What is the best method of assessment for computer science students?

Question

This question is a bit more philosophical so feel free to remove if you like but it's been bugging me for the last 4 years!

As a final year student I find that exams can be often be passed with a couple of days of cramming, without necessarily retaining or understanding the content i.e. a regurgitation of lecture notes is often enough to gain high marks.

A friend of mine is about to graduate with an honours degree whose final year evaluation was based solely on practical work (a project, assignment marks and the creation of a poster) yet all of this work could have been completed by a third party.

Personally I don't think either of these methods of assessment is sufficient as I am currently on track for a 1st class honours in artificial intelligence and computer science and believe this is mostly due to my skill in passing exams not my skill as a programmer or my vast in depth knowledge of any of the subjects I have "studied".

Surely there is a better way to assess our skills - isn't there?

EDIT - Just to clarify about the aforementioned friend, I was not suggesting that it is easy or even likely that a third party would complete the work, I was just trying to illustrate the potential issues with this kind of assessment. In all of my exams we have to produce photographic ID to ensure a friend isn't sitting the exam for us for example but no such similar checks can be done for assessment. Although I don't believe it likely that a student being assessed solely on practical work would be able to find someone to do the work for them but I think they would certainly be able to obtain a lot of assistance for it which could be seen as unfair, after all, I don't get access to stack overflow during my exams ;-)

Answer 1

The problem you're describing applies to education in general, not just college-level Computer Science. Yes, there are better ways to assess a new graduate's skills. In my experience, the companies that place a great deal of weight on a candidate's grades and/or certifications (and relatively little weight on the results of direct evaluation/testing) also tend to have the largest proportion of "dead-weight" programmers.

Just sayin'. Correlation is not causation, of course.

Answer 2

Unfortunately not one that will be taken on any time soon. I am a huge believer in the practical aspects of studies and project based work. It is far easier to study for an exam than it is to get a third party to complete a 6-month practical project.

One can relatively overcome the third party argument by adding in a formal evaluation and presentation that would be overseen by a bench of senior lecturers. They have the qualification and experience to shoot down students who have obviously not done the project themselves.

Fact is, there is not enough time, nor enough resources to formally evaluate let's say 1000 comp sci graduates in a period of one month (which is round about the marking schedule). There would have to be round-robin marking/presentation cycles to cater for the sheer demand and this would be an expensive process in admin and other costs.

One thing you should stand by is the effort you yourself put into your studies. Once on a sheet of paper, most will care simply that you have it. It will be you who is going to apply and try to supersede those who share the qualification. Simply put, you should do your qualification for yourself. Be curious, ask questions and gain insight and experience from those who have it.

Answer 3

As a final year student I find that exams can be often be passed with a couple of days of cramming, without necessarily retaining or understanding the content i.e. a regurgitation of lecture notes is often enough to gain high marks.

It depends on the school I think. Personally, I seriously doubt that you can get high marks just by memorising and not understanding anything. You can probably pass, it's very easy to pass exams in my college, but getting high marks is not so easy. You have to at least understand the basics. You don't need to know how variables are stored in memory in order to work with them for example, so in this sense, it's possible to get high grades without understanding everything, however a student will probably still need to know the language syntax and the algorithms pretty well.

A friend of mine is about to graduate with an honours degree whose final year evaluation was based solely on practical work (a project, assignment marks and the creation of a poster) yet all of this work could have been completed by a third party.

It could have, but do you really think he somehow had third parties help him with ALL of his exams throughout ALL of his college years? There is such a thing as nepotism and people that money can buy, but I seriously doubt any college would give out an honours degree to such a person. They might let him graduate, sure, but with honors? I highly doubt it.

Suppose he doesn't deserve it though. Who cares? What is he going to do next? Nobody will hire him (or they will but he won't last long) if he doesn't show any skill. Favors can only get you so far.

Personally I don't think either of these methods of assessment is sufficient as I am currently on track for a 1st class honours in artificial intelligence and computer science and believe this is mostly due to my skill in passing exams not my skill as a programmer or my vast in depth knowledge of any of the subjects I have "studied".

If you know this, then why aren't you doing anything about it? Why don't you study on your own? No school is going to make you a good programmer (or good anything for that matter) if you do nothing but attend its classes. All of my classes are just introductory ones. If you want to learn more, do it on your own.

Surely there is a better way to assess our skills - isn't there?

Maybe there is maybe there isn't, but I don't really think it matters. It will never be entirely accurate, and I really don't think it means much.

For example, all of our important exams are on paper. I don't think being asked to write a program on paper proves anything. First of all, if it's a complicated task you're bound to make mistakes without access to a computer. How do you determine what's a major enough mistake to cut points for and what can pass as a "it's on paper, this was to be expected" mistake? You can't. If the task is too simple, solving it doesn't say much about your skill. So it's lose-lose, but they still do it.

So basically I think you should just worry about yourself. If you think there are areas you're lacking skill in, start studying on your own and don't wait for someone to come and teach you.