3

My experiment gives data of 12 biological replicates. This would be sufficient to calculate the mean and give a valid standard deviation. However, there might have been some pipetting errors that affect all replicates. So I repeat the experiment from scratch 3 times. Now I have 3x 12 samples. How do I represent the data? Take the mean of the 3 means and the standard deviation of the 3 means?

EDIT: more info;

I am comparing protein expression between constructs transformed into the cells. Thus I have like 10 x 12 biological replicates (10 means that I compare). This whole thing I repeat 2 times. Should I pool the data if I see that they are all really similar?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
SecondLemon
  • 151
  • 4
  • 1
    Mean and standard deviation are just simple descriptives of the data - there's no rules forbidding you to calculate them, it's just that they might not represent what you think. You could add more on why do you need them/what comparisons are you going to do. For example, if I had 12 cell lines and wanted to show the variability between them, I'd first calculate the means for each cell line and then an SD of those 12 means. – juod Jul 31 '17 at 10:34
  • Plot the data! Maybe the ones you suspect pipetting errors stand out, then maybe you should simply remove them? – kjetil b halvorsen Jul 31 '17 at 11:21
  • @kjetil b halvorsen They are really close to each other, I feel like I should be allowed to simply pool the data into 1 big set of 36 points. But ethically I am not allowed to just pick 1 set. So how do I then merge the data. – SecondLemon Jul 31 '17 at 12:33
  • I spoke with some internal people and they said it can be pooled since they are all replicates. It is still good to do it separately as it proves that no mistakes were made if the data is close to each other. – SecondLemon Aug 01 '17 at 09:34

0 Answers0