Asking better research questions: five types and when to use each

The difference between a research project that drags on for months and one that finishes cleanly is almost never effort. It's the question. A well-formed question defines what to read, what to measure, and what to ignore. A vague one makes everything relevant and nothing decisive.

The five question types

Most research questions fall into one of five families. Each serves a different aim, and confusing them is the most common source of stalled research.

• Descriptive — what is happening? Documents characteristics, prevalence, or patterns. Example: "What are the defining characteristics of remote-first companies?"
• Comparative — how do things differ? Contrasts groups, cases, or time periods. Example: "How does onboarding differ between remote and in-office teams?"
• Causal — what causes what? Tests whether one thing drives another. Example: "What is the effect of async-first communication on team productivity?"
• Exploratory — what's going on here? Investigates a poorly understood area with no prior hypothesis. Example: "How do new managers experience their first remote team?"
• Evaluative — how well does it work? Judges effectiveness, value, or quality. Example: "How effective are daily standups for remote teams?"

Pick the type before you pick the topic

The type should follow from what you want to know. If you want to prove that something works, you need a causal or evaluative question, not a descriptive one — no amount of description will prove a cause. If you are entering an area with little existing work, an exploratory question lets you discover patterns instead of forcing a hypothesis too early.

Sharpen the question until it's answerable

A generated question is a starting point, not a finished one. Sharpen it by adding three things: a population, a context, and a time frame. "What is the effect of async communication on productivity?" becomes "What is the effect of async communication on productivity among software teams at mid-size companies over one quarter?" The second version tells you exactly what to study and what to ignore.

1. Add a population — who exactly are you studying?
2. Add a context — where, in what setting, under what conditions?
3. Add a time frame — over what period?
4. Check the scope — can you actually answer this with the data and time you have?

Let the question drive the method

The question type largely determines the method. Causal questions need comparison and control. Descriptive questions need systematic observation. Exploratory questions need flexible, open methods like interviews. When your method fights your question type, that's a signal to revisit one of them before you invest real time.

Ask a better question and the rest of the work gets easier. Ask a vague one and no amount of effort will rescue it.