First, I'm new to statistics, so I only know the very basics like the SD, mean, mode .etc. I've conducted an experiment so that I have two sets of data in hand, I want to know if the difference observed in the experiment set and the control set is significant or not. If I have the sample size, mean and SD of each group, could I calculate the confidence level. Someone suggested T-test and p-value, I surfed the internet for info, but I just cannot grasp the idea clear, so could someone explain them to me? Any help would be appreciate a lot, thanks.

Edited August 27, 201015 yr by dttom

I assume that you have continuous values and that the type of measurements is likely to be normal-distributed? In that case a Student's t-test is applicable. In short, it tests whether two data sets are from the same population or not. The null hypothesis here is that both are from the same population, and if the tests refutes it, it indicates a statistically significant difference between the two data sets. The actual calculation is pretty straightforward and can be found everywhere. In short, it is based on the means and standard deviation of the the data sets. Means that are very far are apart are more likely to be from different populations, whereas a large s.d. reduces it (as in a very flat distribution two values may be farther apart but still have a high likelihood of belonging to the same population). This is reflected in the formula. The resulting value is compared to the t-table and based on the degrees of freedom you can read there with which probability you can refute the null. In most cases a p of <= 0.05 is considered significant.