This paper reports the learning management experiences within an intensive three-month education that helps newly arrived in Sweden nd work as IT professionals. The creation of the Software Development Academy was motivated by the migration crisis and the wider need to help integrate newcomers into the social and professional landscape. Despite having relevant skills and training, many have had their studies and careers disrupted either by con ict, or simply by lacking the pro le and networks needed to restart their careers in a new country. With limited resources and time, combined with the intensive pace and diverse student backgrounds, the program faces many challenges that threaten its success. To mitigate these challenges, an agile and adaptive approach was adopted that employs TEL techniques and pedagogical concepts to ensure the program is continuously improving via short iterations and tight feedback loops. The program has just nished its third o ering and has continuously improved through weekly collection of knowledge, con dence, and experience data that guide interventions and reactions as they are needed. The experience, process, and model presented here may inspire and bene t other courses with similar proles and challenges.
The shortage of IT developers is a substantial challenge for Swedish companies.
According to calculations from the Swedish branch organization, \IT &
Telecom companies," there is a shortfall of 70,000 software developers in Sweden in
2017 [
This is where the Software Development Academy (SDA), a joint initiative between KTH Royal Institute of Technology (a university) and Novare Potential (a recruitment consultancy), contributes to both these challenges. For people with the right prerequisites it is possible to carry out an intensive training with about 500 hours of programming teaching and practice during a relatively short period of time, that is, 3{5 months. This is su cient to meet the skills requirements for a large part of the services where Swedish companies currently have resource shortages. The overall aim with the project is to prepare newcomers to Sweden to become employable in the rapidly expanding IT sector in Sweden and give them the possibility to be employed in a sector where the demand is high. To make sure that the education is relevant for the needs of the job market, the aim is also to involve a number of employers in the project, making sure the contents of the education t their needs, and thereby increasing the chances of employment directly after the education. The education part of the project is organized and implemented by teachers from the ordinary undergraduate and Master's programs in computer science. In parallel with the IT training, an extensive job matching is conducted by the recruitment consultancy. The candidates are also provided with training in soft skills and workplace culture.
The remainder of the paper will describe the SDA program in brief, before describing the main contribution of a model that enables the program to be both agile and adaptive, despite the challenges of intense pace of delivery and a diverse cohort of students. The model is both lightweight in technology requirements as well as data gathering instruments, and as such could be readily adopted in a similar course that faces these challenges. 2
Programming Foundations
Object-oriented Development
Android Development
Group Project
Individual Project The SDA program is structured into a sequence of linked modules (e.g., programming foundations, object-oriented design, android development, as well as two project modules), as shown in Figure 1. Every week there are daily lectures and exercises. Each full day typically contains one two-hour lecture given by the module leader, followed by labs that are continuously sta ed by teaching assistants. As such, there is little room to develop modules from scratch, and instead, existing modules from the traditional degree education are repurposed, reduced and reused. While this helps with the e ciency, it does mean that the normal time given for a new topic to be assimilated into a student's knowledge base is limited. This can result in students falling behind with the topics.
Consequently, there is a greater need to continuously monitor the pace, quality and experience of each module as it is delivered, as there are no opportunities for remedial action to occur before the next module begins. Furthermore, the composition of the student cohort is very diverse. Some come with computer science quali cations already, others only come with a passion and interest to learn the subject. In amongst this, the students are typically older and most often having family and housing concerns that typically are associated with newcomers to Sweden. Given the intensity and diversity, there is a pressing need to provide immediate opportunities to adjust the pace of course, provide reviews, whilst letting students express their con dence and test themselves on a regular basis.
The SDA project draws inspiration from the use of agile methods, as a
strategy to remain adaptive to the diverse needs of the students as well as a means
to continuously improve the quality of the program despite its intensity.
Agile methods have had a profound e ect upon software engineering [
In many ways, traditional education design (following an ADDIE
approach [
Finally, formative assessment provides more opportunities to evaluate
learning, and provides more feedback interventions to help guide student progress,
rather than summative assessment [
ECK
Week 1
Planned Teaching Activities
Experience, Confidence & Knowledge
Evaluation
Reflection Teacher Discussion
Week 2
Reaction & Response
Adapted Teaching Activities
The experience, con dence and knowledge (ECK) model was developed to ensure that the SDA program could be agile and adaptive, despite its intensity and diversity challenges. The rst aspect of the model is the order and timing of key events. Figure 2 shows the sequence of events over a typical two-week period. In the rst week, the course proceeds with the planned teaching activities. At the end of the week, thirty minutes are devoted to running an evaluation session. It consists of two short surveys that evaluate student experience and con dence, and a longer quiz to evaluate their knowledge. Students receive automatic feedback on the knowledge quiz as soon as they have submitted. A teachers meeting follows the student evaluation to discuss the results. The survey and quiz software (all developed using Google Forms) visualises the results immediately for quick and convenient analysis. This meeting allows teachers from all modules to understand the current situation, and plan changes for the next week based on the data. At the beginning of the next week, the rst session presents the results of the surveys and quiz, and also explains the planned response, such as reviewing particularly challenging topics, or adjusting the pace of delivery to either speed up or slow down. Finally, any common trends from the open comments section of the experience survey are addressed during this session.
