Choosing isn’t hard.
You already know this story:
…At a luxury food store in Menlo Park, researchers set up a table offering samples of jam. Sometimes, there were six different flavors to choose from. At other times, there were 24. (In both cases, popular flavors like strawberry were left out.) Shoppers were more likely to stop by the table with more flavors. But after the taste test, those who chose from the smaller number were 10 times more likely to actually buy jam: 30 percent versus 3 percent. Having too many options, it seems, made it harder to settle on a single selection.
Such counterintuitive findings are what researchers live for. It’s led to the success of books such as Sheena Iyengar’s Art of Choosing and Barry Schwartz’s Paradox of choice.
But eventually more studies are done. People identify when exactly your phenomenon occurs, and in the case of choice overload (the name of the jam study effect), the answer is “not very often”.
Scheibehenne replicated Iyengar’s study in an upscale market in Germany and found no choice overload effects. The Germans seem to be able to cope with multiple flavors of jam just fine. [pdf]
Scheibehenne, Greifeneder, and Todd review 50 published and unpublished studies of choice overload and find a mean effect size of zero. Interestingly (but unsurprisingly), they find that studies that do not validate the choice overload hypothesis tend to not get published. Just like the regular media, journals report peculiarities, not the mundane.
Iyengar’s findings remain interesting, and her full professorship at Columbia University is well-earned. (I’ve read about a third of her papers.) But the practical usefulness of her findings have limits–severe limits that are not acknowledged in media accounts. Indeed, if you read her description of her study, you will find this critical information, clearly described [pdf]: (emphasis mine)
[C]areful attention was given to selecting a product with which most customers would be familiar, yet not so familiar that preferences would already be firmly established. Hence, to ensure that potential customers would not just reach for the more traditional flavors such as strawberry and raspberry, these common flavors were removed from the set of 28, leaving a choice set of 24 more exotic flavors.
The reported effect was only shown by removing the standard favorites that might have led to an easy selection and instead leaving customers to pick among such gems/jams as “kiwi”, “three fruits marmalade”, and (mmmm…) “lemon curd”. Customers were not selecting among either six kinds of jam or 24 kinds of jam. They were selecting among either six or 24 bizarro flavors they had no familiarity with. I’m not aware of choice overload having been demonstrated with any real company’s product lines.
Let’s take a look at another venue where choice overload has been observed: mutual fund choices in a 401(k) plan:
When there are only two fund options in a 401(k) plan, 75 percent of eligible employees participate. When the number of options rises to as many as 10, participation gradually slips to about 70 percent. When there are 10 to 30 choices, the participation rate remains stable at 70 percent, and as the number rises from 30 to as many as 60 options, participation steadily declines.
Perhaps that went by too fast. I’m going to type this part very slowly so it sinks in: when there are two options, 75% participate. When there are 30 options, 70% participate.
You have to add 28 options to discourage one additional person out of twenty. And that was sufficient to motivate HR Magazine to publish an article encouraging companies to slash choice in their retirement plans.
The choice overload hypothesis is perfect for a Malcolm-Gladwell-narrated world. It’s counterintuitive enough to be interesting. It’s complicated enough to show up approvingly in the news outlets catering to readers eager to think themselves smart*, but simple enough for them to narrate later at a cocktail party.
What’s more, choice overload is ammunition in a cultural narrative about consumerism and capitalism run wild. It’s the long-sought chaperone to Milton Friedman’s consumer choice orgies. As a result, these small, highly context-sensitive effects that ought to struggle for inclusion in a graduate psychology textbook are have become part of the public consciousness as if they were strong, robust findings that don’t disappear when you let people pick strawberry or try it on a German.
I present as evidence this 2004 piece in the New York Times by Barry Schwartz about why choice overload research means that Social Security should not be privatized. Schwartz cites all the popular choice overload studies and doesn’t seem to pause to consider the vast inferential gap between these narrowly constructed studies and Bush’s plans Social Security. The piece is not a balanced reflection of what exists in the psychology literature. It is a carefully limited set of evidence designed to cater to a narrative.
Which is sad. Because researchers probably do have some important messages about such things. But when the researchers themselves curate their citations to support a pre-selected solution, they are not functioning as scientists. They are advocates.
Photo credit: Wikimedia commons
* This is not a put-down of these sources, all of which I personally read. I am merely stating what I believe to be a fact about their audiences. (I have no evidence for this fact, but I suspect it is true.)