Wednesday, June 04, 2014

Garbage In; Transparent Garbage Out

Bill Fitzgerald:

If you are building or working with a data system, it needs to have these two components:

  • Student dashboard - a student should be able to see everything that is collected about them. More importantly, the application should have a mechanism that