Recent research is exploring the case for choice-induced changes in preferences using the free-choice paradigm of Brehm (1956). Participants are faced with a choice between items that they have given the same rating of liking, two items at a time, and it is found that an item not chosen in one choice has a lower tendency of being chosen in a subsequent choice. This tendency is interpreted as evidence for choice-induced changes in preferences. I argue that this interpretation of the evidence is invalid. Furthermore, I report a novel experiment in which participants were specifically asked to compare the items, allowing for a consistent interpretation of the evidence. I find no evidence of choice-induced changes in preferences after a choice between items where one was viewed as more attractive than the other, but potentially some weak evidence of changes in preferences after a choice between items viewed as equally attractive.