A strong hypothesis should meet three fundamental criteria. It needs to state the hypothesis in the proper conditional phrasing. It needs to clearly establish the relationship between the variables included. Finally, it needs to establish that said relationship is scientifically provable.
Before anything else, the student should make certain that the hypothesis being tested is theoretically and contextually relevant to the assignment at hand. In other words, it should relate to the topic. Once the variables have been chosen, the hypothesis itself should suggest direct causality existing between them. For example, "If temperature affects leaf color, then lowering the temperature will affect that color." The emphasis here is on the "if ... then ... will" construction. It is important to remember that a hypothesis is a statement, not a question, and it is also necessary to avoid hedge words such as may or might in its formulation.
Once a viably stated hypothesis has been obtained, the ability to prove its validity should be testable through empirical methods. In short, this means collectible data. For instance, if the hypothesis relates to the color of leaves and temperature, subsequent testing should create an environment where actual leaf samples are subjected to controlled atmospheric fluctuations. If a hypothesis deals with voters' reactions to certain policy decisions, perhaps the tester could conduct surveys intended to monitor shifts in opinion or voting patterns on specific issues. Most importantly, the tests the original student runs on her hypothesis should be capable of being replicated by others if and when they choose. Others should be able to create the same testable conditions with the same materials and achieve precisely the same results.