I only have n=2 for each group though, so I don't think I can really do these checks...

ooh, not sure you can do any stats if N=2!

I hate to ask, but why have you only two replicates for each treatment. You'll struggle to publish the results with that.

Strictly speaking you can't use ANOVA because you cannot really calculate a variance for N=2. If you can't increase the sample size then you have two options. Firstly, write a descriptive paper and simply describe your observations, this is quite often more difficult to publish. The alternative is to take a statistically ignorant approach and just use the ANOVA, but you'll need to check that the data conforms to the assumptions of ANOVA. If it doesn't your into a world of pain and suffering.

I'd suggest you read Underwood (1997) Experiments in Ecology. Its a whole book, all about ANOVA, so its a good place to start!

Good question about why n=2. I'm doing the masters year of a 4 year PhD, and my lab project only lasts 10 weeks. My sup thinks I've done well to get better than n=1 - there's no way it's going to be published anyway, so it's just about doing what I can in the time available.

In which case, just do it visually and scrap the statistics, your results will be virtually meaningless.

Dan B - now that's the kind of answer I like! To be fair, my supervisors have pretty much said that they're not important. In fact, most academics in my area seem to loathe stats, and see them as a means to an end, rather than actively enjoying putting them in there.

Sorry, I should have perhaps said I did a PhD in statistics and work as a statistician for a living - whilst I'm always happy that people should want to use statistics (and I try and people to get to use more!) sometimes it isn't appropriate. In this case, it's nice you'd like to try something but it really will just give you results that lack any meaning. Sometimes another method actually will have far more meaning, such as displaying the results graphically.

Now I just have to get the people at work to stop using dodgy statistics to get their point across, despite being totally misleading

:-s