When Predicting Other People's Preferences, You're
Probably Wrong
Marketers, job hunters and people looking for mates are all
called upon to predict behavior—and many are probably wrong. The reason: We too
easily make assumptions about what others will like based on their previous
choices, according to new research by Kate Barasz, Tami
Kim, and Leslie John.
The Bachelor is a wildly popular reality dating game show
on which 28 women compete for the hand of a single man. Along with flirting and
fighting and engaging in feats of derring-do, many of the competitors spend ample
time confessing insecurities to the camera.
Leslie K. John admits that she has been watching the show long
enough to notice a pattern in the self-doubt: The women tend to fret when they
see the supposed man of their dreams bonding with women who are dissimilar to
themselves. “One thing that’s commonly said is, ‘Well, if he’s into her,
there’s no way he could be into me,’” says John, an assistant professor in
the Negotiation, Organizations, and Markets unit at Harvard Business School.
“They assume the Bachelor can only like one type of woman.”
It turns out those fretful competitors support new scientific
findings about presuming preferences. When predicting other people’s tastes, we
tend to erroneously assume that liking one thing precludes enjoying another,
dissimilar option, according to a recent set of studies by researchers at
Harvard Business School, which were led by HBS doctoral candidate Kate Barasz
and conducted in collaboration with John and doctoral candidate Tami Kim. Their
research is detailed in their paper The
Role of (Dis)similarity in (Mis)predicting Others’ Preferences,
which will appear in a forthcoming issue of the Journal of Marketing
Research.
The findings have important implications for anyone looking to
impress others, for those who are tasked with forecasting consumer behavior, or
for salespeople who consult with customers on prospective purchases. In short,
it’s dangerous to predict what others will like, and faulty assumptions can
lose you a new job, a sale, or, yes, even a potential mate.
“When you’re observing another person’s choice, the choice
itself becomes very diagnostic,” says Barasz, who graduated in May and will
soon join the faculty of the IESE Business School in Barcelona. “It becomes the
only piece of information that you have to go on, and you kind of anchor to that
choice,” Barasz explains. “So once I see that you wearing a gray sweater, maybe
I assume you don’t like bright floral prints.”
For example, a real estate agent might forego the opportunity to
show a mid-century modern home to a client who previously showed interest in a
Tudor cottage. Or a bookseller might only recommend books that are similar to
previous purchases. Or if someone “likes” a fancy Manhattan hotel on Facebook,
her Facebook friends might not invite her on an overnight hike in the woods.
Or, more seriously, a physician might decide on an aggressive treatment plan
for a terminally ill patient, based on the patient’s previous
choices—neglecting to discuss more palliative options.
And if you knew nothing of Leslie John other than her guilty
pleasure of watching The Bachelor, you’d probably predict that she likes
watching its counterpart, The Bachelorette, too. But you’d be less likely
to predict that John also enjoys conducting social science research at Harvard.
FIVE EXPERIMENTS
The researchers conducted five studies to determine when and why
people assume that others are incapable of liking dissimilar things.
In the first study, 205 participants considered Facebook status
updates of a hypothetical friend, Joe Smith. Half the participants saw the status,
“Just booked a vacation! Headed to a lake.” The other half saw the less
descriptive post, “Just booked a vacation!” Those who knew where Joe was headed
were much more likely to decide that he didn’t like city vacations.
“When you like one lake, people infer that you hate cities,”
Barasz says. “And when you hate one lake, people infer that you love cities.”
In the second study, 297 participants read a scenario about a
consumer named Jane, who was choosing between two products, Widget A and Widget
B. Both widgets could be described by five attributes: price, size, shape,
function, and quality. (Participants received no information about what the
widgets actually were.)
As in the first study, participants were assigned to one of two
conditions. In the “similar” condition, they learned that the two widgets
shared four out of five attributes in common; in the “dissimilar” condition,
the widgets shared only one attribute. In both conditions, Jane always chose
Widget A.
Armed only with that information, participants had to predict
how Jane felt about Widget B.
