Module 1: Producing Data
3. Sources of data > 3.2 Generating your own data > 3.2.3. Generating data from experimentation
iv. Controlling confounding variables
You can try to control confounding variables by following two steps.
STEP 1:
Set up 2 equivalent groups of units/participants. This can be achieved by using randomisation -i.e. by randomly assigning patients to each group.
STEP 2:
Give a treatment to only one group and keep all other variables controlled or the same between the groups. The group that does not receive treatment is called the control group or control.
The second step requires you to work very hard to ensure that the treatment variable is the only variable that changes between the groups and that all other sources of variation between the two groups are controlled.
These steps form the basis of controlled comparative experiments. Randomisation and control are designed to remove confounding variables. By using randomisation any experimental unit has an equal chance of being assigned to any treatment. Randomisation can also be applied to the order of treatments and the type of treatment applied to experimental units.
Control is the situation or group where the treatment has not been administered. The only difference between the control group and the treatment group is that the treatment group has been given the treatment.
Replication, i.e. conducting the experiment more than once, is designed to account for any variation between sample groups. Consistent results indicate whether this has been achieved.
Comparative studies can be observational studies or studies based on available data, but it becomes more difficult to develop equivalent groups.
| 
