News

Fienburg memorial lecture talks sensationalism and statistics

Sir David Spiegelhalter visited from Cambridge to discuss statisitcal reporting in the second Fienburg memorial lecture. (credit: Jacob Paul/Layout Manager) Sir David Spiegelhalter visited from Cambridge to discuss statisitcal reporting in the second Fienburg memorial lecture. (credit: Jacob Paul/Layout Manager)

“Numbers do not speak for themselves,” said Sir David Spiegelhalter, a world-renowned statistician, to a crowd of nearly 100 professors and students who had gathered in McConomy Auditorium to hear him speak. “The stories we tell, the way they’re packaged makes all the difference to their emotional impact.”

We are living in the age of data. Technology plays a bigger role in people’s daily lives than ever before, and many of society’s core issues can be reframed as statistical problems. Yet it is also the age of misinformation, and the field of statistics is facing more challenges than ever before.

Sir David Spiegelhalter’s lecture on Monday, April 22, was the second installment of the Stephen and Joyce Fienberg memorial lecture series. Stephen Fienberg, who died of cancer in 2016, was the head of the Carnegie Mellon statistics department. He was highly respected and influential in the field of statistics for his dedication to applying statistics for societal good. Joyce Fienberg, Stephen Fienbergs’s wife who perished in the Tree of Life synagogue shooting last Oct., worked as a researcher at the Learning Research and Development Center at the University of Pittsburgh and was a great supporter and friend of the statistics department at Carnegie Mellon.

“Steve sets a high bar for how statistic are communicated,” said Sir David Spiegelhalter. In his lecture, he focused on how statistics are manipulated for sensational and persuasive means. Spiegelhalter serves as a chair at the Winton Centre for Risk and Evidence Communication, which is part of the University of Cambridge’s department of mathematics. In his career, he has published over 200 papers and written several books. He has spent the last decade focusing on the application of statistics to everyday life, appearing in numerous T.V. specials on BBC and CNN to discuss issues related to probability and climate change. In 2014, he was knighted for his service to medical statistics, but he jokes that his greatest achievement is an appearance on the game show Winter Wipeout.

Spiegelhalter explained the pipeline of statistical evidence. Initially, data is collected by researchers and statisticians. Their research is then passed on to journals and the commissioners of the research, all of whom may have “selection biases.” Then the research is turned over to the press offices “who decide what to put out and how to package those numbers so that it will get coverage,” said Spiegelhalter. Then, journalists interpret the press release and write an article, editors slap a headline on, “and then finally we see it,” said Spiegelhalter.

“As soon as I see a story with statistics in it, on principle, I don’t believe it,” said Spiegelhalter. For a story to be published in the news, it must have passed through a cycle of distortion, sort of like a game of telephone, and he sees that as reason for skepticism. “The end result is this,” he added, motioning towards a graphic of the Daily Sun’s front pages. Attention grabbing headlines were printed in all-caps on each, including “Snoring raises dementia risk,” “Eat curry to beat dementia,” and “Snoring linked to Alzheimer’s.”

Spiegelhalter delved into multiple case studies of how statistics can be distorted. He showed one Scandinavian study that found a consistent association between high socioeconomic status and higher rates of brain tumors. He explained that this could be an “artifact,” an inconsequential or erroneous finding because “rich people get better health care and are more likely to be diagnosed with anything that’s wrong with them.”

But this caveat did not stop the press offices from publishing the study with the headline “High levels of education linked to increased risk of brain tumor risk.” News sources all over the internet then published articles with headlines that read “University causes brain tumors.” The audience in McConomy erupted into laughter.

To demonstrate how numbers are used to persuade rather than inform, Spiegelhalter deconstructed the Brexit Leave campaign’s infamous red bus ad that read “We send the EU £ 350 million a week.” This bus ad, which he called “nonsense,” is said to have won the Brexit campaign.

“It is a powerful emotional appeal,” but Spiegelhalter showed how the statistic is not exactly truthful. If you calculate that number per British citizen, of which there are 60 million, the number falls dramatically. If you calculate it per day, it’s about eighty pence, the price of a pack of cheese and onion chips. “So: we send the EU a packet of cheese and onion potato chips everyday,” said Spiegelhalter. “It doesn’t sound so good does it?”

At multiple points in the lecture, Spiegelhalter’s clicker malfunctioned. “No, no, no, no — too fast, too fast” Spiegelhalter said as the presentation rapidly progressed through dozens of slides. The audience laughed, and Spiegelhalter, whose sense of humor was on display, played along. “Oh, this is disastrous,” he said, and resorted to using his laptop.

In order to combat the misuse of statistics, Spiegelhalter and his colleagues at the University of Cambridge have been working with the press offices to disprove or mitigate the sensationalism of statistical findings before they cause public health scares. They successfully challenged a statistical finding about the danger of burnt toast in 2017. Yet, not every case can receive so much attention.

Spiegelhalter resolved that statisticians need to demonstrate trustworthiness by being honest about where uncertainty lies in their studies. Research has shown that audiences do not trust a range (between A and B) any less than a single result (A). For accountability, the background information of any study should be made accessible and readable.

Because of this responsibility, Spiegelhalter closed his lecture by calling for statisticians to be trained in ethics and communication, and to be more active in the public sphere.