By definition, a valid hypothesis should be testable. A hypothesis is an educated guess - an attempt to explain a given observation, or answer a particular problem. A testable hypothesis has two properties. First property is that is should be falsifiable, or counterexamples must possible exist to prove it wrong. Second, reproducible results must be possible in proving it wrong or correct. In short, a testable hypothesis is something that can be tested experimentally, and using the proper methods.
For example: Saying that drug A can cure disease X due to a certain mechanism is a valid hypothesis, as you can construct an experimental method to test it. If it works, you get results. If not, you get negative results. You can also make replicate trials to prove the result you first got. On the other hand, the existence of a god or supreme being cannot be made into a hypothesis as there is no such way (in science) to prove or disprove an existence of such entity.
In an experimental set-up, during which a hypothesis will be tested, there are two groups, usually - an experimental group, and a control group. A control group is a separate group where the independent variable being tested cannot influence the result. Usually, the control group has exactly the same conditions as all the other groups, except for the single variable that is being tested. The experimental group, then, is the group which contains this other factor that is being tested. It is the group which has altered conditions, and is subject to test to prove something, or to show something. For example, in trying to see the effect of drug A on individuals, two groups can be made - one takes the drug , the other doesn't; however, both groups shall eat at the same time, and eat the same food. The experimental group is the one that took the drug. Hence, the drug's effectivity can be seen by comparing, after a certain time, the changes that occur from the experimental group and the control group.