Constructivist Approaches to Teaching Programming

(A Critical Review)

By

Matthew Dean

Dissertation submitted to De Montfort University in partial fulfilment of the

requirements for the Degree of Master of Arts in Education Practice

September 2017

Name: Matthew Dean

P Number: P06067159

Course: MA Education Practice

Module Name: Research and Dissertation / Major Project

Module Code: EDUC5022

Supervisor’s Name: Dr. Motje Wolf

Due Date: Friday 8th September 2017

Word count 21,272

Contents

Acknowledgements...................................................................................................... i

Abstract....................................................................................................................... ii

Chapter One: Introduction...........................................................................................1

Constructivism and Problem Based Learning..........................................................1

Curriculum Design...................................................................................................2

IMAT1604 Visual Web Development (First Year)................................................2

IMAT2204 Project Management and Development (Second Year).....................4

IMAT3903 Team Development Project (Final Year)............................................7

The Student Voice...................................................................................................8

Dissertation Structure..............................................................................................8

Chapter 2: Literature Review......................................................................................9

Introduction..............................................................................................................9

The Programming Problem.....................................................................................9

Building Mental Models.........................................................................................13

Constructivism as Pedagogy.................................................................................17

Criticism of Problem Based Learning....................................................................19

Conclusion.............................................................................................................20

Chapter Three: Methodology and Research Methods..............................................21

Introduction............................................................................................................21

Research Methodology..........................................................................................21

Data collection methods........................................................................................24

Ethics.....................................................................................................................25

The Role of the Literature Review.........................................................................26

Questionnaires......................................................................................................26

Programming Preferences Survey.........................................................................27

Security..................................................................................................................27

Informed Consent..................................................................................................28

The Questionnaire Design.....................................................................................29

Your Learning........................................................................................................30

The Teaching.........................................................................................................30

Learning Preferences Survey................................................................................32

Focus Groups........................................................................................................34

The Second Focus Group.....................................................................................36

Data Analysis Procedures.....................................................................................36

Theoretical Sampling.........................................................................................36

Coding and Constant Comparison....................................................................36

The Focus Groups.............................................................................................39

Conclusion.............................................................................................................39

Chapter Four: Findings.............................................................................................40

Introduction............................................................................................................40

Overview of Codes................................................................................................40

Notes on Presentation of Findings.........................................................................42

The First Focus Group...........................................................................................43

The Second Focus Group.....................................................................................45

Constant Comparison and the Generation of Candidate Codes............................46

What helps you learn?...........................................................................................47

Barriers to Learning...............................................................................................49

Tutor as Facilitator.................................................................................................51

Use of Real World Problems.................................................................................52

Use of Complex Problems.....................................................................................53

Collaboration.........................................................................................................54

Codes for Discussion at Focus Group...............................................................55

Summary of Second Focus Group........................................................................56

Student Views on Lectures................................................................................56

Student Views on Programming Pathways........................................................57

Detailed Codes for Programming Preferences Survey..........................................58

Introduction............................................................................................................58

What has helped you to learn programming?........................................................58

Teaching and Learning......................................................................................59

The Teaching.....................................................................................................60

The Content.......................................................................................................61

Delivery..............................................................................................................62

Learning.............................................................................................................63

Social Learning..................................................................................................64

The Tutor...........................................................................................................65

The Team...........................................................................................................66

What has hindered your learning of programming?...............................................67

Hindrances.........................................................................................................68

Obstacles to Student Learning...........................................................................69

Barriers to Individual Learning...........................................................................70

Barriers to Collective Learning...........................................................................71

Issues with Social Learning................................................................................71

The Tutor as Barrier...........................................................................................72

The Team as Barrier..........................................................................................73

Problematic Teaching........................................................................................74

Inappropriate Content........................................................................................75

Poor Delivery.....................................................................................................76

Tutor as Facilitator.................................................................................................77

The Tutor as Facilitator......................................................................................77

Mixed Views on Tutor as Facilitator...................................................................78

Agreement on Tutor as Facilitator......................................................................79

Disagreement with Tutor as Facilitator...............................................................80

Use of Real-World Problems.................................................................................81

Use of Real-World Problems..............................................................................82

Agreement with use of Real-World Problems....................................................83

Mixed views on the use of Real-World Problems...............................................84

Use of Complex Problems.....................................................................................85

Views on Problem Complexity...........................................................................85

Agreement on Problem Complexity...................................................................86

Mixed Views on Problem Complexity.................................................................87

Disagreement with Problem Complexity............................................................88

Team Based Problem Solving...............................................................................89

On Team Based Problem Solving......................................................................90

Agreement on Collaboration..............................................................................91

Mixed Views on Collaboration............................................................................92

Disagreement on Collaboration..........................................................................93

Learning Preferences Survey................................................................................94

Chapter Five: Data Analysis and Discussion............................................................97

Introduction............................................................................................................97

The Problem with Problem Based Learning..........................................................97

Pace and Level......................................................................................................98

The Issue of Cognitive Load..................................................................................99

Learning Modes...................................................................................................100

Individual Learning..............................................................................................101

Social Learning and the Value of the Tutor.........................................................101

Social Learning and the Role of Peers................................................................103

Collective Learning..............................................................................................107

Worked Examples and Module Content..............................................................111

Programming Pathways......................................................................................113

Driving Tests and Sprints....................................................................................113

Happy Learning...................................................................................................116

The Role of the Tutor and Instructional Scaffolding.............................................117

Chapter Six: Conclusions and Recommendations..................................................118

Introduction..........................................................................................................118

Problem Based Learning versus Cognitive Load Theory....................................118

Scaffolding and the More Knowledgeable Other.................................................119

Individual learning................................................................................................119

Social learning.....................................................................................................120

The Demise of the Driving Tests.........................................................................120

Assessment and the Content...............................................................................121

Programming Pathways......................................................................................122

Collective Learning..............................................................................................122

Happy learning and The Student Voice...............................................................123

Appendices................................................................................................................... i

Word Clouds............................................................................................................. i

What has helped you to learn programming?....................................................... i

What has hindered your learning of programming?.............................................x

You learn best when the tutor does not help directly.........................................xix

You learn best working on real world programming problems........................xxvii

You learn best by solving suitably complex programming problems..............xxxvi

You learn best by working together in teams.....................................................xlv

First Focus Group Questions..................................................................................liii

Consent Form........................................................................................................ lix

Participant Information Sheet..................................................................................lx

References................................................................................................................ lxii

Acknowledgements

A huge thank you to all of the students who took part in this work, I could not have

done it without you. Many thanks also to Motje my supervisor for her support

throughout this project. Last but not least thanks also to Jacqui my wife for putting

up with me on this for the last two years.

i

Abstract

This project is a critical evaluation of a range of teaching learning and assessment

strategies implemented as part of BSc Computing at De Montfort University. These

approaches have been put in place over the last decade, inspired by ideas drawn

from Problem Based Learning. Using Grounded Theory and a mixed methods

approach an evaluation of the effectiveness of the pedagogy has been undertaken.

This research follows an interpretivist paradigm in that it attempts to understand the

nature of the curriculum from the point of view of the students in the study. The

research uses qualitative methods to capture the student voice across three

undergraduate levels. Coupled with quantitative data and examination of existing

theory, efforts are made to understand each of these elements in the light of the

other. Whilst some aspects of Problem Based Learning, such as use of real-world

problems and collaborative learning find strong favour with the students, students

raise concerns relating to problem complexity and the role of tutor as facilitator. It

appears that the principles of Problem Based Learning do not fully account for the

empirical findings and that these findings may better be understood via Cognitive

Load Theory coupled with Bruner’s concept of scaffolding. Making problem solving

the single primary focus of learning may be counterproductive especially for novice

programmers. The findings strongly indicate the need for available, approachable

tutors as a vital part of the learning process. The tutor appears to be strongly cast in

the role of More Knowledgeable Other in enabling the students to develop their skills

and knowledge. In addition to the priority placed on the role of the tutor the value of

worked examples is highlighted for example by the use of live-code.

Recommendations are made for the revision of first year delivery to concentrate on a

sprint based model of assessment. Based on student recommendations relating to

the design of the first year this model of assessment and feedback is hoped to

stimulate better social interactions between the tutor, students and peers.

ii

Chapter One: Introduction

As Programme Leader, Module Leader and DMU Teacher Fellow, various

approaches have been used to teach programming to undergraduate students at

three levels of BSc Computing. This research aims to explore the effectiveness of

these approaches primarily from the student perspective. There are three areas

under investigation in this research. The first area relates to the underlying

epistemology i.e. constructivism and its implementation as pedagogy via Problem

Based Learning (PBL). The second area will investigate the implementation of these

ideas within the course at module level in an attempt to ascertain what approaches

appear to be working and why. The third area relates to exploring the student

experience of learning programming on the programme. Looking at these three

areas simultaneously will provide an opportunity to reflect on the design of the

curriculum and, it is hoped, create adjustments to the curriculum that are ultimately

beneficial to the students.

Constructivism and Problem Based Learning

PBL draws its theoretical basis from constructivism which suggests that human

knowledge is socially constructed (Boud and Feletti, 2003). Humans construct

knowledge via “functional fictions” providing us with a pragmatic understanding of the

nature of reality (Mahoney, 2004). There may well be a knowable universe “out

there” but our understanding may not map accurately.

1

The driving forces of a PBL curriculum are appropriately complex problems to

provide opportunities for students to construct knowledge (Biggs, 2006). To support

them in knowledge construction, students work together in groups to solve the

problem at hand (Boud and Feletti, 2003). Another principle of PBL is the idea of

tutor as facilitator of learning. By this principle the tutor is not always the best

qualified to lead learning as they are too far advanced in the process of constructing

the knowledge (Hoover, 2007). Student learning should also be apprenticeship

based, for example, if a student is learning to be a programmer then they should

encounter and solve real world programming problems (Savin-Baden, and Major,

2004).

Curriculum Design

The next element for consideration is to explore the effectiveness of the curriculum

design. Much module development has grown from initial work exploring work-based

simulations for teaching programming. As the taught content has developed it has

turned into a three year plan for teaching programming running across three

modules; IMAT1604 Visual Web Development, IMAT2204 Project Management and

Development and IMAT3903 Team Development Project.

IMAT1604 Visual Web Development (First Year)

Background and StructureVisual Web Development (VWD) is a first year undergraduate module teaching

three-tier, C-Sharp database-programming to novices on Computing. The module

works on the assumption that students need to first understand the terminology and

concepts of programming before they are in the position to think and act like a

programmer. By the end of the module some students may be able to solve small

programming problems but many will not. All students should however understand

key concepts related to programming and be able to write key code structures such

as loops, functions etc.

2

The module is taught via one, two-hour lab session each week and a one hour

lecture which runs for the first term only. The lab sessions typically start with an

introduction to the work for the session with a live demo of how the code might be

created. Students are encouraged to rely on self study as much as possible. All

materials are available on-line with an extensive library of video tutorials based on

the labs and lectures. There are two assessment tools on the module, The Driving

Test and two lab based phase tests.

The Driving TestsAs the name suggests, the core concept of the Driving Tests is that the student may

fail the assessment multiple times until they finally pass. The Driving Test requires

students to complete code to solve a problem and then answer questions individually

about the nature of that code. As far as possible students are expected to self-

manage how they arrive at solutions to problems and when they take the

assessment. In order to place controls on the assessment process a protocol has

been designed. The protocol is intended to encourage the students to take the

assessment in a timely manner and limit the number of assessments in a given

week. A system of “three strikes and you are out” is applied to reduce the number of

assessment points in a week, additionally the assessment may only be taken once

during the week’s timetabled session

The assessment requires the students to work collaboratively (either formally or

informally) to solve the set problem. The intention is that the tutor stands back from

directly guiding the student. Lastly the test attempts to mirror real-world

programming problems for example students are required to write a program that

might allow for a business customers to make orders for goods.

3

The Phase TestsThe module also includes two phase tests at the end of each term. The tests are

open-book lab based assessments where students are expected to write code from

scratch. At the start of the module, students are presented with a sample phase test

question along with a possible solution. On the day of the test they are presented

with a randomly selected test of similar nature to the one from the example.

IMAT2204 Project Management and Development (Second Year)

Background and StructureProject Management and Development (PMD) is a second year undergraduate

module, teaching programming skills against the backdrop of wider systems analysis

and design. It follows directly on from VWD and is delivered to students on

Computing and students on Information Communications Technology. The module

assumes that students will currently be able to understand the terminology and

concepts of programming but will not as yet be equipped with sufficient problem

solving skills to build a system of medium complexity. The starting point for the

development is founded in the process of analysis and design. The first term of the

module involves the development of design blue-prints, allowing the developers to

understand the problem and express it as a series of diagrams, e.g. Use Cases,

Class Diagrams, Entity Relationship Diagrams and Sequence Diagrams. The

assumption is that students are more likely to understand a system expressed

diagrammatically rather than it being expressed primarily as code. The analysis and

design documentation is then used as a jumping off point for building the code for

the system in term two.

4

Programming PathwaysProgramming Pathways is a step by step approach to programming using Test

Driven Development (TDD). Using this document the students build their system

from small “bricks” of code arising from the testing. By assembling these units the

whole system is built mapping on to the original design documentation. In the same

way a house is built from architectural blue-prints the final system may be viewed

against initial designs to demonstrate it has met the final requirements. By viewing

the system as a set of design documents and resulting code, the assumption is that

the two views of the design re-enforce student understanding of both code and

documentation.

Like VWD the module is split into two terms. The first term contains the bulk of the

taught material to explain the theoretical underpinnings of the work. Labs and

lectures are closely aligned to cover the theoretical practical aspects each week.

Much of the worked examples for PMD call back to those used on VWD. The idea is

that providing a continuity of examples helps to reinforce understanding. Like VWD

there is one two-hour lab in addition to the one-hour lecture. During the second term

the emphasis shifts from delivering taught content in lectures to assessing the

student’s work in labs as they develop the software via sprints.

5

The SprintThe idea of a sprint is taken from agile software development and refers to a time

constrained period of activity undertaken by a development team in completing a

project, typically of one to four weeks (Chromatic, 2008). The students work in

teams of three to five on a shared software project. Each student is required to work

on a unique component of the whole and the assessment is calculated individually.

Every fortnight the team meets with the tutor both collectively and individually. The

individual meeting is strictly timed to five minutes and students have the opportunity

to present partially completed work and obtain both verbal feedback along with a

provisional grade for that component. It is assumed that this opportunity for dialogue

and discussion plays an important role in student learning, not only with the tutor but

with fellow team members. The assessment follows a feed-forward model in that

marks claimed at a meeting may be improved upon at a later meeting thus allowing

students to benefit from tutor feedback.

The eGridAs part of the assessment process for this module eGrids have been used to allow

students to self regulate their progress (Dean, 2012). eGrids are spreadsheets that

contain a detailed breakdown of each assessed component. Students have

individual access to their own eGrid to preserve privacy, along with read-only access

to avoid any tampering with the grade. Students may however download their eGrid

and enter grades locally in order to model their final grade. The tutor has full access

to the eGrid updating the contents of the grid at each five minute meeting. The eGrid

is the pivotal tool in generating the feed-forward assessment and feedback, in that at

each meeting the eGrid is updated with the latest marks allowing the student to

regulate their progress.

