthesis on a particular question from individual contributions, each critically assessed, sharing some of its nature with the “wisdom of crowds” effect where the overall estimation of the group is typically better than any individual estimation. One pedagogical example of this is in setting of open-ended examination questions, where the individual answers from students could be usefully combined into a near-complete solution to a real problem in practice (Anderson, 2007); I have seen this effect in group work, where in a simulation of computer networks using playing cards a Foundation-level class ‘discovered’ fundamental aspects of both collision-avoidance through progressive backoff and token-ring Ethernet.

The remaining question to address in this essay is whether, and if so to what extent, structured discussion or elements from it can be used to teach the more foundational computing skills. There is some evidence from industrial practice that there are benefits to working in pairs, both to the individuals involved and to the software product being developed. In the set of programming practices known as “Extreme Programming” (Beck, 2000), one technique used is known as “pair-programming”, where two professionals (sometimes of equal experience, sometimes one more senior than the other in a mentoring relationship) work in pairs, where there is only one keyboard and one programmer types the programme at the explicit direction of the other. In the pure form of this, if the programmer currently typing wishes to contribute to the design of the programme, he or she must relinquish the keyboard to the other. Programming in this style could be incorporated into lab sessions, though as with many practices the benefits are likely to be felt after a substantial time of immersion in the practice rather than somewhat superficial exposure.

Another weakness of the pair-programming approach in pedagogy is that it somewhat presupposes that at least one of the pair is competent to drive the programming, or at least has spent some time working on a design in advance – analogous to the “working alone” first phase of structured discussion. Typically this is not the case for the cohorts of Computing students I have taught, no matter how far in advance the lab material is made available. If elements of structured discussion or group work is to be incorporated into lab work in this way, it might be possible to require students attending a practical session to complete a pre-practical exercise (Carnduff and Reid, 2003) before being admitted, though there will be a tension between enforcing this rule strictly and the pair programming being effective, perhaps particularly for the less engaged students who would benefit the most from peer mentoring.

Another model for incorporating elements of structured discussion into laboratory work is to have the students pair off to develop the design of programmes rather than the detailed implementation. This is probably a more natural fit for the pairing, but again the dependency on some individual work being done before the pairs being formed is problematic: the breadth of experience in the laboratory session is such that some students are likely to be able to complete the assigned work individually with no need for peer or tutorial support, while another section of the cohort is likely to be floundering. As one student has put it, “there are three groups in the class – those who are bored, those who are OK and those who are lost.” (quoted in Fry et al., 2008, chap. 16).

We have seen in this reflection that there are a number of places in the Computing curriculum where elements of structured discussion could be used to help in the delivery of foundational and technical material. The Computing discipline is different from many in that it has traditionally had as a prerequisite to effective practice a skillset that is highly technical, atomistic as well as holistic, and not generally possessed by undergraduate entrants, and hence has typically taken up substantial amounts of learning time, and hence it is perhaps natural that structured discussion is a novelty to me in teaching and learning.

The idea of incorporating elements of structured discussion into lab sessions, if it can be refined and made to work, seems to me to be the most promising; there is a view that Computational Thinking (Wing, 2006) – not the medium of its expression (programming) – is the fundamental academic aspect of the Computing discipline, and encouraging the development of such Computational Thinking through communication and explanation amongst peers has potential. One difficulty may come in measuring the effect, for it remains true at the moment that the most clear expression of Computational Thinking is through programming computers, and consequently it is difficult – though not impossible – to assess the two skills separately. Exploring those details will be the subject of further work.

Anderson, R (2007) Phishing, students, and cheating at the lottery [accessed 23 August 2014]

Androutsopoulos, K, Gorogiannis, N, Loomes, M, Margolis, M, Primiero, G, Raimondi, F, Varsani, P, Weldin, N and Zivanovic, A (2014) ‘A Racket-Based Robot to Teach First-Year Computer Science’

Beck, K (2000) ‘Extreme Programming Explained’, Addison-Wesley Professional

Carnduff, J and Reid, N (2003) ‘Enhancing Undergraduate Chemistry Laboratories’, Royal Society of Chemistry, London

Fry, H, Ketteridge, S and Marshall, S (2008) ‘A Handbook for Teaching and Learning in Higher Education’, Routledge

Gibbs, G (1991) ‘Teaching Students to Learn: A student-centred approach’, Open University Press

Making Mathematics Count (2004) HMSO, London [accessed 22 August 2014]

Northedge, A (1975) ‘Learning through discussion in the Open University’, Teaching at a Distance, Open University, No.2

Wing, J M (2006) ‘Computational Thinking’, Communications of the ACM 49(3)