Designing an experiment
So you have made it through the first steps of the scientific method, and you have a hypothesis to test. Now you need to design an experiment to test it. This is the most difficult part.
First, identify the dependent variable, which will typically be the variable you are measuring. Then, identify the independent variable, which is the variable you will be in control of and will change. Remember, the dependent variable can be a yes/no variable, such as weather the patient has lived or died. Or it can be a variable that changes in value, such as the height the model rocket flies to. All other variables must not be allowed to change. In other words, they need to be held constant. Hence, they are called constants.
Let's look at two examples:
Testing a new drug
The Megadrug company has discovered a potential cure for arthritis in the bark of a tree in China. It wishes to test the drug's effectiveness for possible FDA approval. To do this the company asks for volunteers for the test. Each patient must have a specific type of arthritis and no other disease of the joints. The patients are split into 4 groups. One group is given a sugar pill, while each member of the other groups are given some of the new drug according to the following table.
| Test group | Amount of drug/kg of body mass |
| 1 | 1 |
| 2 | 2 |
| 3 | 3 |
| 4 | 0 (placebo control group) |
The independent variable, or the variable being changed, is the amount of the drug being given.
The dependent variable, or the variable being measured, is whether the patient is cured. This is a yes/no variable, although you could just as easily measure a variable such how tightly the arthritis patient can squeeze a testing device. This shows one big difference between the dependent and independent variable: there is only one independent variable, but there can be a number of difference dependent variables that you might be measuring. And some that you do not.
In this typical drug testing procedure, group 4 is your control group. The sugar pill (placebo) does not cure the arthritis, but does show how many patients might fake a cure, or pretend to be cured, or have the disease just naturally cure itself. It is the control group, because the independent variable (the amount of the drug given to the patients) is unchanged from normal.
Constants would be the fact that you want to make sure the patients have no other diseases of the joints. You would also want to make sure they are healthy, and that they are taking no other medicine for their disease. These variable are not allowed to change, and hence are constants.
An experiment to see how the amount of sunlight changes the speed of plant growth
This is a typical classroom experiment. To do this, you might plant 7 seeds to try to determine if controlling the time in the sunlight will change the speed at which a plant grows. You would want to get 7 seeds, place them in 7 cups, all with the same amount of soil (and the same type of soil). You would give each the same amount of water, at the same time during the day. These variable are all held constant. One plant, the control, you would want to leave on the window sill. The other plants you might cover with paper cones to reduce the amount of time each gets sun. You might cover plant 1 for 1 hour, plant 2 for 2 hours, and so on. Thus, the value you are changing is the amount of time the plant gets sun, and this is the independent variable. Each day you might measure the height of the plant. This is the measured value, and is the dependent variable. You might also decide to count the number of leaves, which could be another dependent variable. But you are only changing one variable. Good design requires that you change only one variable.