Business Problem to Research Question

What is the difference between a business question and a research question?

The business question is the choice the organisation faces, and it contains the options: retain, restructure or withdraw. The research question is what a study must establish so that choice can be made, and it contains no decision verbs. If your research question still contains the word "should", the two have not been separated.

How do I write a good research question?

Start from the decision, not the topic. Write the decision in one sentence with at least two options and a date, then write the single question a study would have to answer for the decision owner to choose. Test it three ways: can it be answered by observing, measuring or listening to someone; does it presuppose its own answer; and would a finished report visibly answer it or not.

How many research objectives should a study have?

Fewer than a brief usually arrives with. Three to six sub-questions is a workable range for a single study. The constraint is physical rather than stylistic: interview length, fieldwork budget and analysis time are finite, and every question added takes capacity from another. Rank as must, should and could, and cut from the bottom until the set fits.

What is the difference between research objectives and information needs?

An objective is what the study commits to deliver. An information need is the specific thing that must be measured or understood to deliver it. "Understand price sensitivity" is an objective. "The price at which each user's chosen alternative becomes preferable, by user type" is an information need. Skipping the information needs layer is what produces questionnaires that miss the obvious.

What questions can research not answer?

Four classes. Questions about the future, because research measures the present and the recalled past. Counterfactual questions, which need a comparison that does not exist. Questions about what people would do, because stated intention ranks options well and predicts rates badly. And causal questions with no comparison group, which can only be answered as an account of what people believe caused something. Each can sometimes be converted into an answerable form, provided what was lost in the conversion is written down.

How do I know whether research is needed at all?

Ask what would be done differently depending on the answer. If every plausible answer leads to the same action, no research is needed. The same applies when the answer already exists in the organisation's own data, when a recent study answered it, when the decision must be made before fieldwork could report, or when the cost of the research exceeds the value of the decision.

What if the decision has already been made?

Say so, and do not produce a question set. Research commissioned to ratify a decision cannot change anything, and a polished set of objectives makes that harder to see rather than easier. The correct output is a short statement naming the decision, the evidence already relied on, and the fact that no result would change the course.

Should every research question have a hypothesis?

No. A hypothesis is worth stating where prior evidence or theory gives a genuine direction and a result could refute it. Where there is no prior basis, a hypothesis is a stakeholder hunch encoded into the instrument, and it will bias the design that follows. State that none is warranted rather than filling the section.

How do I decide which stakeholder questions to cut?

Score each on three things: whether the decision changes depending on the answer, whether it can be answered as posed, and whether it is already answered elsewhere. Anything failing one of the three is cut. Publish the cuts with the reason and the requester's name attached. Questions removed silently return at the debrief.

When should the research question be written?

After the brief has been interrogated and before any method, sample or instrument decision. Choosing a method first quietly redefines the population as whoever the method can reach, and that redefinition never appears in the final report.