6

IMAT3903 Team Development Project (Final Year)

Background and StructureThe Team Development Project (TDP) was conceived as a more “pure”

implementation of PBL principles. In the two modules above there is, by design,

considerable supporting content and tutor support. In the design of this module the

students have access to the content from the previous two levels; however no new

content is available for this cohort. TDP requires the students to work collaboratively

on shared projects of significant complexity. Working in teams of around four they

must negotiate the design of shared components within their respective systems,

very much like a software house in industry would. This process begins with the

creation of design documentation followed by the implementation of the system

code. Students work in sprints of about four weeks meeting the tutor for feedback,

guidance and to obtain marks. At the end of the module students will have complete

a project of significant complexity. This is broken down over the year into the project

proposal, the initial design, the written report and a final presentation. Like PMD this

module makes use of eGrids coupled with sprints allowing students to monitor their

performance.

7

The Student Voice

The final area for investigation within the research is the student perception of the

above teaching and learning strategies. There have been anecdotal indications that

some of the principles of PBL do not actually map to reality. This appears to be

especially true when it comes to novice programmers specifically relating to their

views on the role of tutor as facilitator and negotiating problem complexity. The

impression is that students find programming very difficult. Results for programming

modules are typically lower than other modules for the same students. There are

currently discussions within the Faculty about the best way to address this issue.

Additional support is being proposed of a similar nature to that provided for Maths

and English. Discussions are also on-going, related to revising teaching, learning

and assessment strategies. Are lectures the best way to teach programming and

should they be abandoned completely? What are the best tools for assessment? By

listening to the student voice perhaps some insight may be obtained into these

questions.

Dissertation Structure

This dissertation begins with a literature review, discussing the existing theoretical

basis for teaching programming and PBL. How the research has been undertaken

will be discussed in chapter three. Chapter four will look at the findings and chapter

five will consider the implications of these findings in the light of appropriate theory.

The report will close with appropriate conclusions and recommendations drawn from

the findings.

8

Chapter 2: Literature Review

Introduction

A comment that often arises at exam boards when reviewing computing modules

with low pass rates is the statement “yes but it is a programming module.” Within

this statement is the suggestion that to teach programming is to naturally anticipate

low pass rates. This statement is borne out by evidence. Dehnadu and Bornat,

(2006) note that between 30% and 60% of first year intake in Computer Science fail

the first year. Nor is there an even distribution of failure rates, rather than a bell

shaped curve the trend is towards a “two humped camel” (Dehnadu and Bornat

2006). Perkins et al., (1989) identify two camps within the cohort of novice

programmers. “Stoppers” are those students who, when confronted with the task of

writing code appear to give up. “Movers” represent the camp that keep trying and

acquire the appropriate skills. This literature review covers some of the issues

related to teaching programming, especially to novices, and then goes on to explore

the theory of constructivism and its implementation as pedagogy in Problem Based

Learning.

The Programming Problem

Behind the problematic progression rates there are two issues to consider. Firstly,

learning programming is difficult for students with no background in the subject and

secondly educators in Computer Science have no formal background in education

(Muller and Kidd, 2014). Programming is a process that requires a variety of skills.

A programmer needs to problem solve, plan a solution, write the solution in the

selected language, perform testing, read existing code and have “real world”

knowledge of the application domain of the completed system (Pennington, 1987).

With technology encroaching on a wide range of professions there is also a growing

requirement to educate non computer scientists in programming. For example

geography students now have a requirement to learn programming whilst interacting

with computer technology (Muller and Kidd, 2014). Lastly, it has been suggested that

it takes about ten years for a novice programmer to become an expert (Winslow,

9

1996). In the light of the above points there needs to be effective strategies in place

for supporting both learners and tutors along with realistic expectations of what can

be achieved during an undergraduate degree program.

Different approaches have been taken to addressing the problematical nature of

teaching programming. Dehnadi, (2009) outlines some of these approaches to

teaching the subject. The programming first approach has traditionally been used on

initial programming courses. This approach focuses on the code and the individual

elements of the programming language. The programming first approach however

typically results in many concepts related to object oriented programming being

sidelined and has thus become problematic. The second approach is the concepts

first approach and relies on the use of analogy to teach programming. In this

approach teachers try to make a link between real world concepts and activities to

programming concepts. An example of concepts first might be using children’s

board games to illustrate programming concepts (Curzon, 1999). Thirdly there is the

objects first approach which explores programming from the perspective of object

oriented development often at the expense of traditional procedural concepts.

Pennington, (1987) outlines two types of knowledge required to understanding a

computer program. Text structure knowledge arises from understanding the

“alphabet” of a program, for instance “sequence” and “repetition”. Plan knowledge

takes a higher, level view. In this view a program is a set of outcomes achieved by

combinations of code units. In the former position, the code is viewed as units of

code, in the latter; a program is a desired outcome. The former approach to

teaching programming has traditionally been the approach taken by computing

curricula (Robins, Rountree, Rountree, 2003). For example such an approach might

concentrate on the building blocks of a language e.g. loops, variables etc. This

alphabet approach however may fail to cover the wider application of those building

blocks within a finished system. For example, this might create a situation where a

student understands how to code a “for-loop” but has no idea when best to

implement that control structure within a finished system (Davies, 1993).

10

Lister et al, (2004) Note that that many graduates are unable to write code upon

graduating. When asked to predict the outcome of a section of code students were

unable to do so demonstrating poor problem solving skills. Dehnadi and Bornat,

(2006) paint a bleak picture in collective efforts to teach programming. They

describe the initial enthusiasm of staff and students entering the first year that turns

into a state of demoralization for both.

“They struggle on to the end of the course, for reasons of personal pride, family

pressure or because of a lack of pastoral support and a dearth of escape routes,

disillusioned and with a growing sense of personal failure. Those who can learn, on

the other hand, are frustrated by the slow breeze of teaching, a wind tempered to the

struggling shorn lambs. Teachers become demoralised that once again they’ve failed

to crack the problem. Everybody tries their best, and almost everybody ends up

unhappy – until the next year when a new intake begins a new course with their

motivational batteries fully charged.” (Dehnadi and Bornat, 2006, p. 2)

Du Boulay, (1989) identifies five areas of potential difficulty that students must

overcome in learning programming. Firstly there is orientation where a student

understands what programs are and their potential. Secondly there is the idea of the

notional machine which is a model of the computer as it relates to the program.

Thirdly there is notation, i.e. the syntax of the language. Fourthly there is the concept

of structures, e.g. the common algorithms. Lastly there is the pragmatics, which is

the skill of constructing code along with associated activities e.g. testing and

documentation.

Du Boulay goes on to note that one of the problems of learning programming is that

the student must address these issues all at once, thus generating something of a

shock.

“None of these issues are entirely separable from the others, and much of the

‘shock’ [. . .] of the first few encounters between the learner and the system are

compounded by the student’s attempt to deal with all these different kinds of difficulty

at once.” (du Boulay, 1989, p. 284)

11

In order to comprehend each of these five domains students need to build mental

models of what is going on within the computer (Robins, Rountree, Rountree, 2003).

If these mental models are faulty then it becomes difficult to grasp and predict what a

program is doing. Ben-Ari (1998) illustrates this with an example similar to the

following:

A = 3; //A is assigned the value of 3

A = 8; //A is assigned the value of 8

B = 5; //B is assigned the value of 5

Display(A,A,B); //Display the data held in A, and B

One common misconception among novice programmers is the idea that a variable

may store two values simultaneously. If a novice programmer believes this, then they

might erroneously conclude that the output from “Display” would be 3, 8 and 5 rather

than the correct answer 8, 8 and 5. (The initial value of 3 assigned to A has been

over-written by the new value of 8.)

There may also be a tendency to anthropomorphise the computer system. A novice

programmer may write a section of code, expecting the computer to understand their

intent. They wrongly imagine that the computer will “get” what they as a programmer

mean by their efforts at code, but fail to appreciate that the computer is simply a

machine (Spohrer and Soloway 1989). There are also issues related to the

confidence of individual students. Even if the student “knows” the answer to a

programming problem this knowledge may be described as “fragile knowledge”

(Perkins and Martin, 1986). In the case of fragile knowledge they may know the

answer but fail to act on that knowledge.

There is evidence suggesting that the formulation of these mental models has an

impact on how successful in assessment a novice programmer will be. Dehnadi,

(2009) studied students to understand their mental models of programming. Who

held what mental model, was correlated with test results from experimental data

along with final test results. Within his sample he identified two groups of students

who perform significantly differently. More than half of novices in the sample appear

to build and consistently apply mental models of programming whilst the rest do not.

12

In his findings the first group performed much better in their final examination with an

85% success rate whilst the second group achieved 48%.

The questions therefore arise how are these mental models refined and developed

and what can be done by educators to facilitate this process? There are many

different theories of learning, however, there is insufficient scope to look at each of

these in detail but we shall provide a very brief introduction and concentrate on

theories related to constructivism and ultimately Problem Based Learning.

Building Mental Models

In classical thought Plato suggested that knowledge is “a recollection of something

the soul has already learned previously” (Silverman, 2014). By this logic if you did

not know something “previously” then there is no point attempting to learn it. This

view however still leaves the basic question unanswered, that is how did the

knowledge get there in the first place? John Locke in the 17 th century however

suggested an alternative view. Rather than knowledge being already present in the

subject, humans start as a “blank slate”. Rather than the knowledge being built in, it

is the ability to acquire knowledge that is inherent (Phillips, 2015).

In order to understand the world around us it is necessary to make assumptions

relating to its nature, i.e. an ontology or world view. For this discussion we shall

consider two such views, they are logical positivism and constructivism. As a world

view, logical positivism takes the view that knowledge exists, “out there as objective

truth” (Boud and Feletti, 1998, p127). In a logical positivist position, there is an

objective reality to be understood and most significantly, personal subjective

experience plays no part in understanding this reality. Data from this objective reality

is:

“passed through mental structures by cognitive mechanism in an analogous way that

a message flows through a telephone wire. The process of learning becomes one of

mapping this external knowledge into some form of internal representation that more

or less approximates the external ‘objects’” (Knuth and Cunningham, 1993, p 164).

13

Constructivism takes a different view of reality, knowledge and ultimately how we

learn. In this view, knowledge is “constructed” by the learner via their direct personal

experience of reality, rather than “poured into their heads” via text books and

lectures (Ben-Ari, 1998). Rather than starting with a “blank-slate” our view of external

reality is filtered and coloured by a number of factors including our prior knowledge,

expectations, emotions and prejudices. The internal beliefs and perceptions of the

individual will inevitably have a direct impact on their behaviour. For example

Morrison (1998) notes, “if I believe there is a mouse under the table, I will act as

though there is a mouse under the table, whether there is or not.” Constructivism

takes the view that there is a knowable universe “out there” however our internal

constructed version of it will inevitably fail to map (Boud and Feletti, 1998).

Critics of constructivism voice the concern that it leads to solipsism, which is the

belief that there is no external reality only ones own thoughts. Matthews (1994, p.

151) comments with concern that, “‘we cannot know reality,’ is endemic in

constructivist writing”. It is not within the scope of this discussion to resolve these

questions, however the more extreme versions of constructivism may possibly be

tempered by the acknowledgement of a knowable universe “out there”, whilst

recognising that our personal constructed version of it is imperfect (Boud and Feletti.

2003). As a counterpoint to such criticisms of constructivism it is also worth noting

that taking a solely logical positivist view of reality should also come with some

humility. Kuhn, (1970) notes that science has a history of revolutionary change and

long held “firm” scientific ideas may over-time be overturned by on-going scientific

discovery.

14

The development of mental models to correctly understand the outside world is not

unique to those learning programming. Piaget (Piaget et al., 1956) devised the three

mountain task to examine learning in children. Children are shown a scale model of

three mountain peaks and asked to describe the mountains from a different

perspective. Piaget found that children between four and six could understand the

question but were unable to do what was asked of them (Jarrett and Ginsburg,

2008). Another concept driving learning in Piaget’s view is that of accommodation

and adaptation (Glaserfield, 1989). In this case learning and change come about

when a person encounters an unexpected result. It is in fact the problems we

encounter that provoke learning. Upon encountering a problem we accommodate the

unexpected outcome with new information (accommodation) and ultimately adjust by

modifying our internal models (adaptation).

A contemporary of Piaget, Vygotsky did much to advance theories related to learning

in humans. As mentioned above, some considered intellectual capacity to be innate

in children. Such capacities were therefore waiting for an opportunity for them to

manifest, rather like a plant growing to its mature state. Vygotsky argued against

existing views on intellectual development. He argued that there were differences

between the way that plants might mature to their potential, or that an animal might

respond to its environment. Humans have an additional dimension to learning,

which is language, giving rise to social interaction (Vygotsky, 1978). In this work he

uses the following example, a baby grabs for something that they want and the

parent provides this to them. At some point the link is made between grasping for an

object and communicating with a person. The act of grasping turns into the act of

pointing which becomes part of the repertoire of human communication.

15

By interacting with the external environment sophisticated mental processes are

developed. In Vygotsky’s view these higher functions are determined by the culture

an individual develops in. For example, in a literate culture we may use note taking

to aid memory; however in pre-literate societies such devices as tying knots, carrying

pebbles or verbal repetition may be used (McLeod, 2007). Vygotsky continues with

the idea that learning occurs also via interaction with a skilled tutor or More

Knowledgeable Other. Shaffer (1996) uses the example of a child completing a

jigsaw. The parent initially provides some basic strategies for attempting the puzzle,

yet over time allows the child greater autonomy as they become more competent at

the problem. In this example we see the parent taking the role of Vygotsky’s More

Knowledgeable Other illustrating the kind of social interaction using cooperative and

collaborative dialogue promoting cognitive development.

Another concept put forward by Vygotsky is the Zone of Proximal Development.

Related to the concept of More Knowledgeable Other it relates to those tasks a child

can complete without assistance compared to those tasks a child needs guidance

and support (McLeod, 2007).

“The zone of proximal development defines functions that have not matured yet, but

are in a process of maturing, that will mature tomorrow, that are currently in an

embryonic state; these functions could be called the buds of development, the

flowers of development, rather than the fruits of development, that is, what is only

just maturing” (Vygotsky, 1978, p. 86).

Vygotsky sees the zone of proximal development as the area where guidance and

instruction should be given, facilitating the development of further higher mental

functions.

16

Constructivism as Pedagogy

Constructivism being a theory of learning does not favour or reject its implementation

by any specific means of delivery. Mason, (1994, p 197) remarks dryly “Many

educators espousing constructivism have been known to attend lectures on

constructivism, and even to have enjoyed them!’’ Based on constructivist principles

passive learning however is likely to be unsuccessful as learners are bringing with

them existing knowledge, experience and different strategies for constructing

knowledge. Learning in this view must be active with the participation of peers and

the tutor in order to develop mental models (Ben-Ari, 1998).

Whereas constructivism describes the educational theory, Problem Based Learning

represents a pedagogy arising out of these theories. The origins of PBL date back to

the 1960s at McMaster University Medical School (Schugurensky, 1996). Being a

new hospital and medical school the opportunity was taken for devising new

approaches to medical education (Baden, Major, and Savin-Baden, 2004). PBL was

hoping to address findings that “medical students and residents for the most part did

not seem to think at all. Some gathered data ritualistically and then tried to add it up

afterwards, while others came up with a diagnosis based on some symptom or sign,

never considering possible alternatives.” (Barrows and Tamblyn, 1980). Prior to the

introduction of PBL, medical schools typically imparted knowledge via a lecture

based curriculum. The lecture based approach was structured as an “organ based”

curriculum focussing on essential areas of medicine e.g. anatomy, physiology,

pathology and treatment. McMaster University continued to cover these topics but

within the context of case based problems to facilitate student learning

(Schugurensky, 1996).

