Chris Wild, Department of Statistics, The University of Auckland

We all know the universe of data, big and otherwise, is growing explosively—in volume, in the areas it reaches into, in how it is constituted, and in what you can do with it. By comparison, changes in what students experience are glacial. Statistics education is simply far too slow. The window we open up on this rapidly expanding universe is a tiny porthole. We have to find ways to get much further, much faster—to open our students’ eyes to an enormously wider world of possibility, opportunity, and excitement.

This is necessary to prepare our students for a rapidly changing world, but the self-interest of statisticians also calls for it. We need to adapt to prevent the data domain we have thought of as our own being completely overrun by species that are fleeter of foot (most notably computer scientists). Otherwise, we may expend our last breaths picking up a well-deserved Darwin Award.

This year, by agreeing to be my university’s MOOC guinea pig, I was given the space and production support to prototype an ambitious and radical re-thinking of intro statistics—a much-further-much-faster, more-data-more-quickly, introductory statistics course that emphasizes data visualization. This new approach was created as much for the statistical community as for the MOOC audience. We have called it “Data to Insight: A First Course in Data Analysis.”

Most of the content is delivered in 42 five-minute videos. The received wisdom from MOOC analytics is that viewership really starts to drop off at about five minutes. It is a daunting discipline (I am guilty of minor lapses), but I have been amazed by what you can get into five minutes if you “don’t sweat the small stuff.” It really puts the pressure on to convey just the essence of ideas efficiently.

The course has an eight-week, “three hours a week” structure and features just 30 minutes each week of instructional video. But it is not passive. It is a hands-on data analysis course that takes beginners and quickly gets them working with up to five numeric and categorical variables simultaneously in the same analysis (over the first four weeks and just two “teaching” hours). Students extract meaning from a 10,000-observation, 70-variable data set derived from a large observational study (NHANES) and tackle key questions of development and health using 30 country-level indicators over the last 50 years from Gapminder. For motivational reasons, it catches students up in the potential of data analysis for discovery before being mired in the “limitations” swamp. Serious consideration of systematic biases and confounding and random error (all crucial topics) does not come until the fifth week, which also forms the bridge to bootstrap confidence intervals (Week 6) and experimentation and randomization tests (Week 7). The course concludes (Week 8) with an unrelated, lighter bonus: exploring sets of time series.

Strategies Used for Acceleration

The most novel strategies used are the following:

• Being intensely visual and driving all argument off things you can see supplemented by metaphor
• Building software solutions that prevent “how do I get this out of the software?”, limiting the speed at which students can encounter new situations and new ideas
• Finding some powerful, conceptually undemanding “extender-capabilities” that immediately open much wider horizons

Other strategies are more obvious: limiting messages to just those most critical for real-world learning from data; stripping concepts back to their barest bones; and exploiting, feeding, and reshaping primary intuition. Additionally, we use vivid images (verbal and visual) to make key messages linger.

Can It Work?

What can you pack into a total of four hours of movies? Can it possibly work? Come and see. A page has been set up at 4StatEducators specifically for statisticians and educators. It lets you jump straight from a linked course outline to particular movies. You are encouraged to sample and “steal” ideas, because this MOOC is a prototype built for this community and a course designed to benefit its students.

Editor’s Note: Though the site is officially closed, it will stay open “silently” until December 20 for Amstat News readers. After signing up and receiving a login, you will have access for many months (instructions at 4StatEducators).