Stronger wording only paid off when the speaker knew the subject

On-site experience

Reader Note:: Study 4A was an online experiment with 527 Prolific participants in a used-car scenario. The source was either someone whose knowledge was relevant, "a mechanic you trust," or someone whose knowledge was not, "someone who doesn't know much about cars." Each then said either "Almost certainly, this car is in good shape" or "It feels like this car is in good shape," and persuasion was rated on the same seven-point scale used in Study 1. With the mechanic the stronger wording won, 4.88 against 4.53 (p = .014); with the non-expert the gap disappeared, 3.64 against 3.68 (p = .781). Note that the non-expert scored about a point lower in absolute terms either way, so this is about whose wording is worth tuning, not about rescuing a source nobody trusts. The paper does not say when the experiment ran.
Evidence:: Experimental
Vendor interest:: None
Sample:: Six online experiments, mostly with Prolific participants; disclosed samples include N=15, 21, 111, 111, 400, 527, and 456, plus 3 independent coders in Study 1.

Good for

Backing budget and pricing decisions in similar situations.

Situations far from the one tested, without checking again.
This was measured in one market. It may not hold where you are.
The source never says when this data was collected. It could be considerably older than the publication date suggests.