The PBL approach to teaching was considered to revitalise the teaching and learning

process for students. Student engagement appeared to improve, and the focus on

problem-solving skills appeared more appropriate to a rapidly advancing field of

knowledge rather than expecting students to rote learn knowledge that may be out of

date by the time they graduate (Boud and Feletti, 2003).

17

In its initial inception PBL consisted of the following core concepts: Students are

presented with complex real world problems with no single “correct” answer. The

students work in teams to devise solutions to the problem at hand gaining knowledge

through self-directed learning. Members of staff take on the role of facilitators of

learning rather than taking on the role of expert (Baden, Major, and Savin-Baden,

2004). The problems presented to the students should be drawn from real world

practice; if a student is learning to be a Doctor then they should encounter and solve

real world medical problems (Baden, Major, and Savin-Baden, 2004). Students are

given opportunity to interact with each other in teams or groups. As a group of

students learn, each student in that group though working on the same task will have

different perspectives and insights in understanding what is required. The

assumption is that social interaction is an important factor in learning (Boud and

Feletti, 2003). In framing the role of the tutor as facilitator the assumption is that the

tutor is not the best qualified to teach the students as they are too far ahead in their

knowledge. The best people to teach the students at any point in time are the

students themselves. The role of the Tutor therefore is to create opportunities for

students to construct their own knowledge whilst maintaining the role of facilitator

(Hoover, 2007).

Another perceived benefit of PBL is that it creates a holistic approach to knowledge.

For example, if students were presented with the topic of global warming as a

problem to solve it would also bring into the discussion issues such as politics,

culture and meteorology. Rather than compartmentalising knowledge in individual

lecture sessions the issues are viewed as interconnected ideas (Boud and Feletti,

2003).

PBL is however not a simple recipe for pedagogic design. As PBL has spread to

disciplines beyond medicine it has been adapted to address discipline specific

issues. In technical subjects, such as engineering there may be a prerequisite body

of knowledge such as mathematical skills that need to be acquired before the

student may appropriately engage with suitable problems and in some

undergraduate programmes PBL may take place in the latter stages of the degree

(Baden-Savin and Savin-Baden, 2000).

18

A PBL curriculum may be delivered in a wide variety of ways. For example, lectures

may be used to deliver core concepts whilst problems may be explored in small

group sessions, case studies may be used to discuss issues or “closed loop” PBL

where the emphasis is on the mechanics of the problem-solving process and how it

might be developed. A PBL curriculum may include cross disciplinary knowledge,

emphasis on knowledge acquisition skills rather than (but not to the exclusion of) the

knowledge itself, a focus on student led (peer) assessment and an emphasis on

interpersonal and communication skills (Baden-Savin and Savin-Baden, 2000).

Criticism of Problem Based Learning

PBL is not without its critics and there are perceived issues. PBL being an unfamiliar

approach to pedagogy may present issues for both staff and students. Staff and

students may require additional training/guidance to become familiar with the

approach and students may not use the time for independent study wisely (Overton,

2010). There may also be issues related to cognitive load. Should the student be

presented with too many elements to process simultaneously it may place too great

a demand on their ability to process data (Gerjets, Scheiter and Catrambone 2004).

Sweller, (2006) considered the learning process from two perspectives. Firstly there

is that knowledge already acquired in long term memory that has been “borrowed”

from the knowledge of others. Secondly there is that learning that is acquired via

short term memory. The issue presented here is that the capacity of the brain to

process new information in short term memory is relatively small, whilst the brain can

process a much larger quantity of data already stored in long term memory.

Knowledge in long term memory may be “borrowed” from others via worked

examples, whereas processing problems in short term memory requires an element

of trial and error which is both time consuming and demanding for the student. In this

view it is by combining problems with worked examples that provokes learning.

19

Lastly there is the issue of unfettered discovery (Ben-Ari, 1998). In its extreme form

the students are presented with a problem and left to find a solution with minimal

input from the tutor. This can be very time consuming, and have a detrimental

impact on student engagement (Provan, 2011). This is also supported by Bruner

(Bruner, 1962) who differentiates between episodic constructivism where the student

acquires unconnected facts contrasted with cumulative constructivism where the

process of discovery is organised.

Conclusion

In teaching programming there are clear challenges in engaging students, especially

for novice programmers. This difficulty is compounded by a lack of educational

background amongst computing staff. Constructivism coupled with Problem Based

Learning, perhaps offers a way to approach some of these issues by facilitating the

development of appropriate mental models. PBL is not without criticism though, both

philosophically and practically.

20

Chapter Three: Methodology and Research Methods

Introduction

The primary focus in this research has been grounded theory combined with mixed

methods. This project represents an opportunity to listen to the student voice and

gauge the merits of the implemented pedagogy. The research included the use of

questionnaires and focus groups triangulated with a study of the existing body of

research on the subject. At all points, appropriate ethical and security considerations

have been in place.

Research Methodology

To introduce the selection of overall research methodology and research paradigm

we shall contrast two events; firstly the physics of a cannon ball and secondly the

dynamics of a classroom.

When a cannon ball is fired, there are several aspects to this event. Firstly, the

event is measurable. We may record the distance traversed by the ball and we may

record the arc of the trajectory. Secondly the event is repeatable and controllable. It

is possible to repeat the event having control over the variables, such as the size of

the ball, the quantity of explosive and the angle of the cannon. Having the possibility

of adjusting various aspects of the event allows for the creation of controls. We are

now well placed to discover correlations between the different variables by creating

controlled experiments.

21

Would it not be useful to have the same insight and control into the inner workings of

a taught session? There are certainly variables that may be measured during and

after the class, perhaps we only need to measure progression rates and attendance

data? There are however several other issues to consider too. The class “event”

unlike the firing of the cannon is not repeatable and controllable. How many times

have we taught the same lesson in a week knowing we have a “good” group and a

“bad” group? Asking a single group of students to sit through the same class

repeatedly while we adjust individual aspects of the activity would produce several

variables out of our control, not least the fatigue and interest levels of the students

and the tutor. Perhaps we might atomise the experience via a controlled

experiment? But would this not in of itself create a new set of issues, for example are

the results of this experiment still applicable to the real-world activity? Unlike a

cannon ball the class is a personal, potentially private, subjective experience for tutor

and students alike.

There are a multitude of variables out of the tutor’s control and inaccessible to direct

measurement. These make measuring the following aspects difficult. The first

aspect relates to causality (Halai, 2004). How can we know what specific feature of

teaching practice had benefit to which area of student learning? The second aspect

relates to trying to measure the presence of learning. How do we know that learning

has taken place in a class? Is it conceivable that even though a student has failed

the assessment they themselves have learnt a great deal from the class?

22

Broadly speaking educational research may be divided into two paradigms, positivist,

and interpretivist (Cohen, Manion and Morrison, 2007). The positivist paradigm

draws from the natural sciences making certain assumptions about the nature of

knowledge and reality. In the positivist paradigm knowledge and reality exist

independently from the observer. Positivism seeks generalised laws and principles

that explain why things behave the way they do (Denscombe, 2003). Applying

positivism to an educational context, we would need to find repeatable patterns

perhaps by means of experimentation and large data sets. Like the cannon ball we

want to learn the “physics” of the situation. This may be useful in ascertaining a big

picture view of learning and teaching however it struggles to provide insight in to the

simple question of “how did a class go and why?”

Alternatively the interpretivist paradigm sees knowledge and reality as subjective,

personal, and unique. Rather than looking for general objective laws that govern a

situation we seek individual perspectives on knowledge and reality (Cohen, Manion

and Morrison, 2007). In the context of understanding “how did a class go?” we could

simply ask the participants of the class. However, in this paradigm there is

potentially an issue of rigour. How can we be sure that the subjective views of the

participants are in fact representative of what happened (Angen, 2000)?

Rather than create a false dichotomy by selecting between the above paradigms,

there is a third option that is the mixed paradigm (Cohen, Manion and Morrison,

2007). The mixed paradigm draws from both the above paradigms taking a position

that both views have merit. One of the benefits of the mixed paradigm is that it

encourages triangulation (Denscombe, 2003). Triangulation draws its analogy from

sailing whereby knowing three points, a vessel can calculate its location. By using

findings for example from a quantitative (positivistic) study, comparing the results

with the qualitative (interpretivist) findings in conjunction with existing research we

may use the points of correlation in the three points to get a better picture of where

the truth may be.

23

Data collection methods

Grounded theory first originated with Glaser and Strauss (1999). In the case of

grounded theory, theories arise out of empirical research rather than starting with a

theory and then looking for the supporting evidence (Denscombe, 2003). The

starting point of the research is data gathered in the field of research. From this data,

patterns form and the theories arise. There are certain common elements to the

various forms of grounded theory namely theoretical sampling, coding, constant

comparison, the core variable, and saturation (Cohen, Manion and Morrison, 2007).

Theoretical sampling involves collecting data, reviewing the data, and then deciding

where to go next. In this process the emergent theory should become visible. This

process continues until saturation, at which point no new insights into the theory are

produced. Coding is the process of taking the data e.g. interview transcripts and

breaking it down into smaller sections until themes start to emerge. From the coding

process, categories start to emerge and by the process of constant comparison new

data is compared with the existing data to form a harmony of codes and categories.

If there is a poor fit between data, categories, or theory then the theory must be

reviewed (Cohen, Manion and Morrison, 2007). As this process continues the core

variable is identified. “The core variable is that variable that integrates the greatest

number of codes, categories and concepts, and to which most of them are related

and with which they are connected.” (Cohen, Manion and Morrison, 2007, p 601).

There is a note of caution on the matter of reaching saturation point that is the

understanding of the research topic comes in layers. It is possible to think you have

arrived at the top of the peak only to see a new one arise over the horizon. There are

times when the constraints of grounded research are not based on having found “the

answer” but on running out of time (Bell, and Waters, 2014). This leads to some of

the drawbacks of grounded theory. Since the exact point of saturation is not known

at the start, grounded theory does not lend itself to precise planning. It is a voyage

of discovery where the end is not in sight when we start. Some also contend that in

the worst case grounded theory might be a euphemism for bad research. To counter

this there needs to be a clear idea of the research topic along with a clear basis for

selecting the sample, avoiding a “try it and see approach” (Denscombe, 2003).

24

Grounded theory does not specify any data-collection tool. It does tend to favour

unstructured approaches such as open ended questionnaires, field notes from

observation and unstructured interviews (Denscombe, 2003). In this research

project, a literature review, questionnaires, interviews, and focus groups have been

the main tools.

Ethics

Unlike the physics of a projectile the mechanisms at work in a learning environment

are personal, subjective, and potentially private. At all points during the research,

ethical considerations are paramount. There are three areas that need to be

considered in undertaking ethical research. Firstly, the participants’ interests should

be protected; secondly the researcher should not deceive or misrepresent the

research and thirdly participants should provide informed consent (Denscombe,

2003). On the first point the researcher needs to consider the physical safety of

participants. Even though physical harm may be unlikely a sensible thoughtful

approach is required. Additionally, the researcher needs to consider the

psychological safety of the participants. For example, as a potentially perceived

authority figure the researcher needs to be aware how this might impact on the

sense of wellbeing of the participants. Lastly there is an issue of reputational harm,

for example, if information disclosed falls into the public domain that may be private

and potentially damaging to the participant. As discussed below there are issues to

consider related to security of the data gathered along with a sensitive unbiased

approach to the focus groups. As far as possible there has been transparency

related to how the data is captured what the data is for and the availability of

findings. By implementing appropriate levels of transparency deception should not

be an issue, however, it is important to ensure that the participants are not

misrepresented in any way so impartiality is maintained as far as possible. Lastly at

all points participants have been provided opportunity to give informed consent. This

is especially important when gathering data in class. There may be potential issues

with students feeling pressured to conform to providing data in a class of mostly

cooperative students complying with the tutor, an authority figure. It is important that

not only informed consent is provided but also mechanisms to opt out do not

adversely draw attention to those individuals.

25

The Role of the Literature Review

Within grounded theory there is debate on the question of does an initial literature

review potentially pollute the process (Cohen, Manion and Morrison, 2007)? Purists

might argue that grounded theory should start with no preconceptions on the subject

therefore a literature review would be out of the question. However, this extreme

position is not usually taken; existing theories are viewed as open to question in the

light of the empirical findings (Denscombe, 2003). The researcher should start the

line of research with an open mind. An open mind does not however equate with a

blank mind. The researcher needs to be aware of existing theories related to the

subject at hand (Denscombe, 2003). Part of the research question under

consideration is the match of empirical data to constructivist theories. To produce

this critical analysis of PBL it is necessary to understand the theoretical basis. That

being the case a literature review has been undertaken into constructivism and PBL.

Questionnaires

Questionnaires are often used as a mechanism for capturing data from the “horse’s

mouth” (Denscombe, 2003). In this research project the sample is something of a

captive audience in that it is based on the taught cohort of students. It was made

clear that students have the option to decline participation should they wish.

Denscombe (2003) raises several points about questionnaires that certainly relate to

this research project. Questionnaires are typically used to capture a large quantity of

data which helps to average out the margin of error (Bell and Waters, 2014).

Questionnaires allow for standardisation of the data by applying the same questions

to the sample cohort. Questionnaires require a great deal of skill to construct. There

are potential issues related to a mismatch between what the author means in their

questions and how the respondents interpret the questions. In writing a

questionnaire it is important to select words carefully, with clear unambiguous

questions that do not lead the respondents in a certain direction. It is also important

that the language is pitched at a level suitable for the sample cohort (Bell and

Waters, 2014).

26

Programming Preferences Survey

The first questionnaire involved the use of the Programming Preferences Survey.

The survey was intended to gauge student preferences specifically in learning

programming. Considering the substantial administration and data processing

requirements it was decided to deliver the questionnaire electronically. A long-

standing web site frequently used by the students was used to deliver the

questionnaire (Dean, 2017).

Security

Security can be a mixed blessing when gathering data. On one hand, it is a

requirement for any web system to avoid unauthorised access; on the other hand it

can easily become a barrier if the security requirements are too stringent. Due to the

bespoke nature of the questionnaire it was decided not to deliver it through

proprietary systems such as Blackboard, so some sort of custom security would

need to be implemented. One important aspect of a secure system is secure

communication. The intention here is to stop access to any data transmitted

between source and destination. All access to the questionnaire is via a secure

connection using appropriate data encryption. Having previously created systems

granting students access to open ended questions with full anonymity it had been

noted that this was on occasions open to abuse. This abuse took the form of those

outside of the study accessing the system and some students acting in an

inappropriate manner. This being the case there needed to be a mechanism for at

least in part controlling access yet also maintaining anonymity.

27

Students have already been registered as part of an existing system on the server.

This meant that any valid users of the system could be identified by looking up their

student numbers in this existing system. This would not however allow for

anonymous access since as soon as their student number was known then it would

be possible to look up individual students. To resolve this, the system generates a

hash key based on the student’s number. A hash key is a long alphanumeric string

that is commonly used to secure passwords on servers. From a string of characters

e.g. “mypassword” the hash key is generated for example

“8CEE563B099CCF58C98D0BD55F712F1400281809”. Unlike encryption where it is

possible to calculate the original text from the encoded text, the reverse processing

is much harder with a hash. To add to the complexity of the hash, participants were

invited to input their date of birth. By combining the date of birth with their student

number a much more secure hash is generated.

Informed Consent

As part of the ethical considerations of the research it is important that all

participants have informed consent. This allows them to understand such things as

the background to the study, implications of involvement and the right to withdraw

(See Appendix).

