Aagh!! - % values - parametric or non parametric?

#1
Hi good people

I am a very confused PhD student, and despite my best efforts over the past 24 hours have been unable to get a definitive answer to my question. So this forum is my last resort!

In one of my chapters analysis is done on percentage values - without boring you all too much i am looking at the percentage of cells responding to a stimulus per field of view of a microscope. So assuming i have 10 fields of view, and in each field of view i have 20 cells; therefore for each field i count the number of cells responding out of the 20 and then calculate the corresponding percentage value. I have two groups - stimulated and unstimulated. Now, in order to compare between stimulated and unstimulated systems (10 percentage values each) i have been using an unpaired student t test. However, it seems that i can't use this test because the fact i am using percentages means that my data is actually non-parametric. I was told this by my friend (who heard from someone else who is not around to ask anymore) but now we have been asking around no-one seems to know and the only thing they have suggested is doing a normality test to decide whether the data is normally distributed or not. But this won't actually help me because even if my data comes out as normally distributed its not going to tell me whether percentage values should be classed as normal or not!

Could someone please let me know if using percentage values automatically dictates that a data set is non-parametric. I don't want to change my graphs unless absolutely necessary (non-parametric will mean mann whitney tests and box and whisker plots which seem pretty **** scary to me and i can't do them on Excel).

Any help would be appreciated. Thanks
 

dtae

New Member
#2
I am pretty sure percentages can be considered continuous and non-parametric..especially since you could relatively easily convert your percentages into a numberic value.
The normality of the distribution is a seperate issue which you should still examine.

I would not worry about the percentage thing.

However, you also should take at least a basic stats class...trust me it will be helpful for your own research and understanding others publications.