What is the maximum no. of dimensions to be included in an observation instrument; and what is the ideal time interval for observing students' behaviour doing computer activities in class?
Coding is a polysemous word. Its meaning is different if you argue in a quantitative context or a qualitative one. Unfortunately, with observation both ways are possible. So your question is to narrow to be answered. Ian Dey has written nicely on codes and categories [Dey (2000): Grounded Theory. Guidelines for qualitative inquiry. San Diego: Academic Press].
I used systematic observations myself at some point in my research and struggled with the same question.
On reflection I think that the following issues are important in deciding how many categories you will use and with what time interval:
1. Both are positively related: so the more observation categories you use, the more difficult it becomes to apply them in a short time interval, and the way around. So, if you use a lot of categories, I would recommend using longer intervals. For me it was possible to evaluate every 10 seconds three people using one variable with six values.
2. What is theoretically important: if you need info on many variables, then obviously you will need a more elaborated coding scheme.
3. The resources you have: some people can observe in teams, some alone. If you observe in a team, you can probably divide the work and observe different things (so more variables) and/or in shorter time intervals.
4. The context: I did classroom observations and because of seating arrangements, I would only observe a small number of students well each lesson period (i.e.: those in front of me and not with their back to me). So if you have a context where you can/should observe many people, you might want to lower the number of variable and increase the time length.
5. How much what you observe changes within a period of time and the extent to which it is important to measure this: if what you want to observe changes quickly, and you need to measure this, then I would recommend using smaller time intervals and smaller coding schemes.
6. Practice: the more you practice, the faster you get, so give yourself some time to get used to your system before deciding to change it. Related to this: conduct various pilots of different systems and use what seems the most feasible in practice and theoretically interesting.
These are just some reflections. You might want to check other studies for examples and I can certainly recommend the following textbook (but applied mainly in educational contexts):
Systematic Classroom Observation: Second Edition (Social Research & Educational Studies) Hardcover – April 15, 200