In this type of testing, you must begin by stating a null hypothesis before calculating the likelihood the statement is true. You will need to test the calculation of likelihood multiple times to ensure your observations are not due to pure chance but rather to a real difference and your statement can only be tested for rejection or failure of rejection. It can be figured out by calculating the simple mean of odds, a sample variance or correlation. An example of a null hypothesis would be "a tomato plant does not grow any faster when expose to the sun."
To test a cause and effect hypothesis, you need a different type of statement. Most often, this type of hypothesis will go as follows: "When I do ABC, then XYZ happens." When testing a cause versus an effect, only one variable at a time should be changed. For example if your hypothesis might state "When I add fertilizer to a plant it will grow bigger." To test this hypotheses, you should only change the cause, which is the fertilizer, and test by using various products, or none at all and comparing the results.
In some cases, it is not possible to show how changing the cause of the hypothesis will produce a specific effect. However, you can demonstrate the link between cause and effect by testing its correlation. In this case, you may not be able to explain how a cause produces a specific effect, but you will have gathered sufficient data to demonstrate how two variables may be related. In the example of the plant, exposure to light, quantity of water and temperature of the room also all impact growth, which means that only correlation can be determined if all of these other factors are not being taken into account.
To make sure you are testing your hypothesis correctly and to know if you are considering all variables, you need to first have a control group that you can then compare to your trials. The control group simply tests the hypothesis as is with no changes in the cause. When you start changing variables for your experiment you can either use a specific control group as part of the experiment, such as is the case when participants in medical research are given placebo medication without knowing it while others are given a new drug. Another way of testing variables against the control group is to test different values for the cause at the same time. For example, a fitness research might compare a group of sedentary people, to one in which participants try new program three times a week, and a third group may have to try the program five times a week.