| There is no apparent reason, all the experiments where conducted identically (minor differences can occur of course, but I can't simply kick out values).
For ilustration: The experiment was done in such 96- wellplates with different substances in each collum. Then each well is measured and I get 8x the same result for each substance. For plotting I have to 'average' them somehow. br>
Eg. 2.064, 2.158, 0.18, 0.088, 0.12, 0.265. Average is: 0.813, Median is: 0.223. Speaking from experience the 2.xxx values are too high and the others are about right, so the Median would be better. (Of coruse it needs an error bar/ devination etc. for complete representation)
Manual clean up of the data isn't a good idea because bias and also because of the sheer volume of numbers (There are quite a few of these plates...)
But is it acceptable use the Median in this case? Because normally people would use the mean without much thinking. So I need to justify it somehow.
|
