I know that in theory I had a statistics class...10 years ago. From which I remember, oh, approximately nothing. Now I am in a situation where knowing a little bit of statistics might help
( Read more... )
So far, the repeated measures ANOVA was good in that it indicated a significant linear trend in the data (a secondary analysis you can pick) (which makes sense and is obvious to the eye when looking at the plots) - and I like knowing that the t-test isn't 100% accurate, because otherwise the results make no sense... according to t-test significance values, the compound 'inhibits' at 0.1 mM, doesn't at 1 mM, and does at 10mM...most likely the real result is that the cutoff is somewhere between 1 and 10 (based on completely unrelated data), and the 0.1mM 'inhibition' is just an artefact and wouldn't hold up if I actually had more than 4 data points.
But in any case, I've gotten what I needed, which was a way to 'judge' which compounds might have a significant effect and which probably don't. I'm going to do another experiment to determine the 'inhibition constant' (yes I know this isn't really the right name for it, but this data is really just to tell us where to go in future experiments and how concentrated we really need things, since solubility is a big issue for some of them) for the control compound, but for that one, I'm going to have 8 data points/condition. At least now I know where to focus the test :-).
For anyone overly concerned about my use of these statistics and how accurate or not accurate they are - its all of my own use and nothing is getting published based on it, so if something is wrong the worse it can do is send me in a minorly wrong direction for a week or so.
So far, the repeated measures ANOVA was good in that it indicated a significant linear trend in the data (a secondary analysis you can pick) (which makes sense and is obvious to the eye when looking at the plots) - and I like knowing that the t-test isn't 100% accurate, because otherwise the results make no sense... according to t-test significance values, the compound 'inhibits' at 0.1 mM, doesn't at 1 mM, and does at 10mM...most likely the real result is that the cutoff is somewhere between 1 and 10 (based on completely unrelated data), and the 0.1mM 'inhibition' is just an artefact and wouldn't hold up if I actually had more than 4 data points.
But in any case, I've gotten what I needed, which was a way to 'judge' which compounds might have a significant effect and which probably don't. I'm going to do another experiment to determine the 'inhibition constant' (yes I know this isn't really the right name for it, but this data is really just to tell us where to go in future experiments and how concentrated we really need things, since solubility is a big issue for some of them) for the control compound, but for that one, I'm going to have 8 data points/condition. At least now I know where to focus the test :-).
For anyone overly concerned about my use of these statistics and how accurate or not accurate they are - its all of my own use and nothing is getting published based on it, so if something is wrong the worse it can do is send me in a minorly wrong direction for a week or so.
Reply
I'll stop writing superfluous code for my own amusement then. :)
Reply
Leave a comment