Hi,

this is a really important question. Maybe, the following link can help you. The web side explains quite descriptively and clearly how the n number of your experiment has to be determined.

http://www.labstats.net/articles/cell_culture_n.html

Good luck!

S.

Hi Joe, based on your onset I would do the following.

Since the 5 fields are from one coverslip and are only used to analyse more cells to get a more valid data set I would average this onset. So lets say you analyse 10cells/field and acquiring five different spots you'll end up with 50cells/CS.

This data can theoreticly used to check for outliers before you average them.You do the same for all four CS and create the mean of these four values.

Repeat this three times in independent experiments and use the three data points for prism. based on your onset you only have a biological n of three, acquiring more images or using more coverslips is only your technical n.

In the case you have a high variety of total cells counted per experiment you might think of calculating a weighted mean.

Thanks for the link Stefanie. For the example I set up, n = 3 according to the link. From what you are both saying, does that mean you should only report and do statistics on a number of values that's equal to your n number whether n is number of animals or number of experiments?

Do people do this in practice? 

Hi Joe, in my opinion this is the way it should be.

You can also do statistics within your groups, but this will only show your inter-experimental variation (technical n).