Original Research
Part of Digital PR campaign reviews
Planning a follow-up study from genuine unanswered questions
Turn documented questions from a PR campaign into a follow-up research decision with a defined population, method and claim boundary.
A follow-up study is justified when the first campaign leaves a specific, worthwhile question its evidence cannot answer. Gather questions from the study record, journalist requests and audience response, then decide which can be investigated responsibly. A request for a bigger headline is not itself a research question.
Build a register of open questions
Record each question in the words used by the person who raised it where possible, along with its source and date. Note what the current asset already answers.
A journalist may ask whether a result differs by region; a reader may ask how an eligibility rule applies to them; an analyst may find a subgroup too small to report. Those are different gaps. Check the first study before calling any of them unanswered.
Separate four cases:
- Already answered:The information exists but was hard to find. Improve the explanation or handover.
- Analysis possible:Collected data may answer the question after a documented calculation and disclosure check.
- New evidence needed:The existing measure, population or period cannot answer it.
- Not researchable as framed:The question assumes a cause, population or concept that needs defining before it can be tested.
Keep the question even when the decision is not to study it. The register helps prevent a later team from treating an unsupported answer as established.
Select a question worth pursuing
Prioritise the gap by reader usefulness, decision value, feasibility and the harm an unsupported answer might cause. Repetition in replies is a lead, not proof of importance to a wider audience. Ask who would use the answer and what they might do differently with it.
For each candidate, write a possible result sentence naming the group, measure and period. Then write the stronger claim the proposed method could not support. If a voluntary customer survey could describe participating customers but the desired headline says ‘Australians’, the design or headline must change.
The Australian Bureau of Statistics explains that non-random samples lack known selection probabilities for the usual estimate of sampling error, making population inference difficult.
Turn the gap into a study brief
State the question, target population, unit of analysis, measure, collection method, planned comparison, likely exclusions and what would make the result unfit for release. Say whether the work seeks to describe a group, compare groups or investigate a possible explanation.
A comparison across periods cannot establish a cause merely because the figures differ.
Suppose several recipients ask whether a finding from one business’s customers also appears elsewhere. The follow-up question concerns another defined population, not a rewording of the customer result.
Until the team has a suitable source and method for that population, the original finding remains about that business’s customers.
End the planning review with study, answer from existing material, redesign the question or hold. Record the reason and owner. Commission the next study only when its likely claim can stay within what its method can establish.