The majority of participants in the dissimilar condition—61.5
percent—predicted that Jane disliked Widget B, compared with only 14.8 percent
of participants in the similar condition. And when asked what Jane would do if
Widget A was sold out, participants in the dissimilar condition were much less
likely to predict that Jane would settle for Widget B instead.
THE ACCURACY FACTOR
Having established that people don’t believe others can like
dissimilar things, the researchers conducted two studies to investigate the
accuracy of that belief.
In one study, some participants indicated a preference for one
of two movies, a five-star thriller or a five-star documentary, while an
assigned partner observed the preferences.
Next, the choice-making participants were faced with this
scenario: “Suppose that your first-choice movie is no longer available. Which
movie would you select instead?”
Most of the observers assumed—often incorrectly—that their
partners would prefer a lower-quality movie in the first-choice genre to a
higher-quality movie in a different genre. While 68.5 percent of participants
chose the five-star different-genre option for themselves, only 39.3 percent of
the observers predicted their partners would make that choice.
The observers did a better job of predicting their partners’
alternate preferences when the choices were relatively similar.
In the fourth study, when given the choice of two movies in
similar genres (e.g., an action-adventure flick and a thriller), 73.3 percent
of participants opted for the five-star movie in the alternate genre rather
than the three-star option in the preferred genre; 68.7 percent of the
observers correctly predicted that choice.
When considering sets of very different genres (e.g.,
documentary vs. comedy), the majority of participants—64.2 percent—still
preferred the higher-quality dissimilar choice over the lower-quality similar
choice. However, most observers failed to predict their partners’ preference
for quality over similarity: Only 18.1 percent predicted that their partners
would select the higher-quality dissimilar movie.
WHAT DRIVES OUR ASSUMPTIONS ABOUT OTHER PEOPLE’S BELIEFS
In the fifth and final study, the researchers
showed why people mistakenly assume other people dislike dissimilar
vacations, widgets, movie options, and so on.
The study began with 196 participants imagining that an average
consumer wanted to create a five-song playlist from five specific musical
genres. Their task: Predict how many songs the consumer would pick from each
genre.
Next, in a callback to the first study, participants had to
predict how much an unnamed person would enjoy a mountain vacation and a city
vacation, based on the knowledge that the person had enjoyed a previous vacation
at a lake. As the researchers expected, participants who designed the most
homogeneous playlists were also most likely to assume that the lake vacationer
liked mountains but hated cities.
“As our account suggests, people default to the belief that
others have relatively narrow and homogeneous preferences, and thus predict
that dissimilar items are disliked,” the researchers write in “The Role of
(Dis)similarity in (Mis)predicting Others’ Preferences.”
CORRECTING THE PREDICTION ERROR
So what should we do to correct this widespread prediction
error? For starters, we can be mindful of the researchers’ findings.
“From a firm’s perspective, you should never fill in the blank
about an option you don’t necessarily know about,” Barasz says. “Maybe you can
empirically show that people who drive Toyota Priuses don’t like Hummers. But
in more nuanced cases it might behoove you to just ask consumers, ‘Do you like
this thing?’ That way you don’t jump to any conclusions about it.”
It’s also important to keep in mind that people are likely to
jump to conclusions about you—in job interviews, for instance. “Imagine that if
in small talk at the beginning, which is seemingly innocuous, you divulge
something like ‘I like watching The Bachelor,’” John says. “You could seriously
undermine your prospects. Even though your enjoyment of The
Bachelor does not preclude your liking of high-quality content, Kate’s
finding suggests that the interviewer might infer that you’re shallow … that
you do not like intellectual programming, for example.”
Finally, the research findings provide a weapon against
self-doubt.
John recalls a recent conversation with HBS colleague Mike
Norton, who told her that she and another colleague had notably different
approaches to their work. “I was panicking because I thought, ‘Oh no, he can
only like one of us! He can only like one of our styles!’” she laughs. “But
then I remembered this research and thought, ‘Whew, so maybe he doesn’t hate me
after all.”
by Carmen Nobel
http://hbswk.hbs.edu/item/when-predicting-other-people-s-preferences-you-re-probably-wrong?cid=spmailing-13095677-WK%20Newsletter%2006-22-2016%20(1)-June%2022,%202016
No comments:
Post a Comment