Experimental research (the design of)
An experimental research design is a design in which the data of an experimental and a control group are compared to each other.
What makes an experimental research design so special?
This type of design is highly valued by scientists. It compares the data from an experimental and a control group before and after an experimental happening occurred. This design is depicted in this graphical model:
It simply says: this research is done by collecting data at two moments. The first moment is before an experimental happening occurred and the second one is after this occurrence. Both groups are carefully selected. Though the term group is used, nothing has been said about the size of the group; the minimum could be even 1 so it is even applicable on qualitative research or on comparing two case studies.
Examples of research questions for research designs
Some research questions might be:
- Do people who take this medicine live longer than those who do not take get the medicine?
- Do we get less complaints since we reorganised the helpdesk compared to a company who did not reorganised their helpdesk?
- Do students who attended a course in statistics improve their knowledge more about statistics than students who didn’t attend the course?
The analyses of a research design
The analyses that have to be performed are clear too. First of all, the members in the groups should not be different from one another other. This has to be tested and no (statistically significant) difference should appear. Then on t1 the topic of the research is measured. Again no (statistical significant) difference should appear. Finally at t2 the topic of the research is measured again. This time a (statistically significant) difference should appear. Now the conclusion can be drawn that a change occurred due to the experiment.
Though it is likely, it isn’t necessary a (statistically significant) difference appeared between t1 and t2. One group might decline and the other one incline. On the average nothing has changed. Even the scores of both groups can decrease or increase equally. However, the change in the experimental group should be bigger than in the control group. (Check this carefully in your data, because a statistically significant outcome is sometimes the opposite of what was expected.)
Above a lot of tests can be applied. Actually the effect should be tested in one analysis: a RM-ANOVA.
Using a placebo
It is not ethical to withhold people medicine when they are ill. But if you give them the same treatment too, the effect of the treatment cannot be tested. Then the control group should get the treatment as usual. In the design this can be stated as:
The effect of the new medicine now is tested against the usual treatment.
Sometimes the medicine is expected to work, just because people know they are selected for a special treatment. Now the effect of the treatment is mixed up with the expectations of the members. To unravel this clutching, a placebo is used. A placebo is giving the member of a group a similar treatment but not the one of which an effect is expected. For instance, the effect of a briefing film about the negative impacts of pollution, can be replaced by a briefing film about using correct English. In the design this can be stated as:
On this page the most common experimental research designs have been handled. A few more can be distinguished. If you want to do experimental research, you may need a special design for your experiment to exclude alternative explications. It is always possible to make one.
For reasons of preventing complexity in terminology, only the term experimental group and control group have been used. They can be replaced with experimental condition and control condition. These terms are very common as well.
Related topics to experimental research
- Explorative research
- Descriptive research
- Comparative research
- Evaluation research
- Longitudinal research