The Uses of Small Data

Brian Kibby, the president of the Higher Education Group at McGraw-Hill, took to Huffington Post last week to praise Small Data.

"Big data" might be the most hollow, misused term in education. For all of the chest-pounding about how big data has come to education, how are colleges and instructors actually using it --now, in 2014? 

Kibby notes that we've long heard that Big Data will change everything. But in fact "true big data does not exist in education today." Instead of cheering or fearing how Big Data will save us or bring on doomsday, Kibby suggests that we look at how it might actually be useful right now. Big Data could be awesome, and Kibby is a fan in theory, but in practice there are justifiable concerns about privacy as well as a lack of technology that can actually collect, crunch and process the data back to an instructor in any useful way.

Kibby would rather talk about what we can actually do today, and what we can do today are targetted analytics: "things like assignment scores, time spent on the material, progress in an adaptive learning environment."

What we're talking about is short term use, small picture stuff. "These tools aren't paying attention to whether the student had Frosted Flakes or Cheerios for breakfast the morning before a test, but they're looking at what matters most." (I'm guessing Kibby's familiar with the Knewton video) Instead, small data can monitor things like exactly how well the student did on the homework, or how long he worked on it.

What is the upside for McGraw-Hill in the moderately-courageous, slightly-new world?

Odd coincidence-- I just saw the Undercover Boss episode about the chancellor of the University of California at Riverside. He watched a class of 250 do lecture question response with a clicker which allowed the instructor to immediately gauge how well the class was getting it. It's cool stuff. I've seen it in action. I would use it.

What conclusion did the chancellor draw? This is really cool tech that would work well to scale up into other large classes. With this kind of tech, everybody could teach classes of five hundred.

I've always maintained that classroom teachers already do massive amounts of data collection, far more rich and varied than what reformsters have been pushing at us via testing etc. What a teacher collects by looking at, talking to, interacting with, giving small quizes, informal assessments-- we collect a ton of data every day that allows us to develop rich and valuable student assessments that help us make instructional choices and adapt teaching to the individual needs.

But there are limits to that. Beyond a certain number of students, I just can't collect all the data. I can't watch 100 faces to see reactions. I can't informally verbally assess 100 students.

With small data systems in place? Hmmm. Cyber schooling has turned out to be a bust for all but a small sector of the student population. But what about a hybrid, somewhere in between the two extremes. A few hundred students still in a classroom with a live, actual instructor, but interacting through computer tech that allows him to collect and save response data from all those students.

I'm not sure how well it would work. But if it worked well at all, McGraw-Hill could sell a lot of materials, and Anywhere University could cut 50% of that pesky adjunct staff.