Units of Analysis, Theories, and Hypotheses
The text below is a transcript of the video above, edited with AI assistance for clarity and flow. Where the two differ, the video is the authoritative version.
You’ll learn to
Units of analysis, theories and hypotheses are key concepts in empirical research. In this video you will learn to:
- Identify the unit of analysis of a research question.
- Elaborate theories and hypotheses for research questions.
Why care about these concepts?
Knowing our unit of analysis gives us a foundation by establishing the who or the what of our study. Theories and hypotheses give us direction: they tell us what the next steps are when designing a study and analysing the data.
The unit of analysis
To understand the unit of analysis, let’s look at three research questions. They are all asking about similar relationships.
- Are countries with higher literacy rates less likely to experience democratic backsliding?
- Are cities with more universities less likely to vote for authoritarian candidates?
- Are people with college degrees more likely to support democratic norms?
The second question asks whether those cities are less likely to support candidates who will drive democratic backsliding forward. The third looks at individuals, the types of people who, as citizens, will uphold democratic norms and help prevent democratic backsliding: when we look at people’s education levels, does that correlate with their commitment to democratic norms?
These are all questions that pertain to the same topic, to the same research agenda. But you might have noticed already that one thing is critically different across the three questions. Question one is asking about countries, question two is asking about cities, and question three is asking about people. Country, city, and person are the units of analysis of these research questions.
Determining the unit of analysis means knowing the who and the what of our study, and having that clear in your mind is very important when you are designing a study. It is very common for this key concept to get lost. When you don’t know precisely what your unit of analysis is, you are going to be utterly lost whenever you think about what data to collect and how to design the study.
Theories
Once we know our unit of analysis and our research question, we need to formulate a theory. In the context of empirical research, we use the King, Keohane, and Verba definition: a theory is “a reasoned and precise speculation about the answer to a research question” (King, Keohane, and Verba 1994:3). Based on your knowledge and reasoning, what’s a plausible answer to the research question that you are asking?
Example: Health care systems and policy responses
Suppose we want to study the relationship between health care systems and policy responses, and we ask the following.
Research question: How did countries’ health care systems affect their abilities to mitigate the impact of the Covid-19 pandemic?
Suppose that when we approach this research question we don’t know the answer to it yet. We still have to collect the data to find out. But before we do, we should come up with a tentative, speculative answer to that question, based on our knowledge and reasoning.
There is not going to be one single theory for each research question. There will be multiple theories that can answer a research question, multiple ways of using our knowledge and our logical reasoning to come up with a theory. One of the possible theories for this research question would be the following.
Theory: Countries with public health care systems took advantage of their centralised bureaucracies to implement more effective policy responses.
This theory is reasoned and precise, but it is not necessarily correct. Theories won’t always be correct, and they don’t have to be. They have to be reasoned and precise, so that they guide the next steps towards finding an answer to our research question.
Hypotheses
It is easy to mistake a theory for a hypothesis, or a hypothesis for a theory. What’s different is that a hypothesis is an empirically testable implication of a theory. When we formulate a hypothesis, we come up with a statement that describes what we should observe in the world given the theory. If this is the theory, then what is it that we should observe?
In the health care example, we have the research question and we have the theory. A hypothesis follows logically from the theory, and is a statement about what we should observe if the theory is correct.
Hypothesis: Countries with public health care systems had lower Covid death rates than countries with private health care systems.
Clarification
The goal of research is not to prove that our hypotheses are correct. If I were a researcher who normatively believes that public health care systems are better in general, and that countries should have public health care systems, it would be tempting for me to go out, collect my data, conduct my analysis and test my hypothesis in a biased way, looking for confirming evidence of my hypothesis. But this is not how research should work.
The goal is to answer the research question as accurately as we can, whether or not our hypotheses are correct. The attitude of trying to prove one’s point is very common. We see it a lot, and it is a common mistake at the root of bad social science. So it is very important to learn early on that you do not want to approach a quantitative study with the mindset of trying to prove that your hypothesis is correct.
Recap
- The unit of analysis is the “who” or “what” of your study.
- A theory is a plausible speculation about the answer to a research question.
- A hypothesis states what we expect to observe given our theory.
Why this matters
The unit of analysis grounds us: it is the anchor that establishes who or what we are studying. If we are not clear about this, we won’t be able to carry on with our study. Theories and hypotheses give us a sense of direction: they tell us what the next steps are in designing our study, collecting the data and analysing the evidence.