An observation could be in the form of an established law of science. It could be in the form of observable phenomena. Let me explain.
A law of science is just an observation. Observations generate questions. This is very important. The observation will determine the types of hypotheses that that are generated.
(Not all general observations eventually become laws.)
Ask questions about your observation. State in the form of a question what it is that you want to know. Ask questions such as: how, what, when, who, which, why, or where?
Anything that can have any bearing on the problem is considered. Do background research. Find out as much as you can. Accumulate any relevant data.
Predict what the answer to my question will be. A hypothesis is like a trial answer or an educated guess as to what the answer to your problem or question could be. A hypothesis must be falsifiable.
What should I find if my hypothesis is correct?
Theories generate predictions, but there has to be a way of not only verifying the prediction but also refuting it. If the prediction cannot be refuted it is not scientific.
Design tests in order to disprove my hypothesis.
You must always attempt to falsify the hypothesis, not to confirm it.* The hypothesis will become confirmed through the falsification process. *This is a very important point.
All we can say about a hypothesis, which stands up to a test to falsify it is that we failed to disprove it.
Did your hypothesis hold up under every test? If it did, this helps to establish a level of probability. Draw conclusions as to whether the hypothesis is supported or disproved. If the data failed to confirm your initial hypothesis, it's time to go back and come up with a new hypothesis and test it.
Report your conclusions.
Communicate your results to others who have an interest in the topic. Publish your findings if possible. Include a statement that accepts or rejects the hypothesis. Tested hypotheses are used in formulating generalized theories.
Much of the clip art on this site is courtesy of Phillip Martin.