The incredible flipping preference

What we truly prefer is sometimes not what we choose

There is a new coffee place in town. Monday last week temptation got the better of me and I decided to try it out. The Americano was rather pricey at £3.50, but it tasted so very nice, that I went back the next day. The barista is an intriguing person, with an extraordinary memory. Having seen me only once before, she still immediately greeted me by my (admittedly unusual) name. She also has a rather peculiar way of proving that her coffee, costly as it is, is the best value for money.

On that second day, she made me an offer. She would dilute her coffee by 2% and charge me 10p less, and let me compare that with the standard full strength coffee. If preferred the newer one, I would be guaranteed to always get that cheaper variant — just 2% diluted, but at 10p less. So I watched her do her magic, and then sipped alternatingly from the two cups in front of me. Much as I tried, I could not really spot any difference, so rational dude that I am, I went for the cheaper version.

“I thought you would,” the barista said. I smiled, and worked out that I’d just saved myself £25 on my annual coffee budget.

On Wednesday, she made me the same offer: reduce the strength by a further 2%, and another 10p discount, if I preferred this version over Tuesday’s. Once more, I couldn’t notice a difference between today’s and yesterday’s variants, so as before I decided to go for the cheaper coffee. “I thought you would,” she said again, with a twinkle in her eye. I was now going to save £50 a year on my daily coffee. Great!

Thursday, and Friday, the same story.

So the next Monday, one week after my first visit, I was wondering whether she’d be proposing to reduce the strength further, until I’d stick with my last preference. And indeed she asked, “You preferred Friday’s version over all the coffees you tried last week, right?” I nodded. I was now drinking coffee that unnoticeably little less strong, but I was paying 40p less per cup. That would save me more than £100 per year — the price of a very nice meal for two.

“Here it is again, and here’s a cup of my standard coffee to compare it with,” she said, “the one you had on your first visit here. So which do you prefer?” I took a sip of Friday’s variant — it was pretty much what I remembered from before the weekend. Then I tried the original brew again. My goodness, how different it tasted, so much richer and fuller! If I hadn’t seen her prepare both cups, I would have sworn she was playing a trick on me.

“So, what’ll it be? Which is the best value for money?” she asked. There was no way I could envisage myself drinking that weak Friday concoction ever again — and after all, the wonderful, exquisite full-strength brew was barely 40p more. “I’ll have your standard coffee, now, and forever more,” I said. “I thought you would,” she responded.

What just happened violates one of the cornerstones of conventional microeconomics: the principle of transitivity. Rational people should, if they prefer B over A, and C over B, prefer C over A and not the other way round. I had favoured Friday’s coffee over Thursday’s, Thursday’s over Wednesday’s and so on…

…but I had also preferred last Monday’s over Friday’s. And that is, well, not rational.

“Are you by any chance a behavioural economist?” I asked the barista. She just smiled at me and winked*.

Here is a possible explanation for my inconsistent preferences. The difference in quality, resulting from a 2% dilution, between coffees on two consecutive days is insignificant. This makes the other difference, the difference in price, stand out. It is of course rational to go for the cheaper coffee if it appears to have the same quality as the more expensive one. But when the accumulation of quality reductions becomes discernible in comparison with the original coffee, it outweighs the relatively modest saving in cost.

We’re not off the hook, though. Even without such a somewhat convoluted circular comparison chain, our preferences are far from straightforward and stable.

In 1992, Max Bazerman, George Loewenstein and Sally Blount carried out a set of studies on how justice and fairness are perceived. One of their findings was that “individuals’ preferences […] can reverse, depending on whether potential outcomes are evaluated sequentially or simultaneously.”

Well before this research took place, I had been saving up my pocket money for a portable radio. A local department store had a whole range of them on display, and on the way back from school I often popped in to gaze at the models on offer. One particular type had become the focus of my attention: the Satellit 210 was big expensive, but it was so impressive with three times as many wavebands as the other models, that this was the one I wanted! My dad challenged me, though — what on earth would I do with nine shortwave bands? In the end, I saw sense and I got myself a radio with just 4 wavebands, for less than half the price and weighing less than a fifth of the Satellit. (And to be honest, I never once listened to the single shortwave band on this one.)

Tough choices (images: Wikimedia, kbmuseum)

As soon as I stopped comparing radios side by side, and simply considered the options separately on their merits, it was clear that the shortwave bands were a complete red herring, and that an expensive beast of nearly six kilos was not what I needed. My preferences were reversed.

In a recent paper, Nudge co-author Cass Sunstein, the man who writes faster than his shadow (and certainly faster than most of us can read), delves deeper into this phenomenon. Using examples from elsewhere in the literature, he explores how and why our preference across two options might flip, depending on how we consider them. This is not just intriguing, it can also be welfare-reducing. If we have a true, underlying preference, but the context can lead us to believe the opposite, we risk ending up with the worst option.

Evaluating something means judging its features. That is easier to do if we have another option to liken it with: the characteristics in which they differ will then be prominent in our analysis. The problem is that these salient attributes may not be the ones that matter most to us (like the number of wavebands).

In the absence of an alternative, we have less information about the relative position of what is on offer. Even numerical data are often abstract without a reference point. Is a laptop battery life of 6 hours impressive or just mediocre? Is 10,000 entries in a dictionary a lot or a little? With no benchmark to compare it with, we are open to misinterpretation (“10,000 sounds like a lot”).

Cunning salespeople can take advantage of the weaknesses of both situations. If they want you to buy an item with positive features that are generally easy to evaluate on their own, and less positive ones that are harder to assess in isolation, they are better off showing it separately. In the opposite case, they would enable the comparison with another option, and highlight the attributes where the choice they want you to make is superior — irrespective of the degree to which this matters to you.

But as my radio experience illustrates, we are perfectly capable of misleading ourselves, without the help of a devious shop assistant. The trouble is that neither joint nor separate evaluation is inherently better for establishing our true preference. When we need to decide between options, and choose which will serve us best, we ought to look at the features that matter most. But in practice, we are driven to focus on what stands out, and what is easiest to evaluate. And that tendency can trip us up in both cases.Being aware of the problem can help, though. If you come to a conclusion using one type of evaluation, try the other. If it remains unchanged, great! If it flips, you know some more thought is needed. Ask yourself (or get someone else to ask you!) what truly matters, and then make your choice.

That’s the way to outsmart yourself (and any crafty salespeople).

*This story is based on an example in Nick Chater’s online course The Mind is Flat.

Originally published at on July 27, 2018.

Thank you for reading this post. If you enjoyed it, please give it some claps (this helps others find it too) and share it far and wide. You may also want to check out my other posts (they are published every week — see here). Thanks!

Accidental behavioural economist in search of wisdom. Uses insights from (behavioural) economics in organization development. On Twitter as @koenfucius

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store