Research questions · Lesson 2 of 3

Narrowing: who, what, compared with what, when

Narrow by naming who, what, against what, and when — then check that the result still matters.

8 min read · Free — no account needed to read

Most broad questions become answerable by adding four things, and you can do it mechanically.

Who or what — the population or unit. "People" is not a population. "University students aged 18 to 24 in Bangladesh" is. The unit might be a person, an organisation, a document, a country, a transaction.

What exactly — the thing you are claiming changes or differs. "Addiction" is a construct, not a measurement. "Self-reported hours of use" and "score on a validated problematic-use scale" are measurements, and they are not the same thing.

Compared with what — most interesting questions are comparative even when they do not sound it. Compared with other groups, other times, other conditions, or a baseline. A claim with no comparison is usually untestable.

When and where — the period and setting. Evidence from 2013 social media is not evidence about 2026 social media, and this is easy to forget because the words did not change.

Then the counter-check: is it still worth answering? A question can be narrowed until the answer interests nobody, including you. If the honest answer to "so what?" is nothing, widen one element back out.

See it happen

Your question is "Why are people addicted to social media?"

The read that arrives on its own

It is a real question, people care about it, and there is plenty written — so you begin.

The same situation, with this idea

You apply the four. Who: university students aged 18–24. What: not "addiction" but a measurable pattern — daily use, night-time use, and score on a validated scale. Compared with: students with different use patterns, within the same population. When: current platforms, last two years. You get "which patterns of social media use are associated with higher problematic-use scores among university students aged 18–24?" — and note that "why" has quietly become "which is associated with", because a survey cannot deliver why.

After this, you can

see when your question is asking for a kind of answer your method cannot produce.

The takeaway

Narrow by naming who, what, against what, and when — then check that the result still matters.

Easy to take the wrong way

That "why" questions are bad. They are excellent and they are demanding: they usually need a design that can speak to mechanism, which module 08 covers.

Try this

Take a broad question in your field and write four versions, adding one element each time. Notice at which version you could start searching.