Constructing a hypothesis is an integral and required step when utilizing the scientific method to test a research query. The hypothesis is formulated by the researcher and it is a prediction of what the researcher believes the results will be. Since it is a prediction, the hypothesis does not need to be accurate.
When constructing a hypothesis, the first step is making sure that the hypothesis is testable. Having a hypothesis which is not testable will be hard to work with. For example, asking whether Creationism is true is not testable.
Secondly, another important concept to understand is falsifiable hypotheses. A good a hypothesis is one which can be falsified. Basically a hypothesis is considered to be falsifiable if there is a reasonable or logically sound observation which contradicts the hypothesis.
For example: Pretend you are researching the concept of predation and food supply. The species of interest are foxes (predator) and rabbits (prey). You increase the food availability for rabbits by 4X in one pen and do not alter the food availability for rabbits in the other pen (trying to keep normal conditions). You want to know whether a higher food supply will allow the rabbits to be more successful at avoiding predation by the foxes.
Hypothesis: Changing the food availability for the rabbits will result in no change.
When you observe the statistics from the experiment, the results are inconsistent with this hypothesis. Thus, the hypothesis has been falsified and needs to be revised.
Strong hypotheses need to be falsifiable because in science you can never prove that anything is true. Rather, you can only illustrate that a hypothesis is incorrect. Thus, good research is comprised of hypotheses which are both falsifiable and testable.