IMEJ main Wake Forest University Homepage Search articles Archived volumes Table of Content of this issue

High School Students Learning Computer Science over the Web
Arto Haataja, University of Joensuu
Jarkko Suhonen, University of Joensuu
Erkki Sutinen, University of Joensuu
Sirpa Torvinen, University of Joensuu

About the authors...

The challenge to attract high school students to information technology related careers is huge. In collaboration with high school teachers in the populated district of North Karelia, Finland, the Department of Computer Science of the local university created a program to teach university-level computer science studies to high school students. The goal was to provide a flexible and reliable distance-learning environment for the students that would ensure a meaningful way of learning. The educational design is based on a Candle model: the students are faced with authentic learning needs and receive help through exercises, learning tools, and feedback, via a web-based environment. The learning material consists mainly of traditional textbooks. Preliminary results indicate that the scheme chosen has increased students' motivation to learn. The main reasons for dropping out are lack of time and difficulties in program concepts, especially arrays and methods.

1. Introduction
In Finland, the Ministry of Education is funding a three-year project to establish the Virtual University of Finland, during years 2001-2003. One of the particular goals of the project is to develop new methods for science education. The three universities in eastern Finland, University of Joensuu, University of Kuopio, and Lappeenranta University of Technology, work jointly in the virtual university project to create a web-based learning environment for a course in introductory computer science, intended for high school students. From the research perspective of educational technology and computer science education, this task is a particularly challenging one.

The project is called Virtual Certificate; high school students can take 15 credits of computer science studies in one and a half years via the internet. Courses will give students basic knowledge of three main domains: Introduction to Computer Science (5 credits), Basics of Programming with Java (7 credits), and Preliminaries of Computers (3 credits). In the Finnish system, each credit equals 40 hours of studying; 160 credits are required for the Master's degree. Thus, after passing all 15 credits in the program a student completes the first year of computer science studies. Moreover, if the student passes the program with grade 2/3, she is free to enter the university as a computer science major.

Why are we offering computer science studies for high school students? We have three major objectives:

  1. There are very few qualified computer science teachers in most of the schools. Moreover, most of the students live far from universities and each other; therefore, it is not possible to send temporary teachers to these schools or to invite the students to the university.
  2. There is a great need for professionals in the computing industry. Computer science departments have to be able to attract more students. By entering the university with the first courses completed at high school, the students will have a better chance to finish their studies.
  3. From the research point of view, it is important to assess the internet as an educational tool. As the extensive popularity of mobile technology in Finland indicates, high school students are enthusiastic about new opportunities in web-based education. Moreover, they can provide very frank and well-grounded feedback on the systems they use.

We created a model to support designing web-based instruction called the Candle model. Our main objective was to create a package that supplemented the textbook by challenging students to solve authentic learning problems or critical situations in a simple and meaningful way. By providing them with electronic candles to illuminate the learning path, (in the form of self evaluation), we assumed that students could obtain the required facilities by themselves, as they would from a textbook and an encouraging tutor.

Basically, the Candle model of web-based learning requires students to assess the support they need to solve authentic learning problems. Although the system is not yet adaptive, approaches based on prerequisites for learning a given topic and outcomes of learned material, like those used in AHA (De Bra & Calvi 1998) and InterBook (Brusilovsky et al. 1998), may be utilized in our model. However, contrary to the common uses of these systems, students do not access the basic learning materials through the web; even in the future, we will make full use of textbooks. The Candle-based environment consists of various targeted tools that help students, who are studying textbook-based material, to solve authentic learning problems.

Most of the discussion around web-based education, like that which surrounds virtual universities, stresses the importance of users' needs. After all, distance-learning technology, with little human contact compared to regular education, requires a highly motivated student. Whereas in the traditional setting face-to-face instruction or hand-to-hand-guidance compensates for obscure learning goals, the very same obscurity often shadows a web-learner's path, or learning process. Therefore, at least the final goal must be clear enough to encourage a lonely learner's walk. The Candle model is designed to illuminate the path as well.

2. Design Principles
The educational and instructional design of the package was constructed in close collaboration between high school and university teachers. We found that the contribution from the high school teachers was crucial for the development of the Candle model. They helped us to coordinate the courses to the schedules of the schools - which differ quite a lot from school to school! Moreover, the teachers were able to give us a profile of these students that included both their limitations and their desire to be a part of the university: therefore, it was essential to give at least a few real lectures and examinations. A group of teachers monitor the course during the whole period of 1.5 years.

2.1. Educational Design: The Candle Model
The Candle model is designed to support a student

We will elaborate the ideas as follows:

In Candle, almost all teaching is carried out via the internet because the students come from all around the district of North Karelia. Students do not have to spend their time and money traveling from their home to the university.

An example of Candle model. This particlular Flash-animation deals with understanding of If-statement and relational operators. Although the animation is very simple, it gives an idea about what we mean when we talk about activating learning materials/tools.
Require Flash plug-in.

