MCDONALD: Google, please help improve the mean

The data revolution is coming. While online marketing is getting fat off trillions of cookies, biology is exploding with DNA sequences, sports teams are selecting players who are undervalued by traditional metrics and public health researchers are mining queries to Google and posts on Facebook and Twitter to follow and predict disease outbreaks. Even disciplines typically devoid of data dumps, like the humanities, are getting in on the action: e-book libraries allow literary critics to follow specific words over time.

It’s becoming clear that our brave new world requires us to understand numbers and make sense of data, regardless of our field — but are we prepared to do so? Probability and statistics are the branches of mathematics that cover the important tasks of prediction and interpretation, yet few Americans take a full probability and statistics course in high school.

Arthur Benjamin, a colorful ‘mathemagician’ and mathematics professor at Harvey Mudd College has championed the idea that statistics should replace calculus as the focus of high school math. In a 2009 TED talk, Benjamin went so far as to claim that “If all the American citizens knew about probability and statistics, we wouldn’t be in the economic mess we are in today.”

I’m confident (95 percent) that the quants and financiers responsible for the 2008 financial meltdown had taken some probability and statistics courses, so I don’t think that requiring a single high-school course would have been a panacea. And as someone who was awestruck by the world of derivatives and integrals, I find it a bit unfair to jettison calculus for the cold, hard practicality of statistics. But Benjamin makes a good point. Even though I’m in a scientific field, I have never used calculus for any of my experiments. It’s my half-semester of college probability and statistics that I use and wish I had taken more of.

Without engaging too much in the pure math versus applied math debate, I believe probability and statistics should be standard, ideally at both the secondary and collegiate levels. Realistically, even if every American high school student is required to take an introductory stats course, I doubt they’ll remember the difference between a Gaussian and binomial distribution 15 years later. We need a deeper mode of numerical education that highlights examples of probability and statistics in our everyday lives. How about having every K-12 class, regardless of discipline, begin or end with a probability and statistics “problem”?

Is this practical? Probably not. Many teachers will be able to work in a topic-appropriate, ripped-from-the-headlines query that teaches a valuable lesson about statistics in three minutes, but I realize it might not happen for everyone. That’s where data-driven Google could step in.

In terms of essential skills for the 21st-century, being able to search for something online is probably number one — even above understanding the t test. With this in mind, Google created a daily search puzzle, Google-a-day, to help people develop good search skills (and no doubt, for Google to learn about how people search). By the same token, Google may be able to fill a gap in modern education by providing its own daily probability and statistics problem. They could kick it off with a doodle (the famous Google logo decorations) celebrating the March 23 birthday of Pierre-Simon Laplace, a French mathematician and pioneer in the field of probability. Google should then also advertise this service to schools to encourage teachers to begin incorporating probability and statistics into their daily lessons.

Outsourcing education to private companies wouldn’t normally be my first choice, but it’s in the interest of all parties to raise kids who are excited by data and enjoy playing in the ‘big data’ sandbox. Adults, not just students, could dip their toes in as well, helping America deal with the deluge of data inundating almost every industry.

The Google stat problem-a-day won’t be a complete fix, obviously, but I bet that it’ll be statistically significant.

Jessie McDonald is a sixth-year graduate student in the Department of Immunology. Contact her at jessica.mcdonald@yale.edu.

Comments

  • inycepoo

    Americans these days can barely handle reading, writing, and arithmetic, let alone Gaussian distributions and p-values. That’s an exaggeration, obviously, but still somewhat valid nonetheless.