28

The first page participants see when accessing the questionnaire is the participant

information sheet. Students were invited to complete the questionnaire at the end of

their taught laboratory session. This provided opportunity to explain the nature of

informed consent and allow them to fill in the consent form. The voluntary nature of

their engagement was made clear and no attention was drawn to those who elected

not to complete the questionnaire. All consent forms were gathered in at the end of

the taught session completed or otherwise.

Having read the participant information sheet the respondents need to complete the

information for generating the hash key.

Students also identify their year of study, this is not incorporated into the hash and is

used to group responses within specific years. Should a respondent be happy to

break anonymity they may indicate this by ticking the box provided. If a student

elects to do this their student number is recorded along with their submission. Once

the hash key has been generated they may proceed to the questionnaire itself.

The Questionnaire Design

The questionnaire was intended to be qualitative in nature. Since grounded theory is

in large part a listening exercise the design of the questions are such that they offer

students several open-ended questions. To provide some focus to the questionnaire

the following structure was applied.

The questionnaire consists of two main sections.

29

“Your Learning" – to enquire in general terms what has helped or hindered their learning of programming.

The Teaching – eliciting responses to specific questions related to problem based learning.

Your Learning

In this section students were invited to make general comments of what has helped

or hindered their learning of programming so far in their studies. These sections

were deliberately left open ended to allow free expression of response.

So far in the course what has hindered your learning of programming? So far in the course what things helped you to learn programming?

The Teaching

The focus of the research is to try and evaluate the benefit or otherwise of strategies

associated with PBL. It was decided that to do this, students should be given some

sort of indication as to the basic premises associated with PBL.

30

Students were provided with the following list of pointers:

Problem Based Learning is an approach to learning based on the following

assumptions.

You learn best by solving suitably complex programming problemso Students are presented with problems to solve, which are

complex enough to be challenging to them without any clear single solution.

You learn best by working together in teams to solve these programming problems

o By working in teams, different views on the problem are generated which aids the learning of all team members.

You learn best when the tutor does not directly help you to solve these programming problems

o The tutor is not always the best qualified to direct student learning as they know too much to be of use to the students. The tutor should therefore help students to learn without always relying on their tutor.

You learn best working on real world programming problems using real world tools and strategies

o This means that your learning activities at University should be the same as employment activities in industry.

Students were then given opportunity to respond to these points like so:

Explain why you might agree or disagree with the following statements: o "You learn best working on real world programming problems

using real world tools and strategies." o "You learn best by working together in teams to solve

programming problems." o "You learn best by solving suitably complex programming

problems." o "You learn best when the tutor does not directly help you to

solve problems."

Data was gathered from all years of the course in both term one and term two. The

sample contained some students directly taught by the researcher; however,

including students outside of direct contact reduces the risk of students saying what

they think the researcher want to hear and associating final marks with their

response. Coupled with options for anonymity and the large sample size would also

help to improve the quality of the data.

31

Learning Preferences Survey

The second questionnaire was the Learning Preferences Survey. The Learning

Preferences Survey was developed as part of previous Teacher Fellow research.

The survey allows students to express the value students place on different

approaches to teaching in two ways. Firstly, they may specify the value they place

on different teaching strategies in a range of 0 – 5. Secondly there is a Boolean

option identifying if a teaching strategy is essential or not.

Once the data has been gathered from the sample population it allows for ranking of

each element based on the value the students place against them. This allows the

researcher to obtain a sense of the relative value of different teaching and learning

strategies within the cohort. The output from the survey is ranked in tables of

learning strategies.

32

For example, in this case based on their value to student learning we might assume

that lab sessions have greater value than reading class notes.

In this case based on the essentiality of each activity we might assume that both labs

and lectures have a high essentiality rating when compared to other aspects of the

teaching.

The use of the survey came quite late in the research with a limited sample size of 8

final year students. Fortunately the survey has been used year on year and many of

the results are comparable with those from previous years. The survey was modified

in this project to try and gauge the value students place on Programming Pathways.

33

Focus Groups

The results obtained from questionnaires must by their nature be taken at face value.

In the case of interviews and focus groups a much more nuanced understanding of

the data may be obtained. Tone of voice, hesitation and facial expression are all

available clues for the researcher. These are not obtained via a questionnaire (Bell

and Waters, 2014). A focus group is typically between six and nine people brought

together to discuss thoughts feelings and ideas related to the research topic

(Denscombe, 2003). They are often relatively informal and may help to encourage

people to contribute who might by nature be inclined to hold back in other situations.

One of the problems with the above approaches to research is the interviewer effect.

The quality of data obtained from the interviewees may be undermined by the

interviewer’s manner, their perceived level of authority and position relative to the

people interviewed. For instance, if the interviewer appears too harsh or indifferent

these behaviours may influence the respondent’s behaviour. Also, the perception of

the interviewer re age, gender or race may also influence the respondents such that

they may not fully disclose or even say what they think the interviewer wants to hear

(Denscombe, 2003). One issue to consider with focus groups are the mechanisms

for transcribing the data. Typically, this is handled by hand written notes and some

sort of audio / video recording device. Recording of the sessions is helpful but the

presence of recording equipment may inhibit the responses and not everybody may

be happy to be recorded (Bell and Waters, 2014, p. 178). Other issues to consider

are what happens to the recording, how long the recording is to be kept and the

security on devices storing the recording. Also, when the recording is completed

typically a transcription is produced which adds an extra time overhead.

Two focus groups were included in the research. The first focus group was of a more

generalised nature exploring themes of student learning preference and aspects of

problem based learning. The second focus group focussed on initial findings arising

from the Programming Preferences Survey.

34

The first focus group was run mid-way through the first term. This focus group

allowed for detailed discussion about the student’s preferences for learning

programming. All students were invited to participate from all years of the course.

On the day there were two representatives from the second year and two from the

final year. The intention was to keep the discussion as open ended as possible. The

participants were re-assured of confidentiality and that there were no right answers,

only honest one. Students completed the consent form having explained the

contents of the participant information sheet, along with time to read and consider it.

Each student had a printed booklet with the questions for the session upon which

they could make hand written notes (See appendix). As with the questionnaires if

they wished they could elect to break anonymity by adding their student number to

the front of the booklet.

As a general outline to the focus group the following areas were used as a catalyst

for discussion.

Discussion related to learning programming on specific modules Experience of specific teaching and assessment strategies Exploration of Metaphors related to programming

The following metaphor was used as a starting point for the discussion.

“What I did not enjoy about the module was that the first time we were told to go off

and create our own project. It felt as if we were thrown into the pool. As in, we have

had a couple swimming lesson so we knew the basics, but then the instructor takes

you to the English Channel and tells you to meet him at the end.”

The students were invited to comment on the metaphor, decide if they agreed and

come up with their own metaphor. This metaphor had been presented to the

students in advance of the session so that they had time to reflect on their thoughts.

The Second Focus Group

This second focus group ran after the Programming Preferences Survey had been

completed in the second term, the idea being to feed initial findings from the

35

Programming Preferences Survey back into the group to gain further insight into

what they might mean. This time only three of the students could attend, two from

the final year and one from the second year. The questions discussed at this

meeting related to the nature of lectures and live-code along with the student views

on Programming Pathways.

Data Analysis ProceduresTheoretical Sampling

The theoretical sampling was carried out via the above questionnaires and focus

groups. A range of tools have been used in the analysis of the data. Data has been

captured in the most part electronically and processed in Access, Excel and NVivo.

The data being stored in a database allowed for the questions to be grouped, filtered

and exported between applications.

Coding and Constant Comparison

The first tool applied to the data was NVivo’s facility to perform word frequency

analysis to get a sense of the most commonly used words from the Programming

Preferences Survey. This allowed for a large quantity of data to be processed

quickly in order to have some initial findings in a timely manner for the second focus

group. In addition to the rapid turn-around it would also help to reduce confirmation

bias by automatic processing of results. In considering what helps students learn

programming, all responses across all years were turned into word clouds. As well

as being able to analyse the data for each year of the course, data was also

available for specific terms within a year. This allowed for a fine level of data

resolution as module delivery shifts from term one to term two on each year.

36

For example the following word cloud illustrates the question of what helps students

learn how to program based on data for the whole cohort across two terms.

This word cloud is based on the following top 10 auto-generated codes.

37

This second example looks at the same question but this time for year three students

responding in term one only.

In all there were six questions with responses from all years and all terms.

38

The next phase of examining the data involved exporting the results into NVivo and

manually coding the data. This process came in several passes. The first pass

involved grouping the data around key words e.g. “lectures”, followed by multiple

additional passes where these smaller unites were grouped into larger data sets.

From the 1000 individual responses a structure became apparent and this structure

was used to generate the main codes and themes for the research. There are certain

assumptions being made. Firstly when a student states a preference for a certain

aspect of their learning this is viewed as a two sided coin. For example when a

student states that they benefit from working through notes, from the point of view of

the tutor we need to think in terms of the content generated. From the point of view

of the student we need to think about what aspect is this catering for in them. In the

case of notes then this has been grouped as “individual learning”. In the case of a

lecture this has been grouped as collective learning.

The Focus Groups

In the case of the second focus group initial findings from NVivo’s frequency analysis

tool were used as the basis for the discussion. As noted previously, this tool has

allowed for rapid processing of a large amount of data. In the case of both focus

groups a recording was made, manual notes by the researcher and hand written

notes by the participants. The recording was transcribed and all data input into

NVivo for further manual coding.

Conclusion

By approaching the question of how best to teach programming grounded theory has

been used in order to ascertain the student voice. Clearly there are concerns with

the potential value of this data as considered in isolation it is potentially highly

subjective. To counter this, a mixed methods approach has been taken. By

triangulating the qualitative data, quantitative data and existing research it is hoped

that a picture has emerged giving clues to the effectiveness of differing approaches

to teaching.

39

Chapter Four: Findings

Introduction

By constant comparison an iterative approach has been used to process the data to

try and understand the student voice. That being the case the findings are

presented in a manner reflecting the journey through which they were discovered.

By the end of this section the reader will have a picture of what themes are present

and also the way that they came to light.

Overview of Codes

Before looking at the codes within the context of the findings, we will begin by

introducing them.

Pace and level The speed at which a learner is expected to process content coupled with their skill, knowledge and confidence level.

Individual learning Learning the student engages in as an individual, e.g. working through examples, reading notes or books.

Social learning Learning the student engages in with others where there is some measure of interaction. Social Learning includes both when students seek clarification from the tutor or discussing issues with their group / team members.

Collective learning Activities where the student takes a largely passive role alongside other students, for example attending a lecture.

Learning Modes The modes of learning (selected from above) by which a student elects to learn for a constrained period of time

Happy learning The preferred emotional tone of learning in any and all contexts

The Student The person engaged in the learning process.

The Tutor The perceived leader of learning that is not a student’s peer. Vygotsky’s “More Knowledgeable Other”.

The Team Fellow students with whom a student interacts as part of social learning activities.

Lone Rangers Students who do not work well in teams and teams do not work

40

well with them

Content The persistent content of the module

Programming Pathways

A second year guide to programming based around Test Driven Development

Driving Tests and Sprints

A first year VIVA based assessment mechanism contrasted with the sprint model of later years

Structured Time An aspect of assessment where there is a regular schedule of contact between the tutor, individual students and student peers

Sample Code Examples of written code either as full complete programs or smaller samples of code structures

Live code This is the act of demonstrating how code is written during a taught session

41

Notes on Presentation of Findings

In this document are various numbers that are used to express the strength of voice

from the qualitative data. The following sample from later in the document will

illustrate this.

Between “teaching” and “learning”, the strongest voice related to

benefiting individual student learning is the teaching (72) followed by

their learning (38).

The number in brackets e.g. “teaching (72)” is used to indicate the number of

references in the coding. This should not be used as a guide as to how many

students this relates to, it indicates the strength of the student voice coming from

specific data. When the data is expressed as a percentage in a chart like that

above, it is being used to give visual indication of the proportion of different codes to

each other. The scope of these numbers is also limited. In looking at teaching

versus learning above the 72 / 38 values only have meaning within the context of the

question under consideration.

42

The First Focus Group

The first focus group ran in the first term and was an opportunity for a generalised

discussion about the teaching of programming on the course. Four students were

involved in the focus group, two from the second year and two from the final year.

There were two areas of discussion relating to what comments did they have to

make on the teaching, learning and assessment of specific elements of the course

and what metaphors best might describe the experience of learning programming.

The following codes came out of the focus group with their number of references in

brackets.

Pace and level (19) Driving tests (10) Social learning (10) Structured time (8) Sample code (6) The tutor (4)

A strong theme coming through from the focus group is the idea of “pace and level”.

Pace and level relates to the pace of the course between year, and the pace of

teaching within specific modules. The feeling was very much that the second year

was more challenging than the first year and that the second year was much more

representative of the final year. Faster pace on the second year was however seen

as a good thing provoking deeper learning of programming.

43

The primary issue driving “pace and level” between years was the contrast between

the assessment mechanisms in the second and third year, i.e. sprint based

assessment, compared with the assessment on the first year, i.e. the Driving Test.

The sprint’s regular meeting schedule or “structured time” between the student team

and the tutor was seen as beneficial to the learning of all and also helped to

facilitating extracurricular collaboration. This “social learning” was seen as beneficial

by the students in the sample. This is contrasted with the more open-ended

structure of the Driving Test assessment where the contact takes place at the

student’s discretion. Students appear prone to procrastinate when it comes to the

Driving Test due to, feeling intimidated by the tutor, poor time management and

relying too heavily on other assessments.

As well as teaching between year, pace and level relates to the teaching within a

module. Much of the criticism of the first year module related to the slow speed of

travel. There was however recognition that some students on the first year struggle

with even the most basic of concepts. Getting the pace too fast was not necessarily

a bad thing as modules on the second year where there was, rapid progress, high

complexity and little support were considered to facilitate better learning.

There was discussion related to the role of “sample code”. At all levels the sample

code was seen to be of value both of small code sections and completed systems.

The next theme to arise from the first focus group related to the role of “the tutor”.

The tutor was seen both as signpost and as a person to fall back on when facing

problems.

44

When asked to devise metaphors that describe programming the following

suggestions were put forward.

Programming is just like learning Chinese, then being made to speak it in front of a board with Chinese writing on and figure out what it says.

Learning programming feels like having to cross a big road with too many cars driving very fast.

Programming is like learning to drive the more you practice it the better you get.

Programming something from scratch, is like building something from Lego blocks except you never run out of bricks and if there is no brick of a certain shape I can make that brick when I need it.

The Second Focus Group

By the end of administering the Programming Preferences Survey 84 students had

responded, spread over three levels of the programme generating around a 1000

individual responses.

76 Student responses in term 1

Year 1 = 21 Year 2 = 49 Year 3 = 6

45

24 Student responses in term 2

Year 1 = 5 Year 2 = 19 Year 3 = 0

Constant Comparison and the Generation of Candidate Codes

As part of the process of constant comparison NVivo’s automatic word frequency

tool was used to generate candidate codes. Coupled with manual comparison with

the raw data from these initial findings were fed into the second focus group. The

following discussion relates to this initial analysis and how it feeds into the second

focus group.

46

What helps you learn?

What helps you learn? Full Cohort – Entire Year

Looking at the data for the full cohort for the entire year the first candidate code is

“programs”. At this stage we shall assume that this may relate either to the act of

programming, sample code or both. The next code of interest relates to the actual

taught sessions on the module, that is “lectures” and “lab sessions”. The following

candidate codes “work” and “coding” like “programs” possibly suggest that hands-on

activities are important. The candidate code “asking” suggests that communication

is important with the tutor, peers or both. Lastly the idea of “step” perhaps suggests

the idea of pace.

