Predictably Irrational, by Dan Ariely

When we set the price of a Lindt truffle at 15 cents and a Kiss at one cent, we were not surprised to find that our customers acted with a good deal of rationality: they compared the price and quality of the Kiss with the price and quality of the the truffle, and then made their choice: About 73 percent of them chose the truffle and 27 percent chose a Kiss.

Now we decided to see how FREE! might change the situation.  So we offered the Lindt truffle for 14 cents and the Kisses free…

But what a difference FREE! made.  The humble Hershey’s Kiss became a big favorite.  Some 69 percent of our customers (up from 27 percent before) chose the FREE! Kiss, giving up the opportunity to get the LIndt truffle for a very good price. 

That is from Dan Ariely’s new and excellent Predictably Irrational: The Hidden Forces that Shape Our Decisions.  Here is Dan’s book-related blog.  All of a sudden my head is spinning, wondering what a relative price ratio really means (we can’t divide by zero).  Or is this just the Alchian and Allen theorem on steroids, namely the claim that fixed charges encourage the consumption of the higher quality good?  Or I think: "Zero, is there something special about that number?"

There is more on the way in behavioral economics.  There is Sway: The Irresistible Pull of Irrational Behavior, by Ori and Rom Brafman and Nudge: Improving Decisions About Health, Wealth and Happiness, the defense of voluntary paternalism from Richard Thaler and Cass Sunstein, due out later this June and April respectively.


When Tim Harford was here in Seattle a few days ago, the events coordinator introduced him but said, "[b]efore Tim starts, I'd like to let you know about upcoming talks that may be of interest." Among them was Dan Ariely. Everyone had a good laugh. Interestingly, Tim's talk was free whereas Dan's talk is $5. Interpret that as you will.

Was the test done in a store, on the street, in a lab? Many people don't carry change with them so a penny might as well equal a dollar, even before transaction costs.

What in the world is a Lindt Truffle? If I knew it was some sort of awesome candy imported from Switzerland or something similar, I would buy it for 14 cents over the generic kiss. But I have no familiarity with the brand at all.

Another possible skew, mayhaps?


Interesting comment. It hadn't occured to me, but there really is a cost to breaking a dollar. You end up with more pennies in a jar, more change lost, etc.

You also have to reach in your pocket, see if you have fourteen cents, get out your wallet if not, etc. Or you can just grab a free Kiss. I think the transaction cost explanation has a lot of merit.

I think this is easy to understand. We often assume that prices are set in a competitive market, where "you get what you pay for". If so, the one cent candy must be really cheap to produce, or the producer couldn't make a profit. Cheap to produce goods are often low quality. But when the candy is given away free, the producer is obviously not making a profit, so there is no longer any reason to think the candy must be cheap and low quality.

"Some 69 percent of our customers (up from 27 percent before) chose the FREE! Kiss, giving up the opportunity to get the Lindt truffle for a very good price."

Okay, but how many of these "customers" were actually people who, in the control scenario would not have bought anything at all? Certainly free offers attract free-loaders who might otherwise not buy anything at all if a transaction is involved.

Some clarifications:

We did conduct one of the experiments at one of the MIT cafeterias, where the offer (one of the two) was presented to the customers by the cashier at checkout. This means that all customers had purchased something and the cost of the chocolate would be added to their bill -- essentially making their marginal transaction cost 0 (or at least making the transaction cost the same across all conditions). The results in this setup were the same.

Based on this I suspect that transaction cost alone is not sufficient to explain the results.

As for the comment about why they are charging $5 for my talk, maybe it is because they are trying to get cognitive dissonance to work (people who pay might have a greater need to justify their payment and as a consequence might increase their motivation to enjoy my talk...)

BTW In the book there is a chapter about my attempts to charge people for my poetry readings...

An element no one else is raising is the nature of the commodity. Chocolate is a luxury good. You don't "need" a chococlate. It's a thing that either delivers pleasure or why bother? So people may have been thinking "Well, if I'm going to bothering to have a chocolate at all, I'm gonna go for the really good stuff." But then it shifted as soon as the cheap stuff became not just cheap but free. So down to a penny taste and discernment (and hope) overwhelm price consdierations. At the end, though, the attraction of "free" overwhelms the promise of pleasure-excellence. But do we know if this holds when the good being offered is something other than a luxury-type "pleasure" good?

I dunno. I have the impression that the general rules people play by when it comes to art-pleasure-luxury goods are different than the rules people play by when it comes to necessities ... And that they're flukier and more changeable too. Pleasure and subjective experience are not just hard to predict but hard to know, and of course people can fool themselves too. (Lindts may not in fact be any good. But who's to judge? After all, many sucko movies make a fortune ...)

Haven't read the book, and I imagine it's fascinating. But is Dan really arguing that "irrationality" can be predicted? Really? If so, can I introduce him to my wife?

Besides, isn't it of the nature of irrationality to be unpredictable, except maybe in a uselessly general, "shit happens" kind of way?

Plus; Wouldn't you think Hollywood could do a better job of knowing what's going to be a hit if "predictable" were any part of the game? William Goldman: "Nobody knows anything." Decades of experience, and that's the best generalization anyone's ever come up with where what's-gonna-work-in-showbiz is concerned.

