Sunday 14 November 2010

Open Source Development in Malawi

Tuesday 9 November 2010

Experiments: why and how

Thursday 4 November 2010

Top 5 writing mistakes made by project students

A large part of any final year thesis project is the write-up, but every stage of the project will involve some form of writing. Whether you are writing a proposal, an interim report, a draft report, documentation or the final thesis, a very large part of your time will be spent composing text. These are the top ten mistakes we see from students year on year, avoid them and you can save yourself a lot of time (and earn a lot of marks)...

 

1. Unsubstantiated claims

As a general rule, any statement of fact or opinion about your work, or your topic of study should be substantiated in some way. You don't have to worry about the bleeding obvious, we know that 1+1=2 (unless your work is in the "foundations" of mathematics or philosophy, in which case 1+1 may well be undefined), but anything less obvious should be backed up by solid evidence. That evidence can be a reference to literature, an argument you make in your writing, the results of your own experiments, or any other suitably rigorous evidence.

Avoid so-called sweeping statements that are effectively impossible to substantiate "all Object Oriented programs couple algorithms and data". Do they? All of them? "Testing improves the performance of students". Really? These sorts of claims are poor academic practice and suggest that you have been a little sloppy in your thinking about your work, and of course that's not the impression you want to give.

 

2. Dangling pronouns and other references

 This seems to be an almost universal problem with student writing. Consider the following paragraph:

In his 1953 paper, Henry Gordon Rice proved that for any non-trivial property of a partial function there is no general decision method to determine an algorithm that computes a partial function with that property. It has far reaching consequences for compilers, static analysis and other fields in practical computing.

What does the "It" in the second sentence refer to? Rice's theorem, his paper, or something else? It isn't clear from the text, although we can guess that the author meant to discuss the theorem. Better though, to be clear about the meaning in the first place. Every pronoun ("I", "he", "she", "it", "that", "who", etc.) should clearly refer to exactly one noun. The first sentence gets this right, it is clear that "his" refers to the noun "Henry Gordon Rice" and not any other noun in that sentence. So, we could improve the paragraph above by re-writing it:

In his 1953 paper, Henry Gordon Rice proved that for any non-trivial property of a partial function there is no general decision method to determine an algorithm that computes a partial function with that property. Rice's theorem has far reaching consequences for compilers, static analysis and other fields in practical computing.

 

3. Using a secondary source rather than a primary one

It's much easier to read the popular press, blogs and other "informal" media than research papers. Very few marks will be awarded for this, though. In a final year project you need to show that you can perform a small academic study, so marks will be available for reading peer-reviewed academic literature. There are two major pitfalls to avoid here. Firstly, you will occasionally come across some disreputable conference or journal which does not use peer-review. Worse, it is possible to come across "articles" on the Internet which have citations and publishing records and look, to all intents and purposes, like a genuine piece of academic writing, but have actually never been submitted to a journal or conference. This is very poor practice on the part of anyone who puts this kind of thing up on their own blog or website, but it does occasionally happen. A reputable publishing venue will have some sort of statement on their website stating how articles are reviewed (look for "Instructions to Authors"). This should say that every article is reviewed by at least two people and sent back for corrections before being published. Without this sort of peer-review any poor standard of work can be "published" without anyone checking even basic standard of good practice, such as detecting plagiarism. That said, even with peer-review, some poor practice still slips through the net.

Secondly, whatever references you cite should be primary, rather than secondary sources. A primary source is one where the author(s) reports work that s/he (they) have personally carried out. A secondary source is one where an author reports on work that someone else has completed and published elsewhere. Secondary sources include news reports, magazine articles and blog posts about research completed by others.

 

4. Confusing structure and use before definition

Any academic writing should generally be written for a reader who is an expert in the general field of the study, but not necessarily in the specific area of the study. If, for example, your final year project is on genetic algorithms, your work might be marked by an expert in artificial intelligence, or in computer science generally, but not necessarily by an expert in GAs. So, write with that in mind, and make sure that you don't use any specific technical terms without defining them. This is often very difficult to get right at first, especially when you have been working with your own ideas for a very long time. A good plan is to swap drafts of your work with a fellow student who is on the same degree course but working in a completely different field for their final year project. If you can both understand each others work, that's fine. If not, make changes.


5. Claims of "proof"

At the beginning and end of your dissertation you will want to set out the aims of your work and describe the conclusions you have reached. Occasionally we see students writing sentences such as "this study proves that ...", "this thesis will prove ...", and so on. "Proof" is specifically a mathematical method, and if you have genuinely proven a theorem, by all means say so. If not, then don't use the word "prove" and be very careful about what you do claim. For example ...

  • If you have tested a piece of software and it has passed all your test cases, then you have shown that your software is free of the specific errors you have checked for. You have not shown that it is "error-free" or "works".
  • If you have performed some sort of user testing, or any other usability / accessibility testing, then you may have demonstrated that the system under study is "usable" or "accessible" as far as you have tested it. However, without a large-scale study that is as much as you can claim. Be very circumspect about reading research in the area of usability; there is much good research but also much which is over-blown. Be especially careful of authors who also run consultancy practices, make sure you cite their academic literature, and not anything that could be considered advertising. Make sure everything you cite is peer-reviewed.
  • Be very, very careful about using questionnaires. I usually tell all my students to just avoid them altogether. It is very, very hard to produce a questionnaire which holds up under academic scrutiny and you will need an amount of statistical sophistication to produce sensible results. Also, you need a very large sample-size because your questions will be circumscribed (and for other reasons). This makes questionnaires very difficult to use in short, single-person projects. If you are in any doubt, then use a "semi-structured interview" to interview test subjects and the "talk-aloud protocol" or "cognitive dimensions of notation" for usability testing. Before you start, read some papers on evaluation methods, such as Hollingsed and Novick (2007) Usability Inspection Methods after 15 Years of Research and Practice, or Hornbaek and Law (2007) Meta-Analysis of Correlations Among Usability Measures.

 

Posted via email from Pass your final year project