What helps you learn? Full Cohort – Term Two

47

Breaking the year down into terms, the codes for term one are very similar to those

for the whole year, however the term two codes are different. In term two the codes

are, “programming”, “pathways”, “coding”, “book”, “lab”, “work”, “asking”, and “guide”.

This reflects the change in delivery pattern for all years between term 1 and 2. In the

two years of the programme, the lectures are front loaded in the two main

programming modules, this being reflected in the data as lectures are no longer

referenced. The other point of note in this data is the new occurrence of

“programming pathways”.

48

Barriers to Learning

Barriers to Learning. Full Cohort – Entire Year

In looking at what hinders the students in learning programming the initial impression

is that the main barrier to learning programming is programming itself.

Barriers to Learning. Full Cohort – Term 2

When broken down into teaching terms, term one is very similar to the whole year.

Term two appears to add a couple of new potential themes, suggesting that errors

and the ability to fix them are both barriers to learning code. Secondly there is also a

suggestion that module resources may form a barrier to learning notably

Programming Pathways.

49

Barriers to Learning. Final Year – Term 1

Looking at the final year cohort there is a suggestion that peer learning has the

potential to be an issue.

50

Tutor as Facilitator

Tutor as Facilitator. Full Cohort – Entire Year

When considering the question of tutor involvement we shall only look at the year-

long results for the entire cohort as further breakdown reveals no further significant

detail at this stage. The initial impression is that the feeling on this question is mixed,

voicing agreement, disagreement or uncertainty.

51

Use of Real World Problems

Use of Real World Problems. Full Cohort – Entire Year

When looking at the use of real-world basis for the programming problems there was

strong agreement.

52

Use of Complex Problems

Use of Complex Problems: Full Cohort – Entire Year

In looking the use of complex programming problems there is not unanimous

agreement on the matter. Agreement does appear though to be a strong voice.

There are suggestions here that the complexity and nature of the problem plays a

part in how beneficial the problem solving is.

53

Collaboration

Collaboration. Full Cohort – Entire Year

There appears to be general agreement that collaborative work is a positive aspect

of learning, in spite of indications to the contrary from the final year cohort. The initial

impression relates to themes of help, problem solving and diversity.

54

Codes for Discussion at Focus Group

Having introduced the candidate codes above, taking into account the limited time

for discussion and the quantity of discussion already in the questionnaire data it was

decided to take the following two questions to the second focus group.

1. The apparent value placed on lectures and the role of live-code2. The role of Programming Pathways

55

Summary of Second Focus GroupStudent Views on Lectures

Lectures Recorded (7)

Death by PowerPoint (7)

Live-code (6)

Variety (6)

Exposition (5)

Pace (4)

Re-enforcement (4)

Interaction (3)

In discussing lectures, having them recorded and available after the class was an

important feature. “Death by PowerPoint” was very much the strongest criticism of

poorly delivered lectures. This is contrasted by live-code which was seen to

beneficial to the learning process. Variety of delivery and exposition also serve to

enhance the lecture experience. Pace was seen as an issue in lectures. If a

lecture’s pace is too fast it may be ameliorated by interaction during the lecture for

example, tutor sensitivity to body language of the cohort if they appear lost and

lecture recording allowing re-visiting of difficult points. Delayed learning is a feature

of lectures in that the learning is obtained by re-visiting the lecture content. Lectures

should re-enforce the work set in the labs and vice versa.

56

Student Views on Programming Pathways

Programming Pathways Level development (7)

Style (3)

Copy and adapt (2)

Problem resolution (2)

Exhaustive (1)

Step by step (1)

Test driven (1)

On the subject of Programming Pathways there were various points raised. The

book was seen as a useful tool for developing programming skills. The written style

was considered accessible and easy to follow whilst the use of “copy and adapt”

meant that samples could be selected and applied to student code. The book had

good pace and level starting from the smallest unit building to a full system and its

test driven focus means that it is useful for addressing programming problems.

57

Detailed Codes for Programming Preferences Survey

Introduction

Having taken an initial look at the emerging codes as part of the second focus group

we will examine the final codes in detail. NVivo is able to produce candidate codes

automatically however this is no substitute for manually analysing the findings. The

following detailed codes and themes were arrived at by extensive manual coding of

the Programming Preferences Survey and both focus groups. Each section below

will attempt to summarise the student voice and then look in greater detail at the

breakdown of the findings.

What has helped you to learn programming?

Students are helped to learn programming primarily by how and what they are

taught, coupled with their own learning. The body of content is of prime importance

followed by the mechanisms of delivery and then assessment. Content needs to be

available on-line, appropriately paced, detailed, structured and with sample code.

Lectures should reflect what happens in the real-world, supporting materials should

be available and placed on-line. Live code is a useful way of understanding what the

code is doing and valuable in both labs and lectures. In their learning students value

both individual and social learning. The tutor plays a primary role in social learning

supported by peers as part of a team. The tutor is very important to the student and

amongst other things; they want the tutor to be available, helpful and approachable,

facilitating and signposting their learning. Peer learning works when the team offers

an appropriate range of views; it feels safer to express ignorance with peers than

with the tutor. Being in a team is motivating.

58

Teaching and Learning

Between teaching and learning, the strongest voice related to benefit to learning is

the teaching (72) followed by the learning (38).

59

The Teaching

There are three areas in the teaching in order of voice strength, The Content (39),

The Delivery (21) and Assessment (5).

60

The Content

There are a number of attributes associated with The Content.

The Content On-line (11)

Appropriate pacing (9)

Enabling and empowering (6)

Detailed (5)

Structured (4)

Instructive (4)

Clear and comprehensible (3)

Use of external sources (2)

Recorded (1)

Supporting sample code (1)

Challenging (1)

61

Delivery

Delivery is the next strongest voice of the three, in this data delivery comprises of

lectures, live code then labs.

Lectures have the following attributes:

Lectures (11) On-line (1)

Reality based (1)

Additional supporting material (1)

Live Code (3) Provides Insight (3)

Useful in both lectures and labs (1)

Labs (7) Additional supporting material (1)

62

Learning

Learning is split in to four themes, again in order of voice strength, “social learning”

(40), “individual learning” (38), “collective learning” (25) and “prior learning” (6).

63

Social Learning

Social learning covers the learning that occurs through direct interaction with another

person as opposed to collective learning which typically takes place say in a lecture.

Social learning has two focuses, firstly “the tutor” (34) and secondly the team (14).

64

The Tutor

The tutor has the following desirable attributes beneficial to learning:

The Tutor Available (9)

Helpful (3)

Clear (2)

Approachable (1)

Prompt (1)

Reliable (1)

Knowledgeable (1)

Patient (1)

Detailed (1)

Signposting (1)

Facilitating (1)

Challenging (1)

65

The Team

The team has the following attributes.

The Team Diversity (2)

Security (1)

Motivating (1)

66

What has hindered your learning of programming?

The students find their own individual learning is the biggest barrier to learning

programming followed by the teaching. Failing to understand the content, isolation,

low attendance, poor motivation, and lack of access to resources all hinder the

learning process along with lack of confidence and impatience. Failing to connect

with the tutor can be a problem. If the tutor is unavailable, unhelpful,

unapproachable or unclear in their guidance this can create a barrier to learning.

When turning to peers there may be too great a diversity of abilities leading to

impatience amongst team members. Lack of participation and lack of organisation

can create a team which ultimately becomes a barrier to learning. Sometimes the

pace of the content can be too fast for the student level. Code is hard to understand

and out of date, unavailable or limited access to materials does not help. Lectures

can be theoretical, and uninteresting whilst cancellation of classes does not help.

67

Hindrances

When it comes to the question of what hinders learning there are the following voices

to be heard, that is learning (65), teaching (13) and “nothing” (5).

68

Obstacles to Student Learning

Issues that are considered to be obstacles within Learning are individual learning

(38), collective learning (7), and social learning (9).

69

Barriers to Individual Learning

The level of the work is a concern here, coupled with isolation.

Individual Learning Level (19)

Isolation (8)

Motivation (3)

Attendance and Engagement (2)

Access to resources (2)

Lack of on-line access (1)

Confidence (1)

Patience (1)

70

Barriers to Collective Learning

Within collective learning, the only attribute here is attendance and engagement (2)

indicating that lack of attendance at classes disrupts learning.

Issues with Social Learning

For social learning there are considerations related to the tutor (8) and the team (5).

71

The Tutor as Barrier

The tutor as a hindrance to learning has the following attributes.

The Tutor Unavailable (4)

Inexpert (2)

Unhelpful (1)

Unapproachable (1)

Vague (1)

72

The Team as Barrier

The team has the following attributes when it forms a barrier to learning.

The Team Excessive Diversity (2)

Impatience (1)

Lack of Participation (1)

Lack of Organisation (1)

73

Problematic Teaching

There are the following elements of the teaching when it becomes unhelpful to

learning; they are the content (26) and the delivery (7) i.e. lectures.

74

Inappropriate Content

Issues within the content relate to the following.

The Content Pace (7)

Code (6)

Out of date materials (4)

Lack of external sources (2)

Access limitations (1)

75

Poor Delivery

Voices raising issues with the delivery relate solely to lectures.

Lectures Theoretical (1)

Uninteresting (1)

Erratic (1) i.e. frequent cancellations

76

Tutor as Facilitator

When considering the tutor as facilitator there are divided views on the role of the

tutor. The issues come when students are hitting problems and cannot progress,

who do they turn to? Surely the tutor is the fall back position who has the answers

and is there to guide the students? There are however some benefits to getting on

with the work without the tutor in that skill level improves, autonomy and confidence

are built, enhancing employability and appreciation of the need for collaboration.

Some students need more involvement from the tutor though. If the tutor is removed

from the picture it leads to poor learning, stress, struggle and ultimately de-

motivation.

The Tutor as Facilitator

The strongest voice here is that of mixed feelings whilst the remainder are split. The

findings are in the order of strength of voice, Mixed (38), Agree (28) and Disagree

(27).

77

Mixed Views on Tutor as Facilitator

Those who have mixed views about the question cite “the tutor” (30), “pace” (6) and

“level” (4) as considerations to take into account.

The tutor is seen as a resource to turn to in case of difficulty for signposting (13),

clarification (7), expertise (6) and support (2).

78

Agreement on Tutor as Facilitator

Those who agree with the statement cite a number of benefits to individual learning.

Individual Learning Level Improvement (32)

Enhanced employability (3)

Increased motivation (2)

Appreciation of collaboration (1)

Increased confidence (1)

Rising to a challenge (1)

Sense of accomplishment (1)

Increased Autonomy (1)

79

Disagreement with Tutor as Facilitator

The last voice disagrees with the statement citing the tutor (11), level (3), pace (2),

stress (2), struggle (1) and de-motivation (1) as issues to consider.

In a similar manner to those with mixed feelings on the statement, the tutor plays the

following roles.

The Tutor Resource (3)

Clarification (3)

Expertise (3)

Signposting (2)

Support (2)

80

Use of Real-World Problems

What is the point of learning anything if it does not relate to what is going to happen

in the work-place? There are many benefits which include improved ability,

enhanced employability along with motivation and confidence building. There are

concerns though; what if the real world problems are out of the level of the student,

what if there is not sufficient support for the challenge at hand

81

Use of Real-World Problems

The vast majority of voices appear to agree with this statement with the following

distribution of agree (77), mixed (3) and disagree (1).

82

Agreement with use of Real-World Problems

Those who agree cite the following benefits or real-world problems to individual

learning.

Agree Level Improvement (20)

Enhanced Employability (19)

Reality based (15)

Modern content (4)

Relevant content (2)

Builds Confidence (2)

Helps to build future expectations (2)

Increased motivation (2)

Gives meaning (1)

Research contribution potential (1)

83

Mixed views on the use of Real-World Problems

Those with mixed views cite level (2) and support (1) as issues to consider.

Those that disagree stated that there were “unfamiliar with the application of the

question”.

84

Use of Complex Problems

Many agree with the use of complex problems, however combined there are more

students who are either not sure or disagree. Those that agree see benefits to

dealing with complex problems in that it improves skill level, builds motivation and

confidence and improves depth of learning. There are those who are less sure or

even disagree. What if the problem is too complex? What if the tutor and or content

are not available for the task? This may create unacceptable levels of stress. The

support of the tutor and peers are seen as important factors to take into account

here.

Views on Problem Complexity

When asked this question there were three voices that is; agree (37), mixed (23) and

disagree (21).

85

Agreement on Problem Complexity

Those that agree cite the following benefits to Individual Learning.

Agree Level improvement (32)

Increased motivation (3)

Improved confidence (2)

Rising to challenge (1)

Appreciation of collaboration (1)

Experimentation (1)

Improved comprehension (1)

Improved satisfaction (1)

86

Mixed Views on Problem Complexity

Those with a mixed view cite the following issues to consider.

Mixed Pace (20)

Level (8)

Stress (2)

The Tutor (2)

The Content (1)

87

Disagreement with Problem Complexity

Those that disagree have similar concerns, citing in addition the role of peer support.

These responses note pace (10), level (8), motivation (2) and social learning (6) as

issues to consider.

Within social learning, both peer support and the tutor are noted in equal measure

commenting on the need for an approachable tutor and an appropriately diverse

team of peers.

88

Team Based Problem Solving

The majority agree with this statement citing the following benefits; appropriate

diversity is useful and a good mechanism for learning from others, problem solving is

enhanced by combining efforts. With the support of others the process of dealing

with issues is more efficient allowing each person to contribute in their area of

specialism. Working in a team can be a safe space without risking looking foolish in

front of the tutor, where healthy debate can contribute to the learning process in a

similar way that happens in industry. There can however be issues. Sometimes the

diversity can be a problem if some members are too far behind or too advanced.

This can lead to issues with team cohesion. Fairness in assessment and task

allocation are concerns too. It is not right that all members should get the same

grade when some have done no work. There are also Lone Rangers who do not

work well in teams and the team does not work well with them. Some feel that they

can do better on their own without the team, feeling held back. Some may feel lost in

the crowd and not confident to speak out.

89

On Team Based Problem Solving

The voices commenting on this statement are as follows; agree (74), mixed (16) and

disagree (3).

90

Agreement on Collaboration

Those that agree cite the following points.

Agree Diversity (36)

Peer Learning (21)

Problem Solving (21)

Support (12)

Efficiency (9)

Basis in Reality (4)

Specialisation (2)

Confidence (1)

Debate (1)

Safety (1)

91

Mixed Views on Collaboration

Those with a mixed response cite the following points:

Mixed Diversity (9)

Fairness (4)

Lone Rangers (4)

Problem Solving (2)

Cohesion (2)

Lost in Crowd (1)

Task Allocation (1)

Confidence (1)

92

Disagreement on Collaboration

Those that disagree make the following points.

Disagree Individual Level (2)

Diversity (1)

Efficiency (1)

Unfairness (1)

Lost in Crowd (1)

93

Learning Preferences Survey

The final sets of data to consider are the results from the Learning Preferences

Survey. In this study the survey had been modified to try and gauge the value placed

on Programming Pathways.

Students were asked to rate the following learning and teaching approaches by

rating them out of five as to their value for learning and stating their essentiality.

94

When ranked by their value for learning the list is sorted as follows. (The average is

77% with those above it highlighted in green.)

