POLI 210 Lecture Notes - Lecture 5: External Validity, Internal Validity, Observational Study
Document Summary
Observational methods: case study, medium-n: very basic statistics. Use qualitative and quantitative together: large-n: de nitely quantitative methods. Experimental methods: surveys, eld, natural, lab. Experimental approach is the best to determine that two factors are causally related. Can control other factors that might be responsible for the two. This is not possible for the large-n analysis. The experimental because of its nature, we can isolate the factors that we want to study. The only thing that differs is the manipulation. Randomization is the key for establishing causality. Everyone that you are experimenting, you can conclude more easily because the groups are more similar. Time order is not given in other methods. Random assignment prevents self-selection into treatment( control group) Self-selection happens so many ways and having a tool to randomize is an advantage. Experiments might be problematic in external validity. In the lab you are forced to watch the computer screen.