This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Chapter 5 Controlled Experiments and Observational Studies It is useful to distinguish between “controlled experiments” and “observa- tional studies.” In many experiments, for example, often an investigator wishes to know the effect of a treatment (like a vaccine, university education, government support, etc.) on a response (like exposure to disease, earning higher in- come, economic growth, etc.) To learn more about this, the researchers often compares responses of a treatment group with a control group. The control group consists of individuals who are not exposed to some variable which its effects we are interested in, while the treatment group is exposed to the variable of interest. For example, if we wish the know the effects of training on worker’s performance, the control group would be a group that doesn’t receive the training, while the treatment group would be the group of individuals that undergo training. To make a useful comparison between the treatment and control groups, both groups should be as similar as possible with the treatment being the only difference between the groups. For example, having relatively more experienced people in the group undergoing training could bias the result of training. 1 2 CHAPTER 5. CONTROLLED EXPERIMENTS AND OBSERVATIONAL STUDIES But an important question is how do we ensure that both groups consist of a variety of individuals with neither group dominating the other in terms of some characteristic? A smart person could try assigning individuals to the groups using his/her judgement. Unfortunately, human judgement is bound to introduce some sort of bias. The remedy therefore is by ran- dom sampling, a concept introduced and propagated by the great British statistician Sir R.A Fisher. Random sampling is an innovative idea that ensures the similarity in characteristics of the treatment and control groups. Random can be in- terpreted as “equal chance,” that is, random selection means that each individual has an equal chance of being assigned in either group. For our example, the experimenter could toss a coin, with H = treatment group and T = control group. 5.1 Controlled experiments Although the idea of controlled experiments is appealing, it is rather tricky to set up a proper study. The Salk Vaccine Field Trial is a very good illustration of what a controlled experiment might look like. 1 Polio was an epidemic in the U.S for over forty years from 1916. In the 1950’s several vaccines were developed and the most successful one was the Salk vaccine....
View Full Document
This note was uploaded on 03/14/2010 for the course ECON Statistics taught by Professor Yy during the Spring '10 term at Seoul National.
- Spring '10