95

When sorted for essentiality the list is sorted as follows. (The average is 64%)

96

Chapter Five: Data Analysis and Discussion

Introduction

In astronomy, apparent retrograde motion relates to the motion of a celestial body

when it appears to travel backwards. When the best theory of the solar system

stated that the Earth was the centre of the universe this apparent motion was very

difficult to explain. When the Copernicans pointed out that it was the Sun at the

centre of the solar system and that the objects in question were in fact, like the

Earth, planets orbiting the Sun, the movement may then be explained as an illusion

created by viewing one orbiting planet from another (Crockett, 2017). The point is

that if a theory is not quite right then there will be anomalies when compared to

empirical data. Once the theory is refined or replaced, one is better placed to make

sense of these anomalies.

The Problem with Problem Based Learning

The title of this research project is “Constructivist Approaches to Teaching

Programming, (A Critical Review)”. In using grounded theory and mixed methods an

interpretivist approach has been brought to bear on the question. The title is

misleading though, in that it hides a third element of the research, that is the benefit

of PBL to teaching programming.

There are four areas under examination to obtain a sense of the value of PBL, that

is; the tutor acting as facilitator, use of real-world examples, use of complex

problems and collaborative learning. There is evidence of strong agreement with

these ideas for facilitating learning. In the area of real-world examples there is

almost unanimous agreement. In the area of collaborative learning there is strong

support for its benefit to learning. It also appears that students expressing a

preference for PBL style approaches cite them as personally beneficial to their

development. Certainly some students agree with this approach to teaching and

learning; however beyond a certain point the waters become muddied. A first class

honours student whilst remarking on tutor as facilitator responded, “I agree as this

pushes you to find new methods and types of code to overcome an issue.” The

student also simultaneously responding to the question of collaborative learning, “I

97

disagree with this as I prefer to work on my own where possible.” 100% PBL friendly

students do exist, however this comes with no guarantee.

Vygotsky (1978) saw mental development as a set of functions, defined as

functioning psychological systems. By this definition “capacity for intellectual

collaboration” is identified as one such system (Bozhovich, 2009). So how might we

understand a first class honours student who would rather work on their own? Are

they dysfunctional in some way? Are they somehow deficient in their learning ability?

It seems that ignoring the empirical findings at this point would be a disservice to the

student voice. For example in the areas of tutor as facilitator and problem complexity

there are strong notes of caution and disagreement. In both cases when the

challenge becomes too great for the student’s level, the fall-back position is to some

form of social learning. The first port of call is typically the tutor, followed by peers.

In comparing the student voice with PBL orthodoxy there appears to be anomalies. It

is possible these anomalies suggest that the theory is not mapping to reality and

needs adjusting or replacing. Are there are other theories which may explain more

accurately what the findings suggest?

Pace and Level

Two themes that arise from these findings that seem pivotal to the questions of

problem complexity and tutor support are pace and level. We will look first at this

finding and consider a theoretical basis for it. Pace relates to the speed at which a

problem is comprehended, level relates to the skill / knowledge level of the student.

There appears to be a “sweet spot” where, when both fall into line the student is

happy, motivated and engaged. If the pace is too fast, or the student’s level too low

then the results are problematic. It is also of note the language used to describe this

point of miss-match, “stress”, “struggle” and “de-motivation”. This is also paralleled

by the anxiety laden metaphors.

“Programming is just like learning Chinese, then being made to speak it in front of a

board with Chinese writing on and figure out what it says.”

“Learning programming feels like having to cross a big road with too many cars

driving very fast.”

98

Goleman, (1996) in discussing emotional intelligence notes the role of the amygdala

in the human brain in triggering the fight or flight response. This fight or flight

response may spur some into action, whilst others may simply withdraw from the

arena (Goleman, 1996). The indications are that learning programming may invoke

a stress response which may explain why students may be split into the two camps

of novice programmers, “stoppers” and “movers” (Perkins et al., 1989). The issue of

complexity is not to be underestimated in writing code. As an example, the author of

this dissertation with over thirty years of experience observed an error in one of two

almost identical programs. Resolving the error took two days of solid work,

comparing the code for the two systems only to find that a single back-slash

character had been omitted from one line of code in the failing application.

In the case of an experienced programmer there is sufficient confidence, motivation,

skill and discipline to persist with such an error. When considering the level of a

novice it is noted in the findings that “some of the students would be completely lost

with what to do with the problems they are facing”. The level of ability, isolation,

motivation, confidence and patience are all identified as barriers to learning

programming. Issues with pace and level may be stated in terms of processing too

much complexity too rapidly or it might be expressed as excessive cognitive load.

The Issue of Cognitive Load

Miller, (1956) noted that working memory comes with certain limitations suggesting

that it has a limited capacity of seven items plus or minus two. The theory assumes

that human cognition comes in two forms, cognition associated with long term

memory and that associated with short-term memory. In this view much knowledge

residing in a person’s long-term memory has been “borrowed” from another person’s

long term memory. Sweller, (2006) notes that by using long-term memory new

knowledge may by generated by combining existing information. Cognitive

processing associated with the short term memory relates to the “randomness as

genesis principle”, i.e. that through random exploration new knowledge may be

generated. Long term memory ultimately obtains its knowledge via both processes.

Knowledge is borrowed, via imitation, listening or reading and new knowledge is

discovered via randomness as genesis. Cognitive load theory recognises that

problem solving may take a role in the learning process; however, it also asserts that

99

relying solely on problem solving may not be the most efficient way of learning.

Noting that there are limits on people’s ability to process data, the theory proposes

that data enters working memory in “chunks”. These chunks are then internalised as

“schema” (Chase, Simon, 1973).

Cognitive load theory identifies three forms of cognitive load which need to be taken

into account for a learner; intrinsic, extraneous and germane. Intrinsic cognitive load

is that required for learning a task. For example if a student needs to understand

what a function is they must exert a fixed amount of effort in achieving this

understanding. Extraneous cognitive load is generated by the nature of the

instructional design, for example a concept “this is a function” could be describe to a

learner verbally or it might be more easily be demonstrated through the use of a live

demo. Germaine cognitive load relates the internalisation of learned content, i.e.

building the structures in long-term memory. The idea in cognitive load theory is that

instructional design should aim to reduce extraneous cognitive load allowing the

student to focus on internalising the knowledge (Sweller, Van Merriënboer, & Paas

1998). Malamed, (2017) expresses the idea of improving the clarity of the content as

reducing the signal to noise ratio. The students appear to want clarity in

communication of concepts. Students want clarity of content and also clarity from

the tutor. Students identify barriers to learning such issues with out of date content

and access limitations. Clarity of information processing may go some way to

explaining the student preference to live-code. Rather than explaining how code

works, the students state a preference for direct observation of the code being

constructed and used.

Learning Modes

One observation in the qualitative findings is that student learning appears to take

place in three learning modes, that is; individual learning, social learning and

collective learning. It appears to be the case that students shift between these

different modes, some having an apparent preference for one mode over another.

100

Individual Learning

Programming is a practical subject which takes the form of students trying the code

examples themselves. It is at this point that issues arise. Sweller, (1988) notes that

problem solving may follow a strategy called means-ends analysis. In this case a

solution to a problem is attempted and the proximity to the solution is noted. If the

proximity to a solution appears lessened then the strategy is repeated. It seems to

be the case that students see themselves as the problem when it comes to lack of

progress with programming. To resolve this they will seek to address the problem by

further contact with the content and seeking help from the tutor or peers. Problems

arise when they are frustrated at these points.

Social Learning and the Value of the Tutor

The second mode of learning noted within the findings is that of social learning.

Social learning takes the primary focus of asking the tutor followed by discussing

issue with peers. Looking at the findings from the Learning Preferences Survey there

is high value placed on the role of the tutor. When considering the importance of the

tutor to learning, in this study the tutor achieved a value of 100%.

Similar results are noted with previous runs of the survey with a much larger sample

size (Dean, 2013).

101

When looking at how essential to learning the tutor is they are given a high value.

This is also reflected in the results from the previous use of the survey.

In all sets of data the role of the tutor ranks in the top five.

These findings appear to call into question the PBL concept as tutor acting solely as

facilitator when it comes to programming. Some students appear to be happy with

this, yet others see it as a potential threat to their learning. There are several

remarks from students along the following lines.

“I disagree with this as it is very important for a new student to get help from the

tutor, because at times the programming problems are very complex.”

The findings suggest that students require a great deal of support from the tutor in

learning programming. It seems that the students may quickly find themselves lost.

In order to alleviate this problem, the tutor needs to be available and approachable.

The tutor is indeed seen as an expert however there are more attributes required

than that. The students describe the ideal characteristics of the tutor as available to

help, wanting to help and able to help whilst respecting the student in spite of their

potential sense of ignorance of the subject. The students do not just want knowledge

they also want dignity and respect. Nor do they want the tutor to simply tell them the

answer, they want the tutor to enable them via appropriate challenge, signposting

the solutions rather than necessarily giving it to them.

102

Social Learning and the Role of Peers

The other strand to social learning is the role of peers. From the qualitative findings

there are clear voices stating the value of the peer team. When looked at from the

point of view of the Learning Preferences Survey we get a sense of the value placed

on the team compared with that of the tutor.

In terms of the team’s value to learning in this research we see it below the average

at 70%, 30% lower than the tutor.

103

When looked at for the essentiality value in this survey it comes in at 50%.

104

These results are lower than that found in previous findings however the difference

between tutor and team is still observed. For example this data illustrates the value

of the team to learning.

105

The data below illustrates the essentiality of the team to learning.

Cognitive load theory may shed some light on why collaboration is important to

learning. Like PBL, cognitive load theory recognises the value of problem solving to

learning in spite of the inherent limitations on human information processing. There

needs to be problems to solve, to facilitate cognitive processing in the short-term

memory (Sweller, 2006). It is within this context that cognitive load theory

recognises the importance of collaboration. As complexity increases, the individual

learning becomes less efficient bringing in a requirement for social learning.

Through discussion with others the load is spread across the team members

(Kirschner, Paas, and Kirschner, 2009). The findings suggest that the team is the

secondary fall-back position to individual learning after the tutor.

106

The value of the team to the students is illustrated by the following student quote.

“If you work in a team, you get a chance to see lots of different techniques that can

help you solve real life problems, as well as the programming problems.”

When the knowledge has been “distributed” amongst the team and discussed it must

be re-integrated into the individual team members (Malamed, 2017). The findings

suggest that the majority of students have a preference for collaboration, however it

is of note that there are also “lone rangers” who would rather work independently

and, may it also be noted, may also not be very welcome as part of a team.

Collective Learning

The term “collective learning” has been chosen in order to highlight the importance of

the teaching, notably lectures, to the students.

Within the modules under consideration there are two forms of collective learning,

those being lectures and labs. Examining the data from the Learning Preferences

Survey we can see that labs score highly in both their value to learning and their

essentiality.

Historically in this survey, labs have always achieved the number one position, also

proving to be the case in this research.

107

A similar picture emerges when looking the essentiality of labs to student learning.

There is clearly some overlap in the case of labs with collective learning and social

learning. Labs offer potential benefit in a number of areas. Labs provide an

opportunity of the tutor to demonstrate code and present concepts. In addition they

allow student contact with the content and more importantly, when facing difficulty

may turn to social learning, i.e. the tutor supplemented by peers.

108

Lectures typically come in lower than labs in their perceived value to student learning

as illustrated below.

109

Lectures tend to perform better however when looking at their essentiality to

learning.

110

Some insight may be shed on this possible anomaly. In previous research, students

have previously noted the social nature of lectures. In comparing the experience of

learning via on-line video compared with attending a lecture, students have made the

distinction of going to the cinema (a lecture) versus watching the same film at home

(on-line video) (Dean, 2013). Collective learning should in no way imply passive

learning. Lectures may be enhanced by building in a social component for example

discussion and question and answer sessions. When it comes to the issue of student

engagement at lectures they are very clear on certain issues; that is their preference

for recording the session, the use of live-code and their dislike of “death by

PowerPoint”. It is also worth returning to the point above on signal to noise ratio. It

does not appear to be simply the case that a lecture should be just a clear

presentation of facts. Students want to be stimulated in a lecture by a range of

media, intriguing questions, even humour. The first point of contact with the content

is typically some form of collective learning, e.g. a student attends a lecture and

takes some notes. Following the lecture there may be at this point a delay in the

processing of the knowledge; however there is an expectation that there will be a

persistent record of the session in Word, PowerPoint and preferably as a video. It is

also expected that there will be obvious links between the lecture and any hands-on

sessions.

Worked Examples and Module Content

Critics of PBL have noted the issue of unfettered discovery (Ben-Ari, 1998). In

describing the implementation of PBL at the University of British Columbia, Provan,

(2011) argues that inappropriate speculation about a topic runs the risk of

constructing inappropriate knowledge. In the experiencing this curriculum at the

University of British Columbia, students may find themselves in a state of confusion

until they have gathered sufficient information to form correct conclusions.

Making problem solving the sole mechanism for learning runs counter to the way that

human cognition appears to work. In addition to problem solving stimulating the

cognition of the short term memory there needs to be appropriate peer support along

with opportunities for borrowing knowledge via worked examples (Sweller, 2006).

111

One important feature of the lectures for this cohort was the use of live-code,

especially if that demo is recorded and placed on-line. In discussing one final year

module a student made the following remark.

“We did the labs and there were no live demos until at the very end. That was the

most useful thing all year.”

Sweller, (1988 ) argues that some forms of problem solving actually interfere with

learning. Comparing experts with novices, the latter have had opportunity to form

internal knowledge schemas. This makes their problem solving much more efficient.

In the case of a novice, lacking these internal schemas their problem solving is not

just less efficient it may actually be counterproductive. Sweller goes on to argue that

problem solving may result in specific “problem solution” rather than schema

construction. Noting that problem solving as a learning device might present issues

there are implications for instructional design. If students learn solely via problem

solving, then supporting problem solving should become the primary focus of the

content construction. Cognitive load theory takes a different approach. In order for

learning to take place students need to be exposed to worked examples (Gerjets,

Scheiter and Catrambone 2004) in order to “borrow” existing knowledge (Sweller,

2006). Gerjets et al, (2004) discuss two categories of worked example. The first

types are molar examples relating to a recipe approach to problem resolution. The

student has to first identify what sort of problem they are facing and then follow the

recipe to resolve the problem. The second forms of worked example are modular

examples where complex problems are broken down into smaller units of worked

example. Their findings suggest that the latter more effectively facilitate Germaine

cognitive load and thus schema building. The two modes of worked example are not

however mutually exclusive.

112

Programming Pathways

Students noted that Programming Pathways was a very useful tool in helping them

learn how to program. From this year’s data Programming Pathways scores above

average for both its value to learning and its essentiality to learning.

As noted previously Programming Pathways makes use of elements of Test Driven

Development in order to break an entire system down into at times tiny code

examples. The idea is that students start with very small examples well within their

reach. When combined these elements start to form major system components

ultimately ending in a complete system. One student made the following comment.

“It is a step by step guide on how to do it and it also identifies the problems and how

to resolve them it is just like even though it is not recorded it has everything you

need to do. For my system this year I had to do the testing for it and it gave me a

good sense of what I should be testing for my system and it made it easier.”

The idea of “step by step” instructions was a recurring theme in the findings. By

breaking the development process down into modular examples, Programming

Pathways mirrors the modular design structure noted above.

