Mean or Median?

#1
Hi,
maybe someone can help me with this. I have performed 3 independent experiments in which I compared a control group and a treatment group. Let's say I measured the size of cells. Now I want to plot the combined data (3 data points for each group) and check if the groups are statistically different. If the data of each experiment are normally distributed, I would determine the mean for each group/experiment and simply run a t-test.
My problem: while some of the experiments generated data that are normally distributed, others are not. For instance, in experiment 1, the control data are normally distributed, but the treatment Data not.
How can I compare this/which kind of test do I need? Could I plot for experiment 1 the medians and for experiment 2 and 3 the means?
Hope I am not too confusing...
 
#3
Sample size is between 16 and 20
In one case the mean of Control (normal) is 100, whereas the mean from Treatment (unnormal distribution) is 18, but the Median is 0
 

hlsmith

Less is more. Stay pure. Stay poor.
#4
Do you want an estimate of difference or p value style metric. If latter, a permutation test can be applicable.