How to read a best–worst choice in a vaping survey
Identify the two choices inside one product profile and keep attribute levels separate from overall importance.
No Vape · Updated

Reconstruct one profile from the methods
A BMJ Open e-cigarette study describes a case-2 best–worst experiment. Its tasks presented levels of selected product attributes and asked what made participants most and least want to use the product. Figure 1 shows an example task. A level is a specific value shown for a characteristic, such as a particular upfront cost for the purchase-price attribute.
Read the instructions printed above the options. In this study the participant identified two features within one profile. In a different experiment the participant might choose between complete products. Those two designs leave you with different answers. Write the exact judgment requested before comparing charts that both use the word choice.
Try one card with fictional options
Imagine a hypothetical product profile listing a one-time cost of 20 credits, a plain case and a low-contrast display. One person selects the plain case as their most favorable option and the low-contrast display as least favorable. The card gives two comparative judgments. It leaves open the size of any preference, whether the middle item mattered, and whether the person would accept the product at all.
Now replace only the listed cost with 40 credits in another imaginary task. If the cost becomes least favorable, keep the value that was on each card attached to its selection. A tally that merges all prices under “cost” would lose the information that the levels differed.
Distinguish the level result from the attribute summary
The paper reports model estimates for specific levels and a separate summary of importance across attributes. When a plot shows an estimate for a price level, find the model and the comparison the authors use. If a second figure summarizes purchase price as a whole, read its calculation separately. The units and the objects compared can differ even when both figures use the same survey tasks.
Put the chart label beside one of three notes. It may refer to an observed choice on a card, an estimated preference for a specific displayed level, or a summary for the broader attribute. Ask for the missing label or calculation if you cannot tell which it is.
Keep the scenario attached to the conclusion
These comparisons depend on the features, levels and sample that researchers put into their experiment. The example does not tell you what everyone values, which product they would buy, or whether any displayed health wording is accurate. A survey task about product preference has no clinical recommendation for you.
Keep one complete card and its exact decision rule next to the chart you want to explain. This makes it possible to describe what people compared without treating every mention of an attribute as the same judgment.
Questions about this step
Can I read the least favorable feature as an absolute rejection?
The choice orders the shown levels against one another. Look for a separate accept-or-reject item if you need evidence that the participant rejected the profile as a whole.
Sources and scope
Sources inform the guidance. The worksheets and examples are No Vape editorial suggestions, not tested treatments.
Written with AI assistance and checked against the cited sources. No clinical review is claimed. This guide covers practical planning, not individual medical treatment. How we research and write