Thus, cANDle emphasizes real human contacts for the learners. The web site helps the students to communicate with the instructors and one another in multiple ways. Instructors are Master's level computer science teachers working at the university. Furthermore, each school has a tutor-teacher who supports students in different situations. Tutor-teachers do not need to be experts in computer science but their pedagogical knowledge is crucial. They are experienced teachers who can support the students' learning process sufficiently. The students also help each other, both locally and over the web.

In our experience, sophisticated user interfaces in many web-based educational settings might be confusing for a novice. The features tend to be superficial or shallow, and often have no pedagogical importance. Therefore, candLe keeps the user interface simple. To make the interface of the learning environment more student-friendly, we hired a fresh high school graduate to design the pages.

Another major design principle in candlE is to link the printed learning material with an activating learning environment on the web. The printed material gives students an overall structure and knowledge of the domain while the web materials guide the learners step by step. Activating visual tools like Jeliot, Excel, and BlueJ serve as students' virtual laboratory. Exercises play an important role in the learning process. Especially in programming, learning-by-doing is effective. Each student must complete at least 1/3 of all the course exercises; those who exceed the minimum requirement get bonus points for the grade. The bonus point system has been a motivating factor in increasing the response rate. Although we have restrictions about the number of exercises students may submit, we have been as flexible as possible concerning practical matters like timetables to meet the needs of most students.

2.2 Learning Environment and Course Materials 
We have built a specific learning environment for the students through WebCT. The students can connect to the environment with any commonly used browser (Internet Explorer, Netscape); no special equipment is required. WebCT works with a low bandwidth. This is essential because the students often connect to the environment at home by modem. The environment includes learning materials, discussion forums, chat, personal homepages, etc. The environment is closed for outsiders; every student has a personal username and password for the courses.

Course materials on the web that support the printed materials are composed mainly in HTML. The course textbooks provide primary information of the subject; web materials are designed to help students in critical situations. The web material brings the most essential parts of the course to the learners, and it quickly shows them the structure of the domain. Furthermore, web materials provide examples, visualizations, and interactive experimentation to the students. We designed the web material to provide active learning experiences and help the students in their learning process.

In addition we are planning to use video clips as mini-lectures. These lectures will give extra information and improve comprehension in specific or more complicated areas. The clips will be approximately 5 minutes long because we think that longer clips will bore high school students. With short clips we can focus on one or two specific pieces of information or skills.

2.3 Visual Tools
Visualization has long been an important pedagogical tool in computer science education. The use of the web and interactive animations provides opportunities to expand the availability of visualization-based teaching and learning tools. We have been experimenting with BlueJ, Jeliot and Excel that provide activating learning tools. BlueJ is an integrated visual teaching environment and language. BlueJ helps students to understand object-oriented concepts such as objects and classes, message passing, method invocation, and parameter passing (Kölling 2000).

The Jeliot I environment allows a web user to animate Java programs of his own over the internet (Haajanen et al. 1997). The user writes a Java code in a text field of a web page, submits it and gets back its animation. The animation is generated automatically from the source code, and it is displayed on the user's screen. Built on an extensible architecture, Jeliot can be modified to animate most common data structures. Jeliot 2000 is a program animation system intended for teaching computer science, especially to high school students (Ben-Ari et al. 2000). The emphasis is on program animation that demonstrates the execution of input-output, assignment, selection, and loop statements.

A screenshot AVI movie (~400 KB) showing Jeliot.

We have also been studying how to prepare animations of simple algorithms with a Microsoft Excel spreadsheet program. Excel offers a light, adaptive and inspiring platform for creating visualizations of various needs in computer science (Dybdahl et al. 1998, Rautama et al. 1997). A teacher can prepare visualizations for teaching or visualizations projects can be assigned to students. Two standard features of Excel, i.e. data visualization and macro programming with VBA, form together an environment to animate algorithms.

2.4 Additional Teaching Methods 
Most of the courses have a similar structure: students have material to work with; they do weekly assignments; and their learning outcomes are evaluated through an exam. However, we have tried to add flavor to studies by employing a few other teaching methods:

3 Evaluation 
Our first 93 students enrolled in the program in August 2000. After the first two months, there were approximately 80 active students, coming from eleven different schools in the district of North Karelia, as far as 100 km from the university.

3.1 Course: Programming, Part 1 (2 credits) 
We evaluated the second completed course, Programming, Part 1 (2 credits), which covers the introduction to the concepts of programming in Java. The general feedback from the students and the teachers has been mainly positive, which in the Finnish culture indicates that the chosen teaching method has been successful. We have used both questionnaires and informal queries to assess the students and teaching settings during the course. Based on the analyzed feedback, the curriculum was modified to match students' potential. To participate in the examination, a student had to complete 1/3 of the given 50 exercises. Out of 79 students, 76 submitted at least one exercise and 62 of them returned the compulsory number of exercises. Figure 1 (n=76) illustrates the distribution of the submitted exercises.

