Using a true experimental approach in social research is VERY difficult if not impossible. In a true experiment, the subjects must be assigned to the levels of the independent variable randomly. Random assignment is not often possible in social situations. An example where it might be done in a smaller scale study is to compare attitudes of two groups of individuals based on a specific stimuli. In this case, you could randomly assign the subjects to two groups and then randomly decide which group receives which stimulus.

I agree with James E. McLean about true experiments in the social sciences, but I also want to draw attention quasi-experiments as an alternative. Take a look at Reichardt & Little, Quasi-Experimentation: A Guide to Design and Analysis

David L Morgan makes a very good point. I omitted saying anything about quasi-experimental designs as the question was focused on experimental!

thank you all colleagues for the answers