This tutorial will explore observational studies and how they are conducted. Observational studies are a little different than experiments, so you will learn about:
An observational study is a type of study where the researcher can observe but does not administer any treatment. So whatever would normally happen, the researcher has to allow it to happen.
They can't change anything about the people or whatever they're studying. And the researcher can record the variables of interest, but again, can't affect the study. People have to be allowed to do whatever it is they were going to do without interruption.
There are two types of observational studies:
a. Retrospective Study (case-control)
You take people who are sick-- those are called the cases-- versus people that aren't sick-- which are the controls. Then you look back to see what similarities the cases have in common and what similarities the controls have in common.
b. Prospective Study (longitudinal)
Individuals are doing something like smoking or jogging, and they participate. You record what happens as it happens as opposed to trying to look back and trying to figure it out.
The year is 1929 and a cancer doctor has a suspicion that smoking may cause cancer. His cancer patients become his subjects, or "participants" in his study. He asks his subjects, “hey, did you guys happen to smoke before you got cancer?” What he found was an overwhelming majority of his cancer patients did in fact, smoke. So this doctor was the very first person to suggest a link between smoking and cancer.
That inspired some new studies, one of which began in 1934. It dealt with a lot of physicians, so it was a physician’s smoking study. Now notice in the image below, some of these physicians smoked, and some of them didn't.
They took several thousand doctors and conducted this study over the course of 20 years. They found that not every doctor who smoked ended up getting cancer, and not every person who got cancer was a smoker.
But what they did find was that the doctors who smoked ended up getting cancer. It became the most convincing evidence that smoking had an effect on cancer. This was prospective study, because it started with the doctors in 1934 and it followed them through to 1954.
Neither of these types of studies, prospective or a retrospective, can actually prove a cause and effect relationship. The only thing that can prove a cause and effect relationship between two variables is an experiment.
Think about what an experiment involving smoking would look like. That would require assigning certain people to smoke and assigning certain people not to smoke. And that doesn't make any sense.
It's unethical to do it that way because we know how unhealthy smoking can be. Forcing someone to do something that's unhealthy doesn't make sense, not to mention, it’s highly unethical.
But because the link was so strong, and because the link was so consistent over several studies that were done, in 1964 the Surgeon General of the United States released its famous warning about smoking and lung cancer and variations of the warning below appears on every package of cigarettes.
In an observational study, a researcher can observe but not influence the behavior of the participants (called “subjects”). There are types of observational studies from which to choose: you can do a retrospective study, which looks back. Or you can do a prospective study, where you gather your participants and then follow them along as they live their lives.
You can't draw cause and effect relationships via observational studies; only experiments can lead to a cause and effect conclusion. However, sometimes observational studies can provide some pretty convincing evidence.
Source: This work is adapted from Sophia author Jonathan Osters.