PHILADELPHIA — Why is a Wharton School professor publishing in neuroscience journals? 

The University of Pennsylvania business school is renowned for its many strengths, but one might assume a forte in brain science is not one of them. That assumption would be incorrect.

Abraham Wyner, associate professor of statistics and chair of the undergraduate program in statistics, is one of several Wharton faculty members to have contributed expertise to studies in life sciences. His analyses have enabled him and other researchers to find meaning in vast quantities of data, drawing conclusions that move science forward.

Wyner, a sports fan, has long generated and applied new statistical models to answer questions such as, what makes a baseball pitcher successful? Or, can a general manager of a soccer team “buy” a great squad? He has also applied his research to the areas of finance and computer science. But several years ago, he delved into a collaboration with neuroscientists at Penn Medicine, beginning a fruitful partnership.

Those researchers, led by Allan Pack, the John Miclot Professor in the Department of Medicine, investigate how genetics plays a role in sleep behavior using mice. To do so, they examine a mouse’s sleep stages by implanting sensors in the animals’ brains that track electrical activity. Wyner and colleagues offered assistance in interpreting the resulting data to distinguish between bouts of sleep and bouts of wakefulness.

More recent studies have done away with the surgically implanted sensors and instead relied on videos that track the movements of mice to differentiate between their sleep stages.

“My contribution to that was to generate methods of automating tasks that previously required an enormous amount of time and money and were also quite invasive,” Wyner said. “It’s a nice collaboration, and we actually studied and made new statistical techniques to model the behavior of sleep.”

These techniques have helped break new ground in sleep studies, and Wyner and Pack have been authors on several papers, including two this year, in the journals PLoS ONE and Sleep. Looking ahead, the researchers plan to bring engineers into the fold to find better ways to automatically “score” videos of mice sleeping.

Another project to which Wyner applied his expertise in data collection and analysis ended up causing a stir well outside the campus community.

An undergraduate who had come upon a rare set of historical temperature data approached Wyner for his help interpreting it. As he began to examine the records, he realized that the temperatures labeled “daily maximums” were actually collected at 3 p.m. each day.

“Historically they just looked at the thermometer at 3 o’clock in whatever location they were in and they recorded that and they called that the maximum temperature,” Wyner said. “So I started fooling around with the data.”

Examining modern Philadelphia weather records, Wyner found that, on average, the daily maximum temperatures were nearly a degree higher than the temperatures at 3 p.m.

“What this said to me was that we should expect to see a lot more maximum temperatures happen recently than historically because we’re measuring [temperatures] continuously, and we didn’t always do that,” said Wyner. “So this light bulb went off in my head, and I said, wait a minute, people who are studying temperature, in particular climate change, have got to have dealt with these statistical problems.”

But when he began reading climate analyses, he saw that in many cases they did not deal with these problems in what he considered a statistically sound manner. That led to a climate-data investigation of his own, undertaken with Blakeley McShane, Wyner’s former graduate student and currently an assistant professor at Northwestern University’s Kellogg School of Management.

Moving from the student’s dataset to the larger world of climate science, Wyner and McShane analyzed historical temperature proxies, such as data from tree rings and ice cores, which climate scientists have used to estimate temperatures that occurred before records exist. Their paper, published in 2011 in the Annals of Applied Statistics, puts forward the controversial conclusion that these proxies appear to not to be reliable predictors of global historic temperatures, throwing into question the often-cited “hockey stick” of global temperatures that indicates warming in recent decades.  

Wyner says the paper drew praise from statisticians and cast him into notoriety in some climate-scientist circles. Yet he emphasizes that the study’s findings do not mean he does not believe in contemporary climate change; rather, he sees the results as wake-up call for climatologists who have not previously used rigorous and transparent statistical analyses to support their conclusions.

When he spoke on the subject of climate change analyses during a seminar for Penn’s Biology Department, Wyner says he was well received.

“There are actually a few geologists I’m talking with about working together in the future,” he said.

With new technologies to capture biologic and other scientific data, from the movement of cells to variations in sea level, Wyner says that scientists can get overwhelmed with data.

“All of a sudden people turn around and look and say, ‘Oh my gosh we have an incredible amount of data, what can we do with it?’ And then they realize maybe they need a statisticians or a mathematicians or a computer scientist.”

Often, Wyner is happy to help. Though, to really capture his attention, a problem must pose a challenge and offer him an opportunity to invent new techniques to analyze data.

“I do have interest in science qua science, as opposed to statistics,” he said. “But I’m still looking at new methods and new discoveries. That what we research statisticians like to do.”