T test - (Jul/01/2008 )

I am having trouble applying statistics to my data. I am doing a migration assay.I have repeated the experiment twice and each time I had two replicates for each treatment group.My question is whether I will be able to apply a student T test to my data since I was told that I had do the exp atleast three times for a student t test. Please help me with this.

What I wanted to ask was can I use the technical replicates towards the T test so that I will have in all four replicates; two biological and two technical.

The answer is no. You will need to average the replicates from each experiment, they are not independent a you used the same dilutions etc to set them up. If you try to use each one as an independent sample then you will have a huge standard deviation about each point so you will most likely get no difference in your statistical analysis.

Student's T-test relies on large sample numbers (>30) and normal distribution of the data for the test. You should investigate non-parametric analysis that can cope with low sample numbers.

was just wondering what statistics program Immunologist was using. Statistica's non-parametric tests for multiple dependent samples gives you one p-value and so you end up not knowing which groups differed from which. bob1’s right about needing a large sample size. possibly also one of the reasons your data is not normal. you could try transforming your data to try and get it to be normal, for example log10 or arcsine transformation. if your data is normal then you could try a parametric test, if not, good luck with non-parametrics.

Thank you so much for your input.I finally ended up repeating the experiment two more times.