« Request day | Main | Progressive taxation as price discimination »
Choosing an Inferior Alternative
It's well known that people suffer from confirmatory bias, so after they buy a new car they eagerly read the advertisements for that car - the ads, of course, confirm that their purchase was a good one.
Many people also have a predecisional bias, they interpret new information in a way that is biased towards the leading candidate - a confirmation bias in expectation. In Choosing an Inferior Alternative (also here) Russo, Carlson and Meloy show that careful manipulation to take advantage of predecisional bias can actually cause people to choose inferior alternatives.
The authors ask people to rate restaurants, nominally named A,B,C etc., on a series of attributes (atmosphere, hours, parking, dishes and so forth) thus creating a ranking. Two weeks later they ask the same people to rate the same (but renamed) restaurants in a series of pairs. But this time they put the attribute that most favored the inferior restaurant first - thus the inferior restaurant would win the first comparision and further attributes would suffer from predecisional bias. They also put the attribute that had the second strongest favorable rating for the inferior restaurant last to take advantage of any recency effect. The least favored attributes were buried in the middle.
Compared to a control group, twenty percent more of the treatment group chose an alternative that according to their own preferences was inferior. In fact, in the treatment group a majority chose the inferior alternative.
Posted by Alex Tabarrok on February 27, 2007 at 07:36 AM in Economics | Permalink
Comments
Are we supposed to understand that the first ranking expressed authentic preferences while the second ranking didn't? Why? I don't get it. These things are two weeks apart. Wouldn't it be more accurate to say that people don't have stable preferences, and are therefore easily manipulated, rather than embrace an absurdity like counter-preferential choice? Perhaps I should read the actual paper.
Posted by: Will Wilkinson at Feb 27, 2007 8:31:30 AM
Problems comparable to Keynesian beauty contest?
Posted by: GVV at Feb 27, 2007 9:08:21 AM
This principle applies when writing the dreaded "rejection letter". By opening such a letter with something positive, warm, or friendly, it takes the sting out of the line "Unfortunately, we cannot consider your application at this time".
Posted by: Dave at Feb 27, 2007 10:13:17 AM
Will, yes. The problem you raise is the reason there is a control group.
Alex
Posted by: Alex Tabarrok at Feb 27, 2007 10:21:50 AM
They use a group -lets call it "control group"- to control for any changes or differences that may have happened in the two week period. So you could say that 20% of the group showed "counter-preferential choice".
Posted by: economister at Feb 27, 2007 11:07:58 AM
I still don't get how a control group changes how we interpret the result. (Still not having read the paper!) The people in the control group didn't change their preferences, and the people in the test group did. The experimenters have simply identified elements of the context of choice to which our unstable preferences are sensitive, not conditions under which people will choose against their preferences. People have different preferences when you present information to them in different ways. We change our minds a lot in fickle and possibily manuipulable ways (threatening to turn us into money pumps). People who are not manipulated in certain ways won't change their minds in the same way. That's what it really says, right?
(At this point, I think I'll just refuse to actually read it out of perverse principle.)
Posted by: Will Wilkinson at Feb 27, 2007 4:12:37 PM
So Will, if I put you in a box with an electronic version of the questionairre, and the computer that would show you the questions had an exactly 50/50 chance of showing you the questions in a preference-manipulating order, does that mean you both prefer and don't prefer the same restaurant at the same time? ;-)
(Sorry, couldn't resist.)
Posted by: Flynn at Feb 28, 2007 8:18:17 AM
Flynn, Right idea! The phenomenon measured is not independent of the measurement instrument.
Posted by: Will Wilkinson at Mar 1, 2007 9:45:24 AM
大家好,我是臺灣人,從臺灣一個人搬家來到美國,環境很陌生,感覺很孤單。以前在臺灣幾家知名的徵信社工作過,我是一個優秀的徵信工作者,希望早點找到適合自己的工作。希望通過貴站,認識更多的朋友。
Posted by: 謝文豪 at Apr 1, 2008 9:33:09 PM