(b) Con dence survey topics for the rst four weeks.
Week 1 Class de nition Constructors Methods Types of variables Conditionals Boolean expressions Method header Variable data types Return values
Week 2 Collections hierarchy ArrayList Generic types Array types For-loop statements Iterators Documentation Javadoc tool
Week 3 Basic Git Merge con icts Iterators Streams Coupling Cohesion Code duplication Purpose of testing Testing limitations
Week 4 Exceptions Try / catch block Checked vs unchecked IO Streams IO Bu ers Inheritance Extends keyword Super keyword Overriding methods
The rst component of the ECK model is the experience survey. This is a set
of twelve questions that remain the same each week, as well as an area for open
comments. This is a modi ed version of the Learning Experience Questionnaire
developed for all course evaluations at the host university. It is an evidence-based
tool for course evaluation and analysis that helps to examine students' learning
experience based on a number of factors that have been found to improve student
learning in higher education [
The second component is the con dence survey, which comprises a dynamic list of topics taught in the current week. Table 1b shows the topics from the rst four weeks of a module. Responses are modelled using a ve-point Likert scale, from very uncertain ( 2) to very con dent (+2). The highlighted Iterators topic appears twice. This is not a mistake; instead, it re ects how this topic had to be revisited in the subsequent week based on the results of the former week's evaluation, and demonstrates the adaptation in action. Figure 3 shows the con dence results on this particular topic, with a clear shift in con dence towards more certainty. Although in e ect minor, it nevertheless re ects how an issue in con dence was detected and some immediate actions could be taken to generate improvements in student con dence. Making this intervention as early as possible avoids the chance that any topic is left as uncertain for any length of time, and learning activities are modi ed immediately rather than leaving it until later to uncover and address the issue.
(a) Week 2 Results (b) Week 3 Results
The third component is the 20-minute knowledge quiz. It contains 10-20
multiple choice questions. An example quiz can be accessed here: Week 4 Quiz.
Preparation of the knowledge quiz takes signi cantly more e ort than the
experience and con dence surveys. However, after one iteration of the program,
question banks form and speed up subsequent o erings. In line with online
assessment, like CMU's Open Learning Initiative [
Participant data in Table 2 shows some interesting di erences from the normal
experience at IT programs in Sweden in general and at KTH in particular.1 The
SDA project has more freedom to innovate in the selection process of candidates,
with the result that participant statistics di er from the normal distribution of
seats at IT programs within Sweden. In particular, percentages of women
normally tend to range between 15-20 percent in IT programs at Swedish
universities. Across three iterations of SDA, the share of female participants has been
above 40 percent. The share of female participants who manages to nish the
1 The reader should bear in mind, though, that the data set is limited and hence the
statistics are merely descriptive.
course is even higher, which is a result in accordance with more general
statistics on university studies in Sweden. The diversity in the classes is obviously
high with 9-21 nationalities represented. Syria is the most represented country
among the participants, followed by India and Pakistan. The number of
participants getting a job after the program was 86 percent after 5 months of the
completion of SDA1 (i.e., the rst o ering of the SDA program). The same
gure for SDA2 is slightly worse with 61 percent (this percentage has increased
during the last months and more participants are in interview processes). For
SDA3, which ended just a month ago at the time of writing, the gure is already
50 percent. Relative to the date of completion it is the best rate of all o erings.
The ECK model attempts to compress good pedagogical practice, student
evaluation of learning [
For students, regular evaluation was popular to test learning, as well as an outlet to vent frustrations with pace or the environment. For teachers, regular evaluation provided a focus on how to improve the program in real time. On the negative side, the quiz consistency varied across modules, due to the preparation required; modules with multiple iterations could draw previous questions, whilst the development cost for new teachers was a barrier to consistent delivery. Crucially, there was no signi cant technical burden in using the Google Forms interface to implement ECK components for each module.
With regard to the inspiration drawn from agile methodologies mentioned earlier, the general sense from teachers was that the regular feedback from students in terms of their experience, con dence and knowledge acted like a discussion with clients of a project. It provided a focus on where to make improvements and in a sense, deliver on that requirement. This had the knock-on e ect that there never were any surprises at the end of a module or an entire iteration. The commitment to continuous improvement ensured that problems were eliminated quickly, or at the very least they were recognised and discussed immediately with all the participants in the project. Within agile methods, change is welcome and to be expected, and in our experience, taking inspiration within a learning context helped to breakdown the rigidity of the curriculum and schedule and become more exible to the actual experiences of learning.
Ultimately, the process and model presented in this paper worked as intended { there was continuous communication through feedback loops with the students; each week was a new iteration, and topics were treated as prototypes that could be improved based on the feedback. The agile approach also created closer connections with students, and may inspire other intensive courses to adopt these techniques to remain responsive and adaptive. This will become more important in the future, as life-long learning environments will be populated by diverse cohorts of students, and the experiences within this program can act as an example case, with potential solutions towards adaptive learning.