Or is that just a catchy title?

Whatever the merits of the book (and I am very optimistic based on his promise of poetry recitals), Ariely has some darn good brain teasers on his (not the book's) web site under "riddles."

To Michael B.
Predictability has nothing to see with rationality. We can predict very well the behavior of a planet, meteorit, whatever... plants or animals also. No rationality involved.
We could indeed end up with a better theory of human behavior that is completely orthogonal to rationality (and which appears in many cases to be "irrational").

tom s.,

Sorry, zero can be a price. In standard economic theory, if a good is in excess supply, its price is zero. It is only a matter of definition that one "must" have exchange for their to be a definable price.

Regarding the curios nature of the number zero, my late mathematician father was once asked by a young woman in a public lecture if zero is a real number. He replied, "one of the finest, my dear, one of the finest."

And, as for change, I must tell a story involving my (Russian) wife when we visited Moscow in August 1992 during the peak of a hyperinflation. Kopecks are officially worth 1/100 of a rouble, but the value of a kopeck had fallen well below that of the copper in them, so they were being melted down and were becoming scarce. OTOH, in the old days, public pay phones cost two kopecks to use, and did not provide change and had not been replaced yet. When my wife wanted to make a phone call on such a phone, she had to pay three roubles to obtain two kopecks to make the call.

"Many people don't carry change with them [...]"

Man, I feel old-school

Michael -- I am a big fan of Richard Gregory. Another 2 people you might want to look at are

Why is this choice pattern irrational, as opposed to unreasonable -if it is.

Situation A: Choose between O1 and O2, where O1 is cheap but not great and O2 is expensive but great.

Situation B: Choose between O1 and O2, where O1 is free and O2 is not, but O2 is better quality.

Decision Criteria:

C1) When comparing items with prices, try to trade-off price against perceived value.

C2) If something is free, choose it unless the quality of the pricier item is much greater.

I don't see anything formally irrational about these choices based on these decision criteria.

What I am missing?

I don't get it. In the first example, you can get 15 kisses for the price of 1 truffle. In the second, you can get +inf kisses for the price of 1 truffle. Why wouldn't people change their preference? I must've missed something...

One clarification!

In the studies the participants could only get one of the chocolates.

This mens that they got one and gave up one -- and this is why in our case it was irrational to give up the better deal (the better chocolate for 14 more cents) for the free one.

It is basically like standing 2 hours in line to get Free ice cream. By the time the two hours are over you gave up a lot to get the Free! ice cream

I see that Dan beat meto the punch. I do have some questions for him.

The percentages as given suggest that everyone chose one chocolate or the other. is this accurate? if there were those who did not want any chocolate then you should certainly tell us all three percentages. Clearly, if a large number refused chocolate the degree of reversals might be less dramatic.

In a related point, it seems important to consider that some people were indifferent between the choices and hence chose randomly. It is wildly unlikely, of course, to get these results even if everyone chose randomly, but what if the preferences are somewhat random? Consider restaurants. Lots of people tend to eat in the same restaurant fairly often. They don't always order the same meal. Is that irrational? How does it differ from your experiment?

Bernard writes: "It is "irrational" because the preferences are reversed in the two situations. Consider two sets of goods:

(a) truffle
(b) Kiss plus fourteen cents

An individual who buys the truffle rather than the Kiss in the first situation (paying 15 cents) must prefer (a) to (b). IOW, such a person is willing to pay fourteen cents for the difference between a Kiss and a truffle."

Isn't this just analogous to the common ratio effect in expected utility theory?

And do you hold that those violations are also irrational?

It is my recollection that Mark Machina wrote a couple of great papers in the early 80's about these types of problems in the expected utility theory context, but it strikes me that you have a similar problem here.

Further, there isn't even a possible money pump argument here.

1. I am in the line with free HK and the 14 cent truffle. I have 14 cents in my pocket. I take the free HK.

2. I move to the next line, the one cent HK and 15 truffle line. I sell my HK and now have 15 cents. I get my truffle.

3. I move back to the original line, sell my truffle for 14 cents and take the free HK.

I am back at the starting line with no losses. No money pump argument, no even putative irrationality of preference change.

Bernard and Ariely;

Independence is necessary for certain forms of maximizing functions.

But, in this case, I don't see even a small problem because of the absence of the money pump.

You have found a utility function of two variables, U(x,y) that is not linear in the second variable - price.

The next step would be to show that this is consistent with some version of utility maximization.

I will wait for that step to be explained -because as I said earlier, I don't see it.

Very nice article! Thanks for this!

This is some very valueable information, thank you very much.

Is it realistic ?

if u love somebody , send him work for google , if u hate somebody , send him work for google .

I think the "Kiss" experiment only shows that people associate some positive pleasure with receiving something free, as opposed to some presumably low displeasure for paying one cent.

In the "Armchair Economist", the popcorn example shows how easily economists mislead themselves. If the economist's calculation doesn't explain reality then it is his calculation which is wrong. By definition, people's actions are rational, as they reflect people's preferences and beliefs at that instant.

Also, I understand that "pleasure units" are a attractive and convenient tool for maths-craving economists, but they are nonsense in every other way. Let's stop fooling ourselves. Kiss = 5 units of pleasure!?!

Comments for this post are closed