Driving Tests and Sprints

At the heart of much of the discussion related to the way that programming is taught

on the course is the contrast between the Driving Test mode of assessment and the

sprint based assessment. Visual Web Development is the oldest module under

consideration and has become the most problematic over the last two years as

regards pass rates and engagement. The module has historically achieved good

progression rates of around 80% but since 2015 the pass rate before re-sit for the

module is around 60% falling into line with trends noted in the literature review

(Dehnadu and Bornat, 2006).

113

Some students see the Driving Test as a positive aspect of the learning. The

following is a comment from one student.

“I believe that the inclusion of driving tests, in addition to lectures, have helped me

acquire skills in programming. This is because they allow me to incorporate what I

have learned in the lectures in my lab sessions.”

The Driving Test has proved to be a useful tool for uncovering and discussing some

inappropriate mental models related to how code works. There are however a

number of issues with the assessment requiring a re-evaluation of its effectiveness.

When students do engage with the assessment the average percentage grade is

typically high, between 90 and 100%. The problems arise with the numbers of

students actually engagement with the assessment.

The Driving Test is split into four assessments with decreasing student pass rates for

each one.

For example looking at the 2016 cohort there were 143 students.

Driving Test 1 76% engagement average grade 90%

Driving Test 2 62% engagement average grade 88%

Driving Test 3 28% engagement average grade 90%

Driving Test 4 13% engagement average grade 100%

As noted in the focus group students are not continuing to the end of the process but

opting out of the latter tests. If we look at the Driving Test in the light of the above

discussion the levels of extraneous cognitive load could be reduced. One issue

raised in the findings is the requirement for students to manage their own time, i.e.

they decide when to engage with the assessment. This need for structured time is

coupled with the relational issue of building up the courage to approach the tutor.

114

One perfectly capable student made the following comment.

“You get too nervous, you know the stuff but when I do it in front of you I could not

answer. I got nervous that was the hard part.”

The Driving Tests were contrasted with the sprint based assessment used on the

second and final years of the course. At the heart of this approach are fortnightly

formative meetings between the tutor and the team. The feeling from the focus

group was that this was a more positive and supportive assessment mechanism

which should be implemented on the first year. Via the sprints relational ties may be

formed between tutor and student along with engaging with the student more at their

own level. If a student is struggling then appropriate support may be provided, if

however they are performing well appropriate challenge may be included in the

guidance. This tailoring of dialogue goes some way to addressing issues associated

with pace and level.

It is within this context that Programming Pathways sits. Programming Pathways

becomes the driving force in the code writing process. The students have their own

program that they need to write, that is the problem to solve. They work individually

following the worked examples but via the regular meetings with the tutor and the

team they are required to shift into social learning. The content is generated and

discussed in lectures and labs with a persistent resource of multimedia artefacts

available on the module web sites.

115

Happy Learning

The final theme to consider is one that runs through many of the findings, the idea of

“happy learning”. One commentator noted “For scientific purposes, treat people as if

they were human beings” (Harre, and Secord, 1972). One observation whilst

exploring the existing research, following perhaps a more positivistic paradigm, is the

absence of student thoughts and feelings. The students typically take the role of

subjects to be observed and counted, yet rarely do we get a sense of the underlying

humanity. Even in cognitive load theory couching clarity of communication as “signal

to noise” places it in an electro-mechanical domain rather than a relational one.

Within this project there has been a strong focus on the student voice. One

consequence of this is that many aspect of the humanity of the cohort have been

revealed. When it comes to individual learning the students appear to want to feel

safe and guided when studying programming. The sense of failing to make progress

is termed in anxiety laden terms associated with stress and loss. The desirable

attributes of a tutor relate not just to their knowledge but also to many interpersonal

elements such as approachable, patient and reliable. The students do not just want

lectures that are technically accurate; they want to be stimulated, challenged and

perhaps to a certain extent entertained. There is an emotive component running

through the findings.

116

The Role of the Tutor and Instructional Scaffolding

There are problems with this concept of happy learning since happiness is very

much in the eye of the beholder. In order to thrive people need certain conditions to

be in place. These are, basic safety, connection to others, autonomy, self-esteem,

self-expression and realistic limits (Winnicott, 1986). Within the idea of happy

learning are found all of these elements. Perhaps one way to address the issue of

diversity in these areas is the development of appropriate scaffolding (Bruner, 1975).

Scaffolding brings the discussion back to the Zone of Proximal Development and the

role of the More Knowledgeable Other (Vygotsky, 1978). The first assumption is that

novice students find programming very challenging. In the case of scaffolding,

support is put in place to enable the students. As with a building, scaffolding is

gradually removed once the building is secure (Beed, Hawkins and Roller, 1991).

For any level the scaffolding needs to be strong and secure but there must also be

opportunity for students to operate outside of the scaffolding as and when they feel

able.

117

Chapter Six: Conclusions and Recommendations

Introduction

In this chapter we shall consider some of the conclusions arising from the findings

and look at some recommendations to changing the design of the taught content. To

re-write an entire programme based on the above findings is clearly an onerous task,

additionally there is further reflection required to translate the findings into changes

in delivery at all levels. It would be wise though to take the above results and start to

feed them into the teaching, learning and assessment for the programme. The first of

these changes relate to redesigning IMAT1604 Visual Web Development.

Problem Based Learning versus Cognitive Load Theory

Over the years of developing material for the course, PBL has contributed positive

elements to their design. PBL’s emphasis on real-world problems and collaborative

work are both received well by the students. Yet from anecdotal evidence there was

a sense of a risk of causing students distress by withdrawing tutor support and

getting the problem complexity wrong for the level. This did not seem to quite fit the

orthodoxy of PBL but until now there was no other way of perhaps understanding the

problem. The findings suggest that problem solving in isolation is not the best way to

learn, this limitation being a feature of human cognitive processing. Taking the

position of cognitive load theory, more sense may be made of the findings from the

student voice. Sweller, (2006) identifies two principles at work in human cognition

that is the borrowing principle and randomness as genesis. If the sole learning

strategy available to students is problem solving then it runs the risk of generating

excessive cognitive load. This is especially true when it comes to learning

programming. Communicating with an inscrutable computer can potentially be a

frustrating task. If the nature of problem solving is such that we gradually move

closer to a solution via testing successful strategies, how can a student measure

progress when even the error messages appear indecipherable? The temptation to

give up and disengage is high amongst novices.

118

Scaffolding and the More Knowledgeable Other

In order to address these issues there needs to scaffolding in place. Using the

student voice as guide, this scaffolding needs applying in certain key areas, they are;

individual learning, social learning, collective learning and happy learning. More than

that there needs to be built into the module design a high level of quality contact

between tutor and learner.

Individual learning

Lacking appropriate mental models of what the computer is doing or how the code

behaves makes the task of coding an unpredictable even threatening experience. In

order to support individual learning there needs to be not only a wealth of support,

but also scope to allow for autonomy and sufficient challenge for those students who

pick up the skill quickly. This is potentially problematic however there may be certain

areas of focus that might help to address the issues. Individual learning takes place

as students interact with the content often generated by some form of collective

learning. When working at the rock face of writing code it comes down to the

programmer and the computer interacting via the selected language and

environment. It is at this point in individual learning that things go wrong and the

student is at risk of disengaging from the endeavour. It is at this point they will

typically turn to some form of social learning, prioritising the tutor followed by peer

support. In selecting delivery patterns to facilitate this, the lab session seems an

ideal opportunity as it creates a safe space for the student to try coding and then

engage in other support as required. The lab based approach of Visual Web

Development and Project Management and Development appear to be of high value

to the students in the sample. Labs sessions always come at the top of the rankings

generated by the Learning Preferences Survey. It is the intention to retain this

aspect of the module design. Lab time becomes an important nexus of all beneficial

elements, hopefully helping students to build their mental models. Individual learning

is not limited solely to the taught session taking place out of University hours. The

students need access to the taught content via the internet and it needs to make use

of a range of media including text and video. Issues related to access and

availability of tools such as hardware and software must also be addressed where

possible.

119

Social learning

Social learning is very important to the students and isolation is noted as one of the

issues that undermine the learning process. When students arrive on the first year

they are faced with one clear barrier to social learning, i.e. in the most part they do

not know each other and they may find the tutor to be quite an intimidating figure.

These barriers need to be addressed. The original intention with the design of the

Driving Test was that the tutor would to step back from the learning process, thus

pushing the students to peer learning. In the light of these findings this now seems

unwise on the first year. If the relational bonds are still to form, then this policy

potentially becomes a further barrier to facilitating social learning compounding the

level of difficulty the first year students find with learning to program.

The Demise of the Driving Tests

The following recommendations will be applied to the design of IMAT1604 Visual

Web Development. After over ten years of usage the Driving Test will be replaced

with the sprint model of assessment. The structured time created by having

fortnightly meetings should help to build connections and foster social learning

between the tutor, the student and the team.

We should not underestimate the importance students place on the role of the tutor

as More Knowledgeable Other when learning to program. There needs to be access

to the tutor in terms of physical availability and also “access” in terms of

communication and interpersonal attributes. It is hoped that the implementation of

sprint meetings should help to address these issues. Via the sprint model the

students meet individually with the tutor every fortnight, this provides opportunity to

build relational ties, ask questions and start to discuss common errors and

misconceptions. In contrast to the Driving Tests, The sprints have a strong emphasis

on formative assessment, helping to foster a safer learning space. It is hoped that

the rather gentler, formative approach to assessment will help to build bridges,

confidence and ultimately understanding.

120

The second area of social learning facilitated by the sprint model is peer learning.

Being organised into teams and working together on shared tasks should again help

to build relational ties and mutual support. This needs to be organised formally and

as noted previously one potential barrier to learning is a dysfunctional team. Teams

will need to be monitored and managed to address any issues relating to conflict and

disengagement. As with the sprint model on the second and final year the

assessment will be individual. Students stated a desire for fairness in the

assessment and did not appreciate weaker students “hiding” in the team. There

must however be room in the system for “Lone Rangers”. These students exist and

appear to perform well without such a highly structured support network. In this case

such students will have the option to form a “team of one”.

Assessment and the Content

In the light of limits on human cognitive processing there is an argument for “less is

more” so starting small is likely to be beneficial. Visual Web Development will be

adjusted to take a “concepts first” approach which might help to soften the blow of

learning programming along with allowing the students to start building mental

models related to how programming works. The number of assessments will be

reduced for the module. Rather than comprising of four Driving Tests and two phase

tests the assessment load will be reduced to two in an effort to reduce extraneous

cognitive load. The first assessment will consist of a hand written phase test at the

end of the first term. Students will be given a book of questions related to

programming concepts. Combined with the sprint meetings the student will engage

in dialogue with the tutor, providing opportunity to ask questions, show samples of

work so far and resolve problems.

Clearly at some point the students need to try and program for themselves. In

addition to coming to terms with the “softer” conceptual side of development there

must be opportunities for hands-on practise to allow them to come to terms with the

syntax and errors common to all programming. Taking a concepts first approach

does run the risk of producing students who understand code concepts but cannot

actually code. The second assessment will require students to complete an

extension to the lab exercises of their own design. Again they will be supported via

regular sprint meetings hopefully facilitating engagement and social learning.

121

One element of cognitive load theory is the borrowing of knowledge, achieved by

exposure to worked examples. The content for the modules needs to continue the

tradition of clear, step by step instructions and a suitable range of worked examples,

allowing the students to build schema related to programming. Students have also

noted that out of date or conflicting content has capacity to form a barrier to learning.

This is especially true for novice programmers who are coming to terms with so

many new and difficult programming concepts. It is the plan to review and update

the existing tutorial videos as the year commences.

Programming Pathways

One tool that appeared to be of value to the students on the second and final year

was Programming Pathways. The document also received some criticism as it was

not entirely up to date with the latest version of Visual Studio. This will be corrected

for this year. Implementing Programming Pathways on the first year may be

something to consider for the future, however at this stage the resource implications

would be significant.

Collective Learning

In discussing Social Learning and the role of lab sessions there is clearly some

overlap with collective learning. The same is potentially true with lectures by

including some sort of interactive component. Within the modules in question they

have a long standing tradition of using video recording; however it was the use of

live-code combined with video that received the most positive remarks. Live code

would seem to key in to the concept of worked-example. By seeing a task done, the

students are able to borrow that knowledge, building schema in long term memory.

This practice will certainly continue in the module design.

122

Happy learning and The Student Voice

The final note to consider in conclusion is the presence in the findings of the student

voice. Research into cognitive load theory requires measuring physiological

changes in response to how much cognitive load a person is facing. As a scientific

measure this data is clearly of value, yet we also need to keep in mind that students

have an important role in understanding the effectiveness or otherwise of our

pedagogy. On a personal note the student voice within the findings have added a

richness of tone which engages at a more emotive and less impersonal level. At the

back of these findings are learners with fears, hopes and aspiration who for the most

part just want to be happy learning.

123

Appendices

Word Clouds

What has helped you to learn programming?

All responses

i

All responses – term 1

ii

All responses – term 2

iii

Year 1- Term 1

iv

Term 2

v

Year 2 - Term 1

vi

Term 2

vii

Year 3 - Term 1

viii

Term 2No responses

ix

What has hindered your learning of programming?

All responses

x

All responses – term 1

xi

All responses – term 2

xii

Year 1 - Term 1

xiii

Term 2

xiv

Year 2- Term 1

xv

Term 2

xvi

Year 3 - Term 1

xvii

Term 2No results

xviii

You learn best when the tutor does not help directly

All responses

xix

All responses – term 1

xx

All responses – term 2

xxi

Year 1 - Term 1

xxii

Term 2

xxiii

Year 2-Term 1

xxiv

Term 2

xxv

Year 3- Term 1

xxvi

Term 2

You learn best working on real world programming problems

All responses

xxvii

All responses – term 1

xxviii

All responses – term 2

xxix

Year 1-Term 1

xxx

Term 2

xxxi

Year 2-Term 1

xxxii

Term 2

xxxiii

Year 3-Term 1

xxxiv

Term 2No responses

xxxv

You learn best by solving suitably complex programming problems

All responses

xxxvi

All responses – term 1

xxxvii

All responses – term 2

xxxviii

Year 1-Term 1

xxxix

Term 2

xl

Year 2-Term 1

xli

Term 2

xlii

Year 3-Term 1

xliii

Term 2No results

xliv

You learn best by working together in teams

All responses

xlv

All responses – term 1

xlvi

All responses – term 2

xlvii

Year 1-Term 1

xlviii

Term 2

xlix

Year 2-Term 1

l

Term 2

li

Year 3-Term 1

lii

First Focus Group Questions

liii

liv

lv

lvi

lvii

lviii

Consent Form

Title of project: Constructivist Approaches to Teaching Programming (A Critical Review)

Name of researcher: Matthew DeanPlease initial all boxes if you agree

1. I confirm that I have read and understood the information sheet [Version 1, dated October 2013] for the above study. I have had the opportunity to consider the information, ask questions and have had these answered satisfactorily.

2. I understand that my participation is voluntary and that I am free to withdrawat any time without giving any reason.

3. I agree that non identifiable quotes may be published in articles or used in conference presentations.

4. I agree to the interview being digitally audio recorded

5. I understand that data collected during the study may be looked at by a supervisor from De Montfort University. I give permission for the supervisor to have access to my data.

6. I agree to take part in this study

_________________________ ________________ _____________Print name of participant Date Signature

_________________________ ________________ _____________Print name of person taking consent Date Signature

lix

Participant Information Sheet

Title of Project: Exploring ways to help students learn programming.

