In this video I offer a formal definition of a hypothesis, distinguish it from theories and thesis statements, and offer suggestions on how to write one.
A hypothesis is a specific, testable, falsifiable statement describing a relationship between variables. Annoyingly, each of those words is an important part of the definition. Let’s start at the end, because this is the most common mistake students make in writing hypotheses.
A hypothesis needs to describe a relationship between variables. This means that the sentence you write needs to describe two variables – a proposed cause and an effect – and how you think they relate to each other. So for example, the first hypothesis here isn’t going to work because there are too many things going on. It is unclear which relationship the student wants to test and what the actual relationship is. In the second hypothesis, I have chosen a cause and effect and described how they relate to each other (as oil exports increase, allliances increase). You will be looking at alternate explanations in your research, but as control variables – we are going to try to examine which factor has the most explanatory value in a world where they all matter. For this project you need to choose a specific cause and effect.
This also points to the second key element of a hypothesis – it needs to be specific and testable. You don’t need to define every aspect of your variables, but you do need to be relatively specific. So what does “international standing” or “Iranian politics” actually mean to you? You get to choose how you define your terms, but you do need to start that process here and you need to think of ways you could actually observe your variables. Here, “intervention in madrasas” is still a bit vague, but it starts you on a path to being able to identify how you would actually figure out the government is interfering. I have also chosen a measurable way to evaluate religious practice – mosque attendance. Ultimately, whether you do qualitative or quantitative research, you will still need to be able to define what you actually mean by your variables and how you know they are present or absent.
Finally, a hypothesis needs to be falsifiable. I find it easiest to demonstrate this point by highlighting the difference between a hypothesis and a thesis statement. A thesis statement is a sentence that summarizes an argument. You spend an essay describing the elements of that argument and selecting credible pieces of evidence that support it. A hypothesis needs to have a clear way to be proven wrong. The purpose of your research paper is not to find evidence to support what you think is right, but to identify the potential causes of an effect and test which one has the most support. So in this example, it would be hard to prove the statement “economic development leads to democracy” wrong because there are so many things that statement could mean – different types of development, different types of democracy, etc. The hypothesis “As a country’s GDP per capita increases, its Freedom House score increases” makes it clear what your criteria for evaluating the statement are – you have defined your terms, the relationship between them, and it is possible that this relationship is wrong. There is still a lot more to do to explain the relationship between development and democracy as a whole, but scientific research helps with this by breaking the relationship into parts and evaluating those specific, testable parts.
Another way to think about this is to think about the difference between a research question, a theory, and a hypothesis. A research question identifies a causal relationship you are interested in, such as why people protest. A theory is a proposed answer to that question – people protest because there is a gap between their economic expectations (based on their education) and their economic reality (a lack of jobs). A hypothesis is a way to test that theory. “As the unemployment rate for university graduates in a country increases, the number of political protests increases.” This is not the only way to look at the relationship between education, economic opportunity, and protest, but it is more specific and defined – it is one that you can actually test. That is what we are going for here.
You will note that all the hypotheses I have suggested follow a certain format. As one variable changes, another variable changes. This is the format I recommend you use because it forces you to be clear what your variables are, and what the relationship between the variables is. We often think of hypotheses as “if then” statements, and that is fine if you want to use this format. The only reason I don’t recommend it is because it tends to require binary variables (if means something either happens or it doesn’t) and it tends to imply a deterministic relationship. If one thing happens then another thing always happens. That can be true for biology or chemistry, but it is almost never true in political science.
The last thing I will comment on is that in actual political science, you will rarely see hypotheses written this way. Even when authors are clear about what their hypotheses are, they often write it as more of a thesis statement. I encourage you to write “as as” statements not because they are the only way to write hypotheses, but because they are super-clear. They push you to think through variables and relationships, but obviously are not the only way to describe a relationship between variables. I encourage you to learn the formal rules and structure first, then start relaxing those rules. And FYI, the only actual hypothesis here is the one labeled H1 (in the article by Donno).
That’s it on writing hypotheses – in class we will review this, evaluate some sample hypotheses, and then practice writing them.