No Extra Credit for Delivering on Promises
psychologicalscience.org
psychologicalscience.org
Worth Keeping but Not Exceeding: Asymmetric Consequences of Breaking Versus Exceeding Promises
Ayelet Gneezy1⇑
Nicholas Epley2
Does anyone else find this type of study completely unconvincing?
I assume that the students are being asked to pick a number 1 through 10 to express their level of satisfaction. They're just going to pick an arbitrary point which is going to be different for everyone. You can label the points (10 is winning gold at the Olympics, 1 is your spouse dying) but it's really hard to place something like buying concert tickets onto that scale.
The only way you could get actual information out of such a study would be to ask the same student how satisfied they are with a row 10 seat, a better seat, and a worse seat, but that will only tell you the obvious - that they are to some extent more satisfied with the better seat and less satisfied with the worse seat - and not give you any interesting information about whether it's a significant difference.
What you could do is look at actual data (for concert tickets, plane tickets, train tickets, etc) and see if complimentary increases in ticket quality really correspond to increased purchases in the future. You might also be able to rig up an exit survey with objective questions like "do you regret your ticket purchase" and "would you recommend us to your friends" and compare the answers given by regular ticketholders and those given by ticketholders given a complimentary increase in ticket quality. But, again, that wouldn't work in a lab setting.
In other words, they may have proven that people don't value seats very much, but I'm not sure that applies to every possible reward. Something where preferences are less abstract (e.g. cash) might turn out differently for all we know and there's no way to tell if the authors already investigated this or not from just this story.
The odd thing is that I couldn't find this particular experiment mentioned on the linked paper[0]. There were several other experiments that seem more clearly defined in terms of failing/meeting/exceeding expections.
Certainly, i think he hoped we would earn extra goodwill by over-delivering. These studies seem to shoot that down.
But there was also a risk-control aspect. If things went wrong, and we couldn't deliver as effectively as we'd hoped, then we still had a chance of at least meeting the expectation. That still seems useful.
To test that theory, they asked college students to imagine using an online company to buy concert tickets for Row 10. The participants then were randomly instructed to consider receiving seating that was either worse than, better than, or exactly as promised. They then indicated their level of satisfaction with the purchase, as well their likelihood of both using the online ticketing company again and recommending the service to friends.
One hundred and twenty University of Chicago undergraduates (N ¼ 60)