Normative and Empirical Questions
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
The first step towards knowing how to do good quantitative analysis is to know what quantitative analysis is good for. In this video you will learn to:
- Distinguish normative and empirical questions.
- Explain why we’ll focus on empirical questions.
Why care about this distinction?
Some questions ask what should be happening in the world. These are normative questions. Other questions ask what is happening in the world. These are empirical questions. When we are doing quantitative social science, we are concerned with the second type: empirical questions. You are going to learn why.
Normative questions
When we pose normative questions, we are posing questions that necessarily involve value judgements. Some examples:
- Should income inequality be reduced?
- Is compulsory voting desirable?
- What is a fair tax system?
These are all normative questions. They are the questions that political theorists and political philosophers ask and do research on, and important research for that matter. But as important as these questions are, they cannot be answered by data alone. They fall outside the scope of quantitative analysis. The data cannot fully address those questions head on.
Empirical questions
What we can do with data is ask about observable patterns and relationships, tackling questions that are going to help us answer those normative questions. Each of the three questions below is related to one of the normative questions above. They inform those normative debates, but they do not answer the normative questions.
- How unequal is income across regions?
- Does compulsory voting increase turnout?
- How do tax changes affect employment?
These are all questions that statistics can help us answer. They are also crucial questions that are instrumental in our search for answers to normative questions. Because of that, we are going to bracket normative questions in this video series.
Good empirical questions
So we have empirical questions, but empirical questions can be good or bad. For a good empirical question, we want the question to meet additional criteria. We want to avoid factual questions that have a simple answer, and we also want to avoid questions that we can’t answer with data, such as hypothetical questions about imagined scenarios. We want questions that are challenging but not impossible. There is a middle ground in how much of a challenge a question should be if we want to ask the types of questions that empirical researchers ask.
Avoiding factual questions
Some questions are very important, but they are too easy to answer. They can be answered with a quick Google search, which some people would call research, but it is not actually research. For example:
- Which UK regions voted most heavily for Brexit?
- Which party governed the US from 1981 to 1992?
You don’t really need research to find out which UK regions voted most heavily for Brexit, or which party governed the United States from 1981 to 1992. You need a quick Google search. And if you can find the answer with a quick Google search, then you have to think of a different question. We need a real challenge if we want to do empirical research.
Focusing on observables
We need a real challenge, but we must not deal with the impossible. We must remain committed to answering questions that we can answer with data.
For example, someone might wonder: if Bernie Sanders had beaten Hillary Clinton in the 2016 Democratic primary, would he have beaten Trump in the general election? We are never going to know the answer. It is just impossible. Someone chatting in a pub might come up with an answer to this, but serious researchers shouldn’t really go there.
Similarly, consider asking whether something happened in the past when whatever happened didn’t leave any trace. Someone could ask: did ancient Chinese explorers visit the Americas without leaving any trace of their presence? If there is no evidence, no trace that this historical event happened, then it is completely outside the scope of empirical questions. Suppose someone were to state that yes, those ancient Chinese navigators did go to the Americas, but they didn’t leave any record, any trace, no artefact that can be excavated, nothing. We are never going to be able to verify whether a statement like that is correct.
So we must save our efforts for the realm of possibility.
Recap
- Normative questions ask what should be happening in the world.
- Empirical questions ask what is happening.
- Our exclusion of normative questions is methodological, not evaluative. As researchers who work with statistics and data, we bracket normative questions not because we think they are less important. They are very important questions, but tackling them requires thought and judgements beyond what statistics and data alone can answer.
Why this matters
A research method is never an end in itself. A research method is a means to an end: a procedure that we use in order to answer a research question. And choosing an appropriate method for a research question is a core, foundational analytical skill.