Research Lead: ‘The Strength of Weak Ties’ Then and Now, Show Me Your Kale-Face, R.I.P. to Labels ‘Millennial’ and ‘Gen Z,’ and More

You’re reading the Research Lead, a monthly digest connecting you to noteworthy academic and applied research from around the behavioral sciences.

Oh, kale no!

When I (Evan) was a kid, about five or so, I loved Popeye—the cartoon sailor man with ridiculously large forearms and a passion for spinach. In a violation of unspoken kid law, I loved spinach too. My brother Max was around one at the time, and I convinced my mom he needed to get on the Popeye diet. My mom dutifully bought a jar of the green stuff. As the spoon-turned-airplane made its approach, my brother’s face contorted, it did not have permission to land. It was the only food my brother spit out as a baby. 

Max would empathize with the participants of a recent study that explored the taste of leafy greens, albeit they were a bit younger. Recently, a team of scientists wondered how and whether fetuses near their due date, between 32 and 36 weeks, experience taste. To figure it out, they exposed fetuses to the flavor of either carrots or kale, via a pill taken by the mother. Then, they used an ultrasound to observe the real-time facial reactions of the fetuses.

Certain combinations of these facial movements were of particular interest—they combined to form a “cry-face” or a “laughter-face.” If you’ve ever eaten carrots or kale, you can probably guess which caused which. They found that fetuses exposed to kale made a “cry-face” more frequently than those exposed to carrots or to nothing, whereas those exposed to carrots were the most likely to make “laughter-face.” [Psychological Science]

An example of a kale-induced grimace (left) and a carrot-induced grin (right). Source: Ustun, et al. (2022)

Say R.I.P to generational labels like “millennial” and “baby boomer”

Cruises. Napkins. Soap bars. Diamonds. Homeownership. What do these have in common? They’re all things millennials have been accused of killing. Not to be outdone, upstarts Gen Z apparently recently tried to send the thumbs-up emoji to the gallows. Boomers, of course, helped kill things like the environment and unions.

But nevermind, because you shouldn’t pay attention to these generational labels. Why? Because they don’t categorize human behavior all that well. They’re so poor, in fact, that Cort Rudolph and Hannes Zacher decided to pen a tongue-in-cheek obituary to generational labels. 

“Generations are now a ‘big business’ that helps to sell books, talks, and workshops run by organizational consultants who peddle their expertise in these flawed concepts,” they write. Instead, they advocate viewing people through a lifespan perspective: “This approach conceives of aging as a continuous, lifelong process rather than splitting it into distinct groups to represent generations.” [Group & Organization Management, open access]

“Hey Siri, I love you: People feel more attached to gendered technology”

Gendered technology has become an increasingly popular topic, and many people are warning that the trend risks perpetuating gendered stereotypes (e.g., the prevalence of “female” virtual assistants, like Siri and Alexa). Some are heeding the pushback. Others haven’t changed much. A team of researchers investigated how consumers interact with and relate to gendered technology to understand whether companies might be incentivized to keep it. 

They began by analyzing Amazon reviews for a classically anthropomorphized product—robot vacuums. They find that participants who used gendered language in their review, by referring to their vacuum as “she” or “he” for instance, used more attachment language and rated the vacuum more positively than those who don’t. In follow-up studies, they found that assigning a gender to a virtual assistant made it feel more human, which in turn increased feelings of attachment. This was the case regardless of the gender of the product. Male and female products were equally desirable, and both more so than genderless products.

The authors conclude, “Our findings imply that consumer-goods companies gender their products because doing so is beneficial … Our sense is that the benefits of gendering technological devices are accrued primarily by the companies that sell them while the costs (i.e., reified stereotypes) are shared by society at large.” [Journal of Experimental Social Psychology]

REDISCOVERY: “The Strength of Weak Ties” (1973)

In his 1973 paper “The Strength of Weak Ties,” sociologist Mark Granovetter illustrated the importance of “weak ties” to social networks. Traditionally, networks had been defined as a person and their contacts. Granovetter argued a network should be defined as a person, their contacts, and their contacts’ contacts. 

One of Granovetter’s observations was that a person with more weak ties could have a richer network than someone with only strong ties. Every additional indirect contact a person has makes external information more likely to enter their orbit via the weak ties that bridge social groups. However, he explains, “The fewer indirect contacts one has the more encapsulated he will be in terms of knowledge of the world beyond his own friendship circle.” [American Journal of Sociology, open access]

Granovetter describes his original research into the importance of weak ties in the job market. 

The Strength of Weak Ties: 50 Years Later

Fast forward five decades to a team of researchers who recently investigated how Granovetter’s claim about the strength of weak ties persists in the modern-day job market. In 2015 and 2019, LinkedIn experimented with different versions of its “People You May Know” algorithm. Some versions caused users to form more weak ties, and others caused more strong ties.

The researchers capitalized on this variation to investigate how tie-strength influences job applications and “job transmission,” or the number of instances where two users (A and B) who worked at different companies connected, and one year later, user B reported working at the same company as user A (or vice versa).

They found “that moderately weak ties with low interaction intensity—measured by the number of mutual friends between two people—increased job applications and job transmissions the most, whereas strong ties—measured by both the number of mutual friends and interaction intensity—increased job applications and job transmissions the least.” [Science, open access]

Some research on overconfidence might be a little too confident

Previous decision-making research suggests that “belief distributions” are a way to reduce overconfidence. To capture a belief distribution, researchers present a decision maker with the entire range of possible outcomes and ask them to assess the probability of each. However, Beidi Hu and Joseph Simmons were worried that inferring a participant’s level of confidence from a belief distribution may misrepresent their actual confidence-level, as the nature of the task might compel people to assign probabilities to outcomes they actually believe to be impossible.

A belief distribution looks like this—the panel on the left represents what a belief distribution should look like, the panel on the right represents the concern of the authors:

Source: Hu and Simmons (2022)

In line with previous work, the authors believed that, in principle, belief distributions should reduce overconfidence by forcing people to seriously consider options they may have glossed over. However, in a series of ten studies using alternative measures of confidence, they found the opposite. “Providing belief distributions sometimes had no effect but usually increased confidence,” they write. “It seems that providing a belief distribution often convinces people that they were, in fact, right all along.” 

It turned out that participants were allocating probabilities in a way that reinforced their initial beliefs—essentially generating their own evidence to support their original predictions and turning them into more overconfident decision makers. These results came as a surprise but confirmed the authors’ concern that “what appears to be a remedy for overconfidence may instead be a methodological artifact.” [Journal of Experimental Psychology: General, open access]

Measurement and survey design resources

J-PAL (The Abdul Latif Jameel Poverty Action Lab) recently put together a repository of measurement and survey design resources. It includes questionnaires, relevant studies, datasets, lectures, and more. It covers a number of domains, including energy and the environment, financial inclusion, housing, and health. [J-PAL]

The Behavioral Science in the Field Course: Collaborations between universities in the U.S. and East Africa

“The Behavioral Science in the Field Course is a collaboration between MIT GOV/LAB and the Busara Center for Behavioral Economics, to train graduate students from the U.S. and local universities in East Africa in cutting-edge behavioral science research. Conducted in Kenya, the course is structured as an intensive deep dive into interdisciplinary behavioral science and provides students the opportunity to develop novel behavioral science games to answer research that will result in data collection.” Read about the experiments and the experiences of the first dozen students who each conducted their research in Kenya. [MIT GOV/LAB]

Disclosure: Busara provided financial support to Behavioral Scientist as a 2021 organizational partner. Organizational partners do not play a role in the editorial decisions of the magazine.