Some fun facts about statistical studies
Before starting my summary, I must admit this semester is much more busy than I expected, so I have to postpone my first blog one week later than my plan. I will try to write the next one on time and make things up.
I would like to introduce my study on statistics for this semester. One of my statistic course is called Method of Data Analysis II, this class introduces many models include logistic regression, loglinear models, longitudinal data and so on. So far we haven’t cover too much of these models, but I have found something very interesting and want to share with my readers.
Before taking this course, I always wonder about what are the most common type of statistic study? In fact, the researchers often use experiment (also called randomized trial) to test if there are any causal relationships between the treatments and results. One of the characteristics of this type of study is the researchers are controlled the conditions, they are responsible for keep tracking the effect on the observations from the treatment studied or other variables. Experiment is commonly used in the medical, psychology and pharmacy studies. For instance, Stanford Prison Experiment is one of the most famous experiment in psychology history. The study is designed to test whether the contrived environment will cause influences to people’s personalities. The procedures of the study is 24 students were randomly assigned the roles as prisoners or guards, they were put in a simulated prison for two weeks. From this experiment, we should aware of two most important keys of experimental study, the first thing is the participants must be randomly assigned to the control groups, besides, the control groups must be very similar to each other. The differences should only caused by the randomness.
Another commonly used study is observational study. Just like its name, in this type of study, the researchers are “watching” the participants who are measured and recorded without any control to the conditions. The observation are divided in two types based on the status. The first one is called case-control study where the researchers are analyzing based on the outcomes. Oppositely, the other type is based on the exposure status of the observation, prospective study is the researchers involved before there are any results, retrospective study is the researchers getting relevant when the outcome is already showed.
In an observational study, participants are not assigned as treatment or control group anymore. But some of them should have the features that we want to study. For example, in the study relation of obesity with heart diseases, overweighted people would be the treatments and other people would be the controls.
It’s worth to mention that we could only conclude about associations between the variables and outcomes. The causality from the observational study could be misleading sometimes, because some impacts of factors could mixed up and we would not know which one has the real impact, these factors are called confounding variables.
I realize that known and being familiar with these knowledge is very important to my statistics studies, because as a statistician, we will face many research topics, design a study is the very first thing to do when we get the relevant data. We should always be prepared.