CBC show exposing health abuses of major chain restaurants driven by U of T student research

Maybe it’s time to start brown-bagging it: one in four major Canadian chain restaurants has experienced a health violation according to new research from the University of Toronto showcased in an unsettling investigation by CBC Marketplace.

The study will be featured in "Canada's Restaurant Secrets" which airs Friday 11 April at 8pm ET on CBC Television.

It's the largest analysis of its kind, said CBC Marketplace Associate Producer Nelisha Vellani – examining nearly 5,000 health inspection reports from 13 popular restaurants including Swiss Chalet, Starbucks and The Keg in major cities across the country.

“About two million Canadians become ill after eating out every year, yet health inspection records can be hard to access and understand,” said Vellani. “Statisticians Michael Guerzhoy and Nathan Taback at the University of Toronto were key to the success of this ground breaking investigation, which spanned many months. Their data analysis allowed us to crunch the numbers to find out how these restaurants rank when it comes to cleanliness and safety.”

Guerzhoy pursued the work as part of a project in an applied statistics course led by Taback, which pairs students and their number-crunching skills with real-world clients and practical applications.

“With the help of U of T, Canada's Restaurant Secrets gave Canadians a new tool to make more informed choices when deciding where to eat out,” said Vellani. “The project has already spurred reaction, with one province making inspection records available for the first time and some restaurant chains promising to work to improve their records.”

Writer Brianna Goldberg spoke with statistical sciences lecturer Taback about the troubling restaurant data, the impact of statistics and what it’s like to have student work featured on a platform such as the CBC.

Tell us about the course that inspired this student project
The title of the course is "Statistical Consulting," a graduate level course offered by the Department of Statistical Sciences. My primary aim is to mentor students during the academic year who are interested in applying statistical methodologies to problems encountered in areas such as medicine, public health, business and environmental sciences. One of my major goals is to foster students' ability to communicate statistics to non-statisticians.

Students usually come into the course with a theoretical knowledge of statistics, with no experience applying that knowledge to tackle real problems. During the course they meet with clients that have statistical questions and problems, give many oral presentations to the class and present written reports to clients. In many ways, it's an apprenticeship as an applied statistician. I hope that students learn how to gain new knowledge on their own and communicate it to non-statisticians.     

How did one of your students’ projects end up featured on the CBC?
CBC contacted our departmental consulting service; we felt that this would be a great project for the consulting course so I met with the producer. Then Michael and I discussed the statistical approach he would use to rank the restaurants and he presented it to the class. (Photo at right by Jeff Golden via Flickr.)

After that point he worked on it with the CBC with very little supervision. I try to encourage students in the course to become independent thinkers that can weigh the advantages and disadvantages of various approaches to a problem; the course is based on real problems and data so there is no "right answer." Then at the end we wrote the report together.

Are students in this course ever surprised by what they learn working with clients?

Statistics students that are new to applying their knowledge to problems in different domains, such as public health, are often surprised how much background information about the topic is required to develop a solution that is feasible and will help solve a client's problem.  

At the end of the year I ask students to summarize all of their projects and tell me what they learned, and what they wished they had done better. One student's report noted that I should be aware he did not do any statistical analysis for any of his clients. He actually spent most of his time learning about his clients’ problems and advising on what analyses should be done, which didn't feel like a traditional "statistics" course.

I'm certain this course altered his perception on the application of statistics to real problems.

What do you wish people knew about statistics’ role in public health, and their lives?
I wish people knew more about the role statistics plays in public health, including its limitations. A lot of public health policy is based on statistical evidence. Statistics is a powerful tool for developing arguments based on identifying trends such as the dirtiest restaurant chain in Canada. But all arguments, including ones based on statistics, have strengths and weaknesses. It seems that people have a more difficult time evaluating the limitations of arguments based on statistical evidence compared to non-statistical arguments.