Name of Investigator: Matthew Dean

You have been invited to take part in a research study. Before you decide whether to take part it is important for you to understand why the research is being done and what it will involve. Please take time to read the following information carefully and discuss it with friends and relatives if you wish to. Ask us if there is anything that is not clear or if you would like more information. Take time to decide whether you wish to take part or not. Thank you for reading this.

What is the study about?

In an effort to review teaching related to programming I am undertaking research to evaluate three areas. The first area of research relates to your experience of how programming is taught. The second area relates to the effectiveness of the taught material. The third area will be looking at the theoretical underpinnings behind how the curriculum is designed.

You are invited to participate in a qualitative and quantitative research project which forms a part of my MA in Education Practice at De Montfort University. My investigation is seeking to look at the theory of how programming is taught and review practice in the light of your involvement.

The project is being carried out in accordance with the ethical guidelines of the British Education Research Association and under the supervision of the Education staff at De Montfort University.

What does the study involve?

You may be involved in completing some or all of the following. Questionnaires, interviews or focus groups. Your involvement in these activities will help me to understand what you think and feel about how programming is taught and feed into ways that its teaching may be improved.

Why have I been chosen?

You have been chosen to take part in this study as a student studying on the course, alumni or as part of a wider network of stake holders.

Do I have to take part?

It is up to you to decide whether or not to take part. If you do decide to take part, you will be given this information sheet to keep and be asked to sign a consent form. If you decide to take part, you are still free to withdraw at any time and without giving a reason.

lx

What if I agree to take part and then change my mind?

You can withdraw from the study at any time, without giving a reason.

What are the possible benefits of taking part?

By taking part in this research you will help me as course and module leader to devise more effective approaches to teaching programming.

What if something goes wrong? / Who can I complain to?

If you have a complaint regarding anything to do with this study, you can initially approach the lead investigator. If this achieves no satisfactory outcome, you should then contact the Administrator for the Faculty Research Ethics Committee, Research & Commercial Office, Faculty of Health & Life Sciences, 1.25 Edith Murphy House, De Montfort University, The Gateway, Leicester, LE1 9BH or hlsfro@dmu.ac.uk

Will my taking part in this study be kept confidential?

All information which is collected about you during the course of the research will be kept on a password protected database and is strictly confidential. You will be given an ID code which will be used instead of your name. Any identifiable information you may give will be removed and anonymized.

What will happen to the results of the research study?

Findings from the study will be publish as part of the MA Dissertation. This will be made available to participants on request. Innovations and changes derived from the study will feed into future curriculum design.

Who has reviewed the study?

This study has been reviewed and approved by De Montfort University, Faculty of Health and Life Sciences Research Ethics Committee.

Contact for Further Information

If you require further information please contact:

Research investigator: Matthew Dean mjdean@dmu.ac.uk

Research Supervisor: Motje Wolf mwolf@dmu.ac.uk

Thank you for taking part in this study.

lxi

References

Angen, M.J. 2000, Evaluating Interpretive Inquiry: Reviewing the Validity Debate and Opening the Dialogue, Qualitative Health Research, vol. 10, no. 3, pp. 378-395.

Atwood, J. (2017). Why Can't Programmers.. Program?. [online] Blog.codinghorror.com. Available at: https://blog.codinghorror.com/why-cant-programmers-program/ [Accessed 7 Aug. 2017].

Baden, M.-S., Major, C.H. and Savin-Baden, M. (2004) Foundations of problem-based learning. Maidenhead: Society for Research into Higher Education & Open University Press.

Baden-Savin and Savin-Baden, M. (2000) Problem-based learning in higher education: Untold stories. Buckingham, United Kingdom: Society for Research into Higher Education & Open University Press.

Barrows, H.S. and Tamblyn, R.M. (1980) Problem-based learning: An approach to medical education. New York: Springer Publishing Co.Boud, D. and Feletti, G. (2003) The challenge of problem-based learning. London: Kogan Page.

Beed, P., Hawkins, M., & Roller, C. (1991). Moving learners towards independence: the power of scaffolded instruction. The Reading Teacher, 44(9), 648–655.

Bell, J. & Waters, S. 2014, Doing your research project: a guide for first-time researchers, 6th edn, Open University Press/McGraw-Hill Education, Maindehead.

Ben-Ari, M. (1998). Constructivism in computer science education. ACM SIGCSE Bulletin, 30(1), pp.257-261.

Biggs, J. (2006). Teaching for Quality Learning at University, Second Edition. The Society for Research into Higher Education.

Boud, D. and Feletti, G. (1998). The Challenge of Problem-based Learning. London: Kogan Page.

Boulay, B. du. (1989). Some difficulties of learning to program. In E. Soloway & J. C. Spohrer (Eds.), Studying the novice programmer (pp. 283–299). Hillsdale, NJ: Lawrence Erlbaum Associates.

Bozhovich, E.D. (2009) Zone of proximal development, Journal of Russian and East European Psychology, 47(6), pp. 48–69. doi: 10.2753/rpo1061-0405470603.

Bruner, J.S. (1962). On knowing: Essays for the left hand. Cambridge, MA: Harvard University.

Bruner, J. S. (1975). The ontogenesis of speech acts. Journal of Child Language, 2, 1-40.

Chase, W.G.; Simon, H.A. (1973). Perception in chess. Cognitive Psychology. 4 (1): 55–81. doi:10.1016/0010-0285(73)90004-2.

lxii

Clark, R., Nguyen, F., and J. Sweller (2005). Efficiency in Learning: Evidence-Based Guidelines to Manage Cognitive Load.

Cohen, L., Manion, L., Morrison, K. & Bell, R. 2011;2013;1993;, Research methods in education, 7th edn, Routledge, Abingdon.

Crockett, C. (2017). What is retrograde motion? | EarthSky.org. [online] Earthsky.org. Available at: http://earthsky.org/space/what-is-retrograde-motion [Accessed 18 Aug. 2017].

Chromatic. (2008). Extreme Programming Pocket Guide. Sebastopol: O'Reilly Media, Inc.

Cuevas & Fiore (2014). Enhancing learning outcomes in computer-based training via self-generated elaboration. Instructional Science; Vol. 42 Issue 6, p839-859.

Curzon, P. Learning computer science through games and puzzles. In Computers and Fun 2, York, UK, 1999.

Davies, S.P. (1993). Models and theories of programming strategy. International Journal of Man-Machine Studies, 39, 237–267

Dean, M. (2012) Use of Electronic Marking Grids (eGrids) to Enhance Assessment for both Staff and Students. Ireland International Conference in Education (IICE), 286-294

Dean, M., 2013, Simple Video Production for Academics, Conference of the International Journal of Arts & Sciences, CD-ROM. ISSN: 1943-6114 :: 6(3):81–97

Dublin, 16th – 18th April 2012

Dean, M. (2017). Programming Preferences Survey. [online] Matthew's Web Site. Available at: https://g677.dmu.ac.uk/Questions/ [Accessed 14 Aug. 2017].

Dehnadi, S. and Bornat, R. (2006) The camel has two humps (working title). Available at: http://eis.mdx.ac.uk/research/PhDArea/saeed/paper1.pdf (Accessed: 23 January 2017).

Dehnadi, S. (2009). A Cognitive Study of Learning to Program in Introductory Programming Courses. School of Engineering and Information Sciences, Middlesex University.

Denscombe, M. 2003, The good research guide: for small scale research projects, 2nd edn, Open University Press, Maindenhead.

Gerjets, P., Scheiter, K. and Catrambone, R. (2004). Designing Instructional Examples to Reduce Intrinsic Cognitive Load: Molar versus Modular Presentation of Solution Procedures. Instructional Science, 32(1/2), pp.33-58.

Glaser, B.G. & Strauss, A.L. 1999. The discovery of grounded theory: strategies for qualitative research, Aldine de Gruyter, New York.

lxiii

Glaserfield, E von (1989) Cognition, construction of knowledge, and teaching, Synthese, 80, 121-140

Goleman, D. (1996) Emotional intelligence: Why it can matter more than IQ. London: Bloomsbury Publishing PLC.

Greer, McCalla, Cooke, Collins, Kumar, Bishop, & Vassileva, (2000). Integrating cognitive tools for peer help: The intelligent intraNet peer help-desk project. In S.P. Lajoie, eds, Computers as Cognitive Tools (Vol.2): No More Walls, Vol. 5, pp. 69–96. Mahwah, NJ: Erlbaum.

Halai, A. (2004). Action research to study classroom impact: is it possible? Educational Action Research, 12(4), pp.515-534.

Harre, R. and Secord, P. (1972). The Explanation of Social Behaviour. Oxford Blackwell.

Hoover, W.A. (2007) The practice implications of Constructivism - SEDL letter, Constructivism, volume IX, number 3, august 1996. Available at: http://www.sedl.org/pubs/sedletter/v09n03/practice.html (Accessed: 10 November 2016).

Jarrett, C. and Ginsburg, J. (2008). This book has issues. London: Continuum.

Jonassen (1988). Learning strategies in courseware. In David H. Jonassen (Ed.), Instructional designs for microcomputer courseware, (pp. 151-181). Hillsdale:NJ. Lawrence Erlbaum Associates, Publishers.

Kirschner, F., Paas, F., & Kirschner, P (2009). A Cognitive Load Approach to Collaborative Learning: United Brains for Complex Tasks. Educational Psychology Review, Volume 21, Issue 1, pp 31–42.

Knuth, R. A. & Cunningham, D. J. (1993) Tools for constructivism. In Duffy, T. and Jonassen, D. (1993). Designing environments for constructive learning. Berlin: Springer.

Kuhn, T. (1970). The structure of scientific revolutions. University of Chicago Press.

Lister, R., Seppälä, O., Simon, B., Thomas, L., Adams, E., Fitzgerald, S., Fone, W., Hamer, J., Lindholm, M., McCartney, R., Moström, J. and Sanders, K. (2004). A multi-national study of reading and tracing skills in novice programmers. ACM SIGCSE Bulletin, 36(4), p.119.

Malamed, C. (2017). Six Strategies You May Not Be Using To Reduce Cognitive Load. [online] The eLearning Coach. Available at: http://theelearningcoach.com/learning/reduce-cognitive-load/ [Accessed 18 Aug. 2017].

Mason, J. (1994). Enquiry in mathematics and mathematics education. In P. Ernest (Ed.), Constructing mathematical knowledge: Epistemology and mathematics education (190-200). London: The Falmer Press

lxiv

Matthews, M.R. (1994). Science teaching: The role of history and philosophy of science. New York: Routledge

Mayer, Bove, Bryman, Mars, and Tapangco (1996). When Less Is More: Meaningful Learning From Visual and Verbal Summaries of Science Textbook Lessons. Journal of Educational Psychology, 1996, Vol. 88, No. 1,64-73.

Mayer & Moreno (2003). Nine Ways to Reduce Cognitive Load in Multimedia Learning. Educational Psychologist, 38(1), 43–52.

McLeod, S. (2007) Lev Vygotsky. Available at: http://www.simplypsychology.org/vygotsky.html (Accessed: 6 January 2017).

Miller, G. (1956). The magical number seven, plus or minus two: some limits on our capacity for processing information. Psychological Review, 63(2), pp.81-97.

Morrison, K. R. B. (1998) Management Theories for Educational Change, London: Paul Chapman, In Cohen, L., Manion, L., Morrison, K. & Bell, R. 2011;2013;1993;, Research methods in education, 7th edn, Routledge, Abingdon. P 17

Muller, C.L. and Kidd, C. (2014) Debugging geographers: Teaching programming to non-computer scientists, Journal of Geography in Higher Education, 38(2), pp. 175–192. doi: 10.1080/03098265.2014.908275.

Overton, T. (2010). Problem Based Learning: An Introduction. [pdf] HEA Academy. Available at: https://www.heacademy.ac.uk/system/files/pbl.pdf [Accessed 11 Aug. 2017].

Pavlov, I. P. (1927) Conditioned Reflexes, London: Oxford University Press.

Pennington, N. (1987) Stimulus structures and mental representations in expert comprehension of computer programs, Cognitive Psychology, 19(3), pp. 295–341. doi: 10.1016/0010-0285(87)90007-7.

Perkins, D.N., & Martin, F. (1986). Fragile knowledge and neglected strategies in novice programmers. In E. Soloway & S. Iyengar (Eds.), Empirical studies of programmers, First Workshop (pp. 213–229). Norwood, NJ: Ablex.

Perkins, D.N., Hancock, C., Hobbs, R., Martin, F., & Simmons, R. (1989). Conditions of learning in novice programmers. In E. Soloway & J.C. Spohrer (Eds.), Studying the novice programmer (pp. 261–279). Hillsdale, NJ: Lawrence Erlbaum

Phillips, D. (2015) Perspectives on learning. Teachers College Press.

Piaget, J. and Cook, M. (1992) The Origins of Intelligence in Children. Madison, CT: International Universities Press.

Piaget, J., Inhelder, B., LANGDON, F. and LUNZER, J. (1956). [La Representation de l'espace chez l'enfant.] The Child's Conception of Space ... Translated ... by F.J. Langdon & J.L. Lunzer. [With illustrations.]. Pp. xii. 490. Humanities Press: New York; Routledge & Kegan Paul: London; printed in Great Britain.

lxv

Provan, A. (2011). Use of problem-based learning at the University of British Columbia. [ebook] BC Medical Journal. Available at: http://www.bcmj.org/sites/default/files/BCMJ_53_bcmd2b.pdf [Accessed 11 Aug. 2017].

Richardson, K. and Sheldon, S. (1987) Cognitive development to adolescence: A reader. Hove, East Sussex: Lawrence Erlbaum Associates, in association with the Open University.

Robins, A., Rountree, J., & Rountree, N. (2003). Learning and teaching programming: A review and discussion. Computer Science Education, 13, 137–172

Salovey, P. and Mayer, J. D. (1990) Emotional Intelligence, Imagination, Cognition and Personality, 9: 185-211.

Savin-Baden, M. and Major, C. (2004). Foundations of problem-based learning. Maidenhead, England: Society for Research into Higher Education & Open University Press.

Schaffer, R. (1996). Social development. Oxford: Blackwell.

Schugurensky, D. (1996) History of education: Selected moments. Available at: http://schugurensky.faculty.asu.edu/moments/1969mcmaster.html (Accessed: 10 November 2016).

Silverman, A. (2014) Plato’s middle period metaphysics and epistemology. Available at: https://plato.stanford.edu/archives/fall2014/entries/plato-metaphysics/ (Accessed: 24 January 2017).

Spohrer, J.C., & Soloway, E. (1989). Novice mistakes: Are the folk wisdoms correct? In E. Soloway & J.C. Spohrer (Eds.), Studying the novice programmer (pp. 401–416). Hillsdale, NJ: Lawrence Erlbaum.

Sweller, J.; Van Merriënboer, J. & Paas, F. (1998). Cognitive architecture and instructional design. Educational Psychology Review. 10 (3): 251–296. doi:10.1023/A:1022193728205.

Sweller, J (June 1988). Cognitive load during problem solving: Effects on learning. Cognitive Science. 12 (2): 257–285. doi:10.1207/s15516709cog1202_4.

Sweller, J. (2006). The worked example effect and human cognition. Learning and Instruction, 16(2), pp.165-169.

Vygotsky L. S. (1978) Mind and Society, Cambridge MA: Harvard University Press.

Winnicott, D. (1986). Home is where we start from. Reading, Mass: Addison-Wesley Publishing Company, Inc.

Winslow, L. E (1996). Programming pedagogy – A psychological overview. SIGCSE Bulletin, 26, 17-22

lxvi