I haven't read the study yet, but that quote has it's own interesting (to me) subtle bias in it:
nearly half of the people chose the more expensive pen.
fewer than 20 percent...
I think it would be better said as (emph on changes):
In one study, researchers offered people a choice between two different pens.
When the pens were priced at at $2.00 and $3.99, fewer than 50% of the people chose the more expensive pen.
But when the prices changed by a single penny, to $1.99 and $4.00, fewer than 20 percent of people chose the more expensive pen.
Why change from "nearly half" to "fewer than 20%"? Isn't "nearly
half" also "fewer than 50%"? Although it's obvious that "half" is the same as "50%", it bugs me when writers change terms midstream. That's fine for literary works to provide interest, but when presenting data, keep the terms the same. I think the "nearly" half" provides a subtle bias of making the first group seem even larger than it is, relative to the group described as "fewer than X%".
Is there a name for that kind of bias?
edit/add: Further, the study is flawed if they were looking for left digit bias, as the relative value changed (albeit by a small amount). And even though they say they changed by a single penny, the difference is two pennies, and the difference is what matters. A better study would be:
$2.49 and $3.59 pens (a $1.10 difference, and a left digit difference of 1)
$2.99 and $4.09 pens (a $1.10 difference, but a left digit difference of 2)