A few days ago, I posted a blog entry about how the way in which a problem is framed can influence solutions. I'd like to extend that a bit. Although I'm usually a proponent of data-driven thinking, there are times when an over-reliance on data can also artificially constrain the search space for solutions.
The primary danger of data is that, particularly when dealing with humans, what is measurable is usually several layers removed from what is desirable. Even something as seemingly foundational as "learning" is hard to operationalize in practice.
In a university setting, we have "instructors" who "teach" students to help them "learn" things that they did not previously know. How would one measure "effective instruction" or "successful learning"? Do grades represent learning? Test performance? If you get a high score on the final, but then forget most of the material, did you "learn" anything?
Empirically investigating the impact of instructional decisions, student engagement, etc. is undoubtedly a useful endeavor. However, if we only collect metrics that are readily available, such as grades, allow them to exclusively serve as proxies for "learning," and focus our attention solely on interventions that impact those metrics, we may be artificially constraining the search space for useful solutions.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment