In "Section IV Shopping" of Kőszegi / Rabin (A model of reference-dependent preferences, QJE 2006), the example of consumer buying a pair of shoes is given.
They claim that "her disutility from spending on the shoes is between $(1+\eta)p$ and $(1+ \eta \lambda)p$" (p. 1146).
How do they get this result?
I can see where the second term comes from, namely buying at $p$ when expectations were to buy at 0, which generates money-related utility of
$$ -p - \eta \lambda p = - (1+\eta \lambda) p$$
But the first term is beyond me: If she buys at $p$, but expected to buy at some $p' > p$, I would obtain
$$ - p + \eta(p' - p) = -(1+\eta)p + \eta p' $$
which is not the same as the $(1+\eta)p$ stated in the paper.
What am I missing here? Thank you.