how to exclude block effect (I think)

[mkb]

Hi,
I have done a simple experiment, with one control group and one treated group. Each group have between 3-5 replicates. The whole experiment is repeated three times.

If i do a normal student t.test for each experiment i get a significant difference, however if I do a student t.test collecting all replicates from all experiments, there is no difference.

When I look at the numbers it is clear that some other factor caused the cells to grow a lot more in one of the experiments than in the two others.

Can I somehow group my data depending on experiments (experiment 1, exp 2, and exp 3) and then be able to get p-value including all my data?

I really appreciate any help.

FYI: I use R to calculate by statistics

 

[terzi]

I would not recommend grouping your data since the conditions were certainly different. You can use the different experiments as blocks, in order to consider the differences in the test. A two-way ANOVA or its non-parametric equivalent could be of use.

 

[mkb]

Okay thanks. Maybe I just dont know the right words, I didn't know there was a difference of grouping or blocking my data. Anyway, I will try and look at it