This tutorial talks about observational studies. Observational studies are one of two main ways of conducting a statistical study and obtaining data. The other way is an experiment.
In an observational study, the researchers are observing and recording the variables. They're careful not to try to influence the subjects. They also can't control the variables. You can't assign a treatment. You can't determine exactly what's going to happen. You can only watch. Because of this, you can't determine cause and effect. Typically, an experiment is what we would use to determine cause and effect.
Now, we'd use an observational study when an experiment isn't possible, whether because of ethical concerns or practical concerns like time or cost. Now, in an observational study, we have some vocab words that we need to use. The participants are the people that are involved in it, the people that we are studying. Subjects are the animals or things that we're studying. Sometimes people use the word subjects to refer to people as well, so just be aware of this.
There are two main types of observational studies, and we'll look at both of them-- longitudinal studies and case control studies. One type of observational study is a longitudinal study. Now, a longitudinal study occurs over a long period of time. It observes the same set of people and follows the same variables over that chunk of time. It can be as many as several decades. While the study isn't very quick to do, it provides a lot of data, and many different researchers can use this information in a variety of ways.
Now, the studies are typically prospective. That means that it follows a particular group over a set of time. It can also be retrospective. It can take information from the past. For example, it can find a group of people and then examine their old medical records to try and find some patterns.
One example of a longitudinal study is the Framingham Heart Study. It started in 1948 and is still going on today. 5,209 healthy adults from Framingham enrolled in this study. Researchers collected a variety of information about them, including social networks, eating habits, exercise habits, and several markers for heart health. Over 1,000 different papers have been written using this information. Some of these papers have proven that obesity and smoking increase the risk of heart failure. Other papers look at how the social networks tied to obesity risks.
A second example looks at a group of middle-aged truck drivers. These truck drivers have a variety of smoking habits. Some don't smoke, some smoke a little, and some smoke a lot. If you study them over time, you'd expect that the smokers who smoked the most would have the highest rates of lung cancer. We can use the information that we collect to help determine whether or not this is an accurate hypothesis.
Another kind of observational study is a case control study. A case control study is also called retrospective study. In this kind of study, you take a pair of participants. They're both very similar across most variables, except for one major difference. One has a disease, the case, and the other does not-- the control. Because the participants are so similar, you're essentially focusing in on just that disease and seeing how it affects them or seeing what caused the disease.
Now, it's retrospective because it looks in the past. You ask the participants to recall past events or you used information about their past to kind of determine what risk factors there are for the disease. It's similar to a matched pair design in an experiment, but in this case, the researchers are not giving a treatment. They're not doing anything to affect the people. They're only observing and reflecting on people's pasts.
Let's look at an example of how this would look. In this first example, the case group is a group of children age 3 through 14 that have asthma. The control group is a similar group of children who do not have asthma. Now, the researchers were trying to examine whether or not exposure to smoking through passive means-- so the children themselves aren't smoking, just whether or not people around them were-- had an effect on asthma rates.
Now, the data comes from a questionnaire. The children are probably too young to accurately remember, so the parents and other people around the children were used as well. Now, from the questionnaire and whether or not the child had asthma now or not, researchers were able to find that the highest rate of asthma came from children whose maternal caregivers-- like moms, grandmothers-- were smokers.
This has been your tutorial on observational studies. We covered two major kinds of observational studies-- longitudinal studies and case control studies.
This tutorial looks at experiments, one of the two main ways of conducting a statistical study, the other way being an observational study. Now in an experiment, the researchers are manipulating one or many variables while holding all the others constant. The researchers are careful not to influence the subjects involved.
However, because researchers can control the variables, we're able to determine cause and effect. Observational studies cannot do this. So we try to use experiments whenever possible. However, sometimes, due to ethics or practical concerns like time and cost, it's not possible to run an experiment. That's when an observational study would come in handy.
A participant is a person that's in the experiment. A subject or experimental unit is an animal or a thing that's in the experiment. However, some researchers use the word experimental unit or subject to refer to people in experiments as well. So just be aware of this usage. Here are two examples of experiments.
I'll read the first one. Patients with Alzheimer's are divided into two groups. One group is given a new drug, the other the old drug. In this example, the participants are the patients with Alzheimer's. We could also call these experimental units. Remember that people are typically referred to as participants but could be referred to as experimental units or subjects.
The researchers are trying to determine cause and effect. They're trying to determine whether or not the new drug affects the patients, whether or not it slows the progression of Alzheimer's or even reverses the effects.
Now in our second example, rats were divided into two groups, each with an even mix of males and females. One group was given large doses of aspartame. The other was given large doses of sugar. In this example, the rats, those are our experimental units. They could also be called subjects, but we wouldn't refer to them as participants.
In this example, researchers are still trying to determine cause and effect. They're trying to determine whether or not aspartame causes a reaction in rats, whether or not it causes death or tumors or any other host of side effects. These two examples can give us a bit more insight into how vocabulary is used to describe experiments. Other examples of experiments will be given in later tutorials.