Common Experimental Designs

In designing an experiment, in many cases it is not effective to simply test a variable on one group of subjects and note the changes that take place. Often it is more meaningful to have a comparison of two or more groups of subjects that are treated differently. Following are some common designs for this kind of experiment.

Controlled

In a controlled study, at least two groups of subjects are used. The "test" group receives the intervention of the variable, and the "control" group does not. Both groups should be treated alike in all other ways, as much as possible.

For example, we could test the effects of feeding Moringa tree leaves to cattle. If we have twelve cattle, they could be divided into two groups of six. One group would receive their normal feed. The other group would receive the normal feed with Moringa leaves added. Both groups should receive the same amounts of water, pasture, exercise, rest, and so on. The group receiving the Moringa leaves is the "test" group, and the group receiving only the normal feed is the "control" group. Measurements of height, weight, and girth could be taken every day, and then the rates of growth compared between the two groups.

Randomized

A randomized study is simply a controlled study where the subjects in the test and control groups are selected randomly. This is one step that can be taken to avoid any bias on the part of the researcher from affecting the experiment.

When a controlled study is not randomized, it is easy for the researcher's bias to affect the outcome. For example, if we simply hand-pick the cattle for each group, we may unconsciously select our favorite, or the healthiest, cattle for the test group. This unconscious bias would affect the results of the study. So a technique for random selection is used instead.

There are many techniques that can be used to randomly select the groups, such as:

• Drawing straws
• Tossing dice
• Selecting every other (or every third, every fourth, etc.) subject to be in a different group
• Pulling numbers or names out of a "hat"

No matter what technique is used, the important point is that the selection is random.

Experiments using plants or animals as subjects often use the randomized controlled study design. Here is a diagram of a randomized controlled study:

Double-blind

A double-blind study takes even further precautions to keep personal bias from influencing the results. This design is most often used in studies with human subjects.

Studies with human subjects can be influenced by the "placebo effect." Sometimes, simply telling human subjects that they are receiving a new medicine can be enough to create improved effects in those subjects-even if they are really just receiving a sugar pill, or placebo.

A "single-blind" study attempts to avoid the placebo effect by not telling the subjects whether they are receiving the medicine being studied or not. They are kept "blind." For example, the test group would receive a pill of the medication being tested, and the control group would receive an identical pill, but it would only contain sugar or some other edible substance (a placebo).

However, even in a single-blind study, the researcher may treat the subjects differently or give subtle, unconscious clues if they know that certain subjects are receiving the medicine being studied. The researcher's behavior toward the subjects can influence the outcome.

In a "double-blind" study, neither the subjects nor the person administering the intervention know who is in the test group or the control group until after the experiment is finished. This would involve the subjects being selected for different groups and "pills" prescribed to each group by a person other than the physician who actually gives the pills to the subjects. Since the physician does not know who is getting the real medicine, he or she will not treat the subjects differently.

Here is a simple diagram of a double-blind controlled study: