How a badly worded question manufactures an opinion

A question containing an undiscussed assertion obliges one to accept it in order to answer. Asking whether one approves of the way a problem is handled presupposes that the problem exists and that it is being handled, two points on which the person has not been consulted.

The presupposition is all the more effective for being banal. Nobody thinks of contesting a premise stated in passing, and the answer obtained will then serve as proof that the premise was shared.

It joins two ideas in the same item, for instance a general value and a particular means. The person who approves of the first but rejects the second can no longer answer correctly, and their choice becomes arbitrary.

The result is uninterpretable while appearing clear-cut, which makes it the most dangerous defect: nothing in the final figure signals the ambiguity. The marker is simple: if an item contains “and” or “in order to”, one must check that two things are not being measured.

Certain terms carry an evaluation before any debate: waste, laxity, hammering, but also progress, modernisation or common sense. An item containing one measures attraction to or repulsion from the word as much as the position on the question.

The effect is measurable and sometimes considerable: on political questions, replacing a term by a less loaded synonym shifts answers by several points. That is why comparing two surveys without comparing their wordings makes no sense.

An amount cited before the question influences the evaluations that follow, even when it is manifestly arbitrary. This phenomenon of anchoring, widely documented, applies to prices as well as to quantities and durations.

Order acts differently: an item placed after an emotionally strong one does not receive the same answers as in isolation, and a general question asked after several precise ones inherits their framing. Rotating the order between respondents is the usual remedy.

One idea per item, descriptive rather than evaluative vocabulary, a concrete situation rather than a grand principle, and both sides of the question worded with equal care. These rules are simple to state and demanding to apply, because the most natural wording is almost always slanted.

The test a writer can impose on themselves is this: would the person who thinks the opposite of me find this item fair and answerable? If the answer is no, the item will measure adherence to my way of putting the question, and nothing else.

For a reader, the counterpart is a single reflex: look for the exact wording before commenting on a figure. When it is not published, that is in itself information about the seriousness of the survey.