Figure 1. Distribution of the submitted exercises

Out of 62 eligible students, 56 took part in the exam. The exam included four questions on the following themes:

Figure 2 shows the results of the examination (3=highest mark,..,1-= lowest mark). The average grade of the exam was 2-, which is a little bit higher than that of the regular courses in our department.

Figure 2. Grades of the exam

One can easily recognize the reverse Gaussian curve in the diagram of the grades illustrated in Figure 2. The frequency of the lowest and the highest grades is much greater than the middle grades. Although the number of low and failed marks was quite high we think that the results were satisfying.

Three categories of students emerged. The first group is managing rather well (above 2-) and they have a good chance to reach the average mark of 2/3, which allows them to continue their studies in our department upon their enrollment. The second group consists of students who had trouble both in exercises and in the exam (marks 1+,1). Future course assessments reveal that they have a good chance to pass the virtual certificate, but they probably won't get the average mark of 2/3. The third group is comprised of students having major difficulties even in passing the first courses. In many cases, these students have ended their studies either after this course or in following courses.

3.2 Discovered Difficulties in Learning to Program 
Let us now consider the students who did not continue the "Virtual Certificate" after the first programming course. The number of the students who gave feedback is 25. Eight of them decided to quit before taking the examination.

Quite unexpectedly, as many as 48% of the students regarded if-statements as difficult to learn (Table 1). Surprisingly, a total of 72% considered arrays difficult.

Topic of programming n=25
Variables and symbols 24 %
Assignment statement & input/output 32 %
If-statement and logical operations 48 %
Loops: for, while and do-while 24 %
Arrays 72 %
Methods 60 %
Applets and graphics 60 %
Making animations 56 %

Table 1. Most difficult topics in the course

Let us then look at how the topics were scheduled during the course. Figure 3 shows that most of them occurred in the last half of the course and, at the time, there was an exam period in high schools. So quite a number of the students had difficulties synchronizing their high school studies with the "Virtual Certificate" courses.

Figure 3. How difficult topics appeared during the course

4 Future Work 
From the tutoring point of view, the Candle model needs to be developed further. By default, the model assumes that the learner is motivated. Thus, the teaching intervention can be focused on a learner's authentic occasional difficulties, not his or her motivation as a whole. However, it would probably be useful for even a highly motivated learner to understand the relationship between a partial learning step and the overall learning process. To manifest this relationship, one can make use of several visualization schemes where a life-long learner can see and understand his or her learning status within a larger educational setting. Therefore, a new model will be developed in collaboration with the experts in student counseling.

Many computer science professionals think of their field as a truly universal one. However, learning difficulties identified at early stages, as when one is doing the first programming assignments at high school, might help us to recognize contextual difficulties a student might have in comprehending even abstract structures. For instance the terms of the assignments have to be different at the high school level than at the university level. Therefore, we will localize our environment to other cultural contexts, in order to analyze this problem.

Although some high school teachers were involved in the planning phase of the courses, there were still difficulties in synchronizing the Virtual Certificate studies with the schedules of the high schools. Hence, the collaboration with the high schools has to be deepened. In this way we can eliminate unnecessary barriers in the student's learning process.

Traditionally, elementary school teachers, especially those working at small village schools in rural areas, have been called folk candles, bringing the light of the civilization even into most remote places. Hopefully, our Candle model can work in the same direction, at the time when most of the people in the industrialized countries are moving into cities, like the misled children in the famous folklore story on the pied piper of Hamelin. The Candles remain as long as these areas belong to the rest of the networked civilization.

5 Acknowledgments 
We wish to thank Niko Myller for designing the multimedia material. We are also very grateful for the revisers' efforts to enhance the appearance of the article. They have given us valuable new viewpoints during the revision process.

6. References
Ben-Ari M., Levy, R. & Uronen, P.A. (2000). An Extended Experiment with Jeliot 2000. Accepted for publication in the Proc. of the Program Visualization Workshop, Porvoo, Finland.

Brusilovsky, P., Eklund, J. & Schwarz, E. (1998). Web-based education for all: A tool for developing adaptive courseware. Computer Networks and ISDN Systems (Proceedings of Seventh International World Wide Web Conference, 14-18 April 1998) 30 (1-7), 291-300.

De Bra, P. & Calvi, L. (1998). AHA! An open adaptive hypermedia architecture. The New Review of Hypermedia and Multimedia 4, 115-139.

Dybdahl, A., Sutinen, E. & Tarhio, J. (1998). On animation features of Excel. Proc. ITiCSE '98, Dublin, Ireland.

Haajanen, J., Pesonius, M., Sutinen, E., Tarhio, J., Teräsvirta, T. & Vanninen, P. (1997). Animation of user algorithms on the Web. Proc. Visual Languages '97, Capri, Italy.

Kölling, M. (2000) BlueJ - The Interactive Java Environment.

Rautama, E., Sutinen, E. & Tarhio, J. (1997) Excel as an algorithm animation environment. Proc. ITiCSE '97, Integrating Technology into Computer Science Education, ACM, Uppsala.

********** End of Document **********
IMEJ multimedia team member assigned to this paper Yue-Ling Wong