A strong hypothesis is concise, clear, and defines an expected relationship between the dependent and independent variables. This relationship should be stated as explicitly as possible and must be testable.
Having a concise hypothesis makes it obvious to the reader when you transition from background information to a research question without having to state, “This is my research question”. It is important to be as clear as possible; readers should not have to guess what you are trying to test. This is crucial when crafting research proposals, as reviewers need to know exactly what question you hope to answer with the proposed study.
Consider the following examples:
• A weak hypothesis: TV consumption influences sleep.
• A moderate hypothesis: People who watch more TV will experience poorer sleep.
• A strong hypothesis: People who watch more than three hours of TV daily will wake up more frequently during the night than people who watch less than three hours of TV daily.
Having a strong hypothesis is not only important for communicating with others; it also sets up a strong basis for your research. It is important to consider your planned statistical analysis when you start asking your research questions. Reaching the end of data collection and realizing your data is inappropriate to answer your original question is extremely frustrating and demoralizing. This can be avoided by making sure you have the right tools to answer the question (e.g., univariate vs. multivariate statistics; parametric vs. nonparametric data), and that you are collecting data that be used according to the assumptions of your planned tests. For example, if you want to analyze the relationship between TV consumption and sleep quality, you will have to make a few decisions. First, do you want TV consumption to be measured in bins (e.g., 1 hour, 3 hours) or as a continuum (e.g., 321 minutes, 13 minutes)? Is sleep quality going to be measured on a five-point scale, time spent asleep, or number of times the focal individual woke during the night? Each type of datum can tell the reader (or researcher) something, but different tests are required to look for a correlation between these different possibilities. Planning out exactly how you are going to answer your question through statistics will streamline your data collection process and will make statistical analysis much more straightforward when you reach that point in your study.
• Did you start with a how/when/what/where/why question? You cannot form a statement of what will happen if you have not asked a question first.
• Is your hypothesis a statement of an expected result? If you are writing a question, that is not your hypothesis. Make sure the hypothesis is a declaration of an expected outcome. Ideally, this is supported by the background research you have done.
• Is your hypothesis as clear as possible? It is important to be explicit, as the hallmark of a good hypothesis is the potential to refute it as well as support it.
• Are your variables clearly defined? If the expected outcome is well described, it should also include what the dependent and independent variables are.
• Are you going to be able to test your hypothesis? Make sure you are gathering the correct data before you begin so you will be able to use statistical analysis to support or refute your hypothesis