Posttest-only Control Group Design

This diagram is interpreted as follows:

The "Population" is ALL persons to whom this treatment might possibly be applied. For example, all adult-onset diabetic patients in the world.
The arrows and the letter "R" to "Sample" indicates that a sample is randomly selected (Remember that discussion from the previous page?) is drawn from the pool of all potential patients.
The arrows and the next letters "R" indicate that the "Sample" is randomly assigned (ditto) to either experimental or control group.
The Arrow to the letter "E" to the letter "O" is the experimental group. The "E" represents the experimental treatment, and the "O" represents the observation (measurement) of the effect of the treatment.
The other branch, with a "C" instead of an "E", represents the control group. (You can actually subject them to an alternative treatment (such as current therapy), or you can simply do nothing to them. Either way, they are the control group for your hot new idea.) It is extremely important, however, that the only difference in the experimental group and the control group is the differential treatment. They must be identical (statistically) in every other way that might be expected to affect the outcome.
Sources of Error
We shall turn to a discussion of how to analyze the results (statistics) in a moment, but I first want to harp a bit more on extraneous factors. Just a bit, I promise, but please read it.
Let's say, for example, that you actually are using this design, the Posttest-only Control Group Design, to examine the effectiveness of a new approach to educating adult-onset diabetics about the importance of controlling their blood sugar. You use this new approach with the experimental group, and you continue to do what you have always done with the control group.
Is your experiment "clean"? Maybe, but maybe not. Did you (or your colleagues) treat the control group differently than usual just because of knowledge that they were controls and that there was an experiment in progress?
This is just one example of subtle biases that can creep in if you are not careful.
I promised - that's all. Let's turn to statistics.
Statistics
Analyzing data from the Posttest-only Control Group Design is a piece of cake, at least as long as you conform to certain standards. You have one measurement (e.g., average (mean) in-office blood sugar level over a six-month period for experimental and control groups), and you simply compare the measurement from the experimental group with the measurement from the control group. Of course, there will be some difference, so you want to know if the difference is large enough to start getting excited. This is where statistical tests come in.
I have prepared a graphic to help you understand the t-test intuitively before we get into a more technical discussion. Click here to see this graphic and read the discussion.
Return to the discussion of experimental designs.