A hypothesis statement predicts a relationship between two variables. Writing a hypothesis should always precede any actual experiments and is an important part of the scientific method. Remember that the scientific method is used not only in the physical sciences, but also in the social sciences. So you could write a hypothesis on anything from how sunlight affects plant growth or cell multiplication to how the amount of income taxes paid impacts one's perception of the president. A good hypothesis statement makes clear the relationship between the variables and is always testable.

Step 1

Choose a topic that is of interest to you. Observe the world and identify things you feel are worthy of studying more in-depth. Perhaps you've noticed that fewer people seem to be divorcing today than 10 years ago, and you'd like to understand why.

Step 2

Formulate a prediction to help explain the issue you have chosen. You might say that people are divorcing less today because they are choosing to marry later in life. Identify your dependent and independent variables. In this case, divorce is the dependent variable, and the age at which one marries is the independent variable.

Step 3

Indicate how exactly one thing might affect the other. Many hypotheses are unacceptable because the writer fails to articulate the direction he expects a relationship to take. Instead of saying "A person's age when married affects likelihood to divorce," say "The older a person is when he marries, the less likely he is to divorce."

Step 4

Make sure that your hypothesis is testable. You must establish your variables. By then examining the ages at which people married and whether they have divorced, you can control against other variables, such as education, and see whether a correlation exists. Social science researchers may write a null hypothesis, which is the opposite of what they believe will happen. They use the null hypothesis to test a hypothesis. In the divorce example, the null hypothesis would be that the age at which someone gets married has no effect on the likelihood to divorce. If the data show that age when married does impact divorce rates, you can reject the null hypothesis that age makes no difference, and thereby accept the initial hypothesis (called the alternative hypothesis).