A version with more completed responses can still have a lower response rate if many more people were exposed to it. Start with counts and denominators before deciding what a comparison actually shows.
Need help choosing a next step? Scan the contact QR code below or contact Hanlin on WeChat. Send CBPA RESPONSE RATES 1030 and tell us your school year and question.
Calculate the two rates
In an original fictional exercise, version A is shown to 100 people and receives 10 completed responses. Version B is shown to 300 people and receives 15 completed responses.
A has a response rate of 10/100 = 10%; B has 15/300 = 5%. B produces more completed responses, but A has the higher rate. Those are two different observations, not contradictory calculations.
Define the event and exposure consistently
The numerator must refer to the same completion event for both versions. The denominator must describe a comparable eligible exposure, rather than page views for one version and distinct visitors for the other.
If the same person appears several times, explain how the counting handles repeat exposure. A percentage without a clear unit of analysis can conceal a difference in the underlying measurement.
Separate a comparison from a causal conclusion
These fictional percentages alone do not prove that the design caused the difference. The groups might differ, the versions might run at different times, or the assignment process might introduce bias.
A planned comparison should consider suitable assignment, common conditions, a defined observation period and uncertainty. Even random assignment does not make a small observed difference automatically conclusive or guarantee that it will recur.
Choose the question before selecting a winner
Total completed responses and completion efficiency answer different questions. A student report should explain which outcome matters for its stated learning question, rather than choose whichever metric favours a preferred version after seeing the results.
This is a general business-research lesson relevant to CBPA interests, not an organiser's required experiment or a completed project submission. The counts are invented teaching examples. Students should independently design any permitted research and should never relabel these figures as real customer observations.
Continue reading: CBPA Business Skills: A SWOT Label Is Not Evidence.
Contact Hanlin
Planning a first attempt or reviewing a topic you find difficult? Scan the QR code and send CBPA RESPONSE RATES 1030. Include your school year, country or competition region, experience and the question you want to discuss.
On your phone, use the button or save the QR image and scan it within WeChat.
Official information and further reading: Source 1
