A Paradigm Shift in How Scientists Study Kids

Posted by on February 15, 2024 10:01 am
Tags:
Categories: Everything Else

There is an open secret in the study of child development: Most of what we think we know about how babies develop is actually based on a specific subset of kids—those born to families from Western, educated, industrialized, rich, and democratic (a.k.a. WEIRD) nations. The acronym was first coined in an influential 2010 paper to describe the wildly unrepresentative populations that many psychology studies have long relied on. This is an issue in the field generally, and certainly a thorny problem in developmental psychology, which primarily studies children: According to one paper, WEIRD subjects make up 96 percent of the data used in published developmental-science studies but represent only 12 percent of the world’s population.

As a result, it’s hard to be certain whether many things we think we know about babies’ development are truly universal elements of human nature. It means that we tell an incomplete story about the process of our own becoming. Yet the problem has remained hard to fix. Even within the U.S., similar demographic biases have arisen: The families that most often participate in research studies tend to be white, affluent, and highly educated. The type of parent who brings their baby to a study typically lives near a university, many of which are located in cities, and has the resources and free time to travel to a lab and wait. “Some labs can book a single baby for a day” to collect one data point, Elizabeth Bonawitz, a cognitive scientist at Harvard, told me.

The upheaval of the coronavirus pandemic, however, provided an unexpected opportunity. In spring 2020, Laura Schulz, a cognitive scientist at MIT, and collaborators released a tool called Lookit. At the time, in-person studies were hard to do and research groups had started running online ones instead. Lookit recruited families, connected them with institutions that needed subjects, and hosted virtual studies—including game-based experiments, surveys, and video interviews. Last year, the site merged with Children Helping Science, or CHS, a virtual bulletin board (also co-founded by Schulz) where researchers can advertise studies they need online participants for. Today, CHS has enrolled more than 8,000 children for studies spanning more than 200 labs in all 50 U.S. states and on multiple continents.

The basic technology that underlies CHS is straightforward: a combination of video capture, messaging, and gaming interfaces. In a typical experiment, a child might play a computer game devised by a researcher, the subject can be recorded during play, and both in-game responses and the video are reviewed by scholars later. The platform started in 2013 as a side project for Kim Scott, then a graduate student in Schulz’s lab, but it wasn’t an easy sell to many academics. “Some people still have attachments to doing developmental science in front of a child in their lab, controlling the environment,” Schulz told me. The pandemic meant that scholars had no choice but to relinquish some of that control.


Online experiments might have grown out of necessity, but they help address two of modern developmental psychology’s core problems. First, not enough children participate in experiments in general, so researchers are less likely to identify rarer or more subtle behaviors in them. Second, the WEIRD issue: When experiments consider just a slice of the world’s children, can they really claim that their conclusions are universal?

Take the famed marshmallow study of the 1970s, which offered preschoolers either one marshmallow immediately or two of them if they could wait. The study ultimately suggested that children who delayed short-term gratification in favor of a bigger reward had better outcomes later in life. But the original study was both small (32 children) and demographically specific (all were students of Bing Nursery School at Stanford University). Subsequent attempts to replicate the experiment found the effect diminished or absent altogether. In 2020, researchers even demonstrated that for children from low-socioeconomic-status backgrounds, snapping up that treat immediately could predict future success. In unstable environments, “it may be more effective for you to just go ahead, when you have an opportunity, to take advantage,” Candice Mills, a developmental psychologist at the University of Texas at Dallas, told me.

[Read: Why rich kids are so good at the marshmallow test]

Other developmental processes that scientists long thought were universal, such as language acquisition, can be affected by one’s environment too. For years, scientists believed that children gained language through one-on-one interactions with adults, but in an island community in Oceania, children largely learn from one another.

Scientists have tried various methods to tackle the field’s biases. The Stanford psychologist Anne Fernald, for example, traveled in an RV to a low-income community in Northern California in order to collect data on how children learn language. But this took time and money that not every experimenter has. In recent years, broader movements within academia at large—including Open Science and Big Team Science—have embraced sharing data among research groups and collaborating on studies. And in the developmental-science world, tools such as Databrary (a video and audio library) and CHILDES (primarily a repository of language transcripts) help scientists use existing data for new studies.

CHS is an extension of these efforts. Elena Tenenbaum, a clinical psychologist at Duke University, is studying younger siblings of autistic children, who are up to 17 times more likely to receive a diagnosis of autism compared with the general public. Yet this population is a particularly difficult one to bring into the lab. “These families that are already stretched thin from their appointments for their older child—if they need to come into the lab, it gets really challenging, really quickly,” Tenenbaum told me. With CHS, researchers can test this group—for example, measuring how many words they know or whether they can pay attention to and remember faces—to see if early hints precede more obvious symptoms of autism without needing them to come to a lab.

But the implications are bigger than any one study: Online testing tools have the potential to use technology to understand the whole child. For instance, one kid could participate in different studies at different labs—for example, to do with language, or motor skills, or causal reasoning—all connected through CHS. “How does change in one ability relate to changes in another ability?” Schulz said. “We’re going to get a much, much better window into a developing child.”

Crucially, the platform also has the potential to broaden geographic and ethnic sample diversity within the U.S. For instance, data supplied by CHS show that 13.1 percent of its subjects are Latino and 5.5 percent are Black. Many researchers don’t record the demographics of their in-person study subjects, but CHS’s figures are a striking increase from numbers reported by a survey of top journals, which estimated that more than 90 percent of subjects are so-called convenience samples—in other words, people who live near a university or research center, who tend to be white and affluent.

Even with a tool like CHS, developmental psychology still needs to reach more international, rural, and low-socioeconomic-status communities. Most of the world’s children are growing up in Africa and Asia, some of them “in rural settings, very often with some access to electricity, but not necessarily a tablet or easy access to internet,” Alejandrina Cristia, a linguist at the École Normale Supérieure in Paris, told me. And domestically, for CHS to expand further, researchers may need to bring laptops to recreational centers and libraries, Lisa Oakes, a developmental psychologist at UC Davis and an early CHS adopter, told me. Melissa Kline Struhl, the executive director of CHS, hopes that improving the platform’s functionality on smartphones will widen its reach too.

Indeed, forming truly universal theories of how children develop was never going to be an easy task, and still has a long way to go. Yet a shift to online studies is helping provide one thing that the smaller, less representative samples of the past couldn’t: kids who don’t typically come to university labs. For developmental psychology, that alone is a vital step.

Leave a Reply

Your email address will not be published.