The reason a question persuades better than a statement

The same proposition can be delivered as an assertion or arrived at by the audience in response to a question. The second version is remembered longer and defended more readily, because the listener holds it as something they worked out rather than something they were told.

What everyone sees

Rhetorical questions are read as a stylistic flourish for speeches, and their effect is credited to engagement. Engagement is vague enough to be unfalsifiable. The measurable difference concerns who produced the conclusion, and it is bounded by conditions that also explain when the technique backfires.

What is actually happening

Petty, Cacioppo and Heesacker tested rhetorical questions and found they increased the processing of message content when the audience was not already motivated to think, while reducing it when motivation was high, so the effect reverses with involvement. Slamecka and Graf established the generation effect: material produced by the participant is retained better than the same material read. Combined, the two results predict what is observed — questions work by transferring production of the conclusion to the audience, which improves retention and commitment, and fail where the audience was already engaged and experiences the device as an interruption.

Why it stays hidden

The condition hides because speakers see the same reaction in both cases: nods. Whether the conclusion was generated internally or merely accepted is not observable in the room, and it only shows up later in whether the audience can restate the argument. Nobody measures that, so the technique is either adopted universally or dismissed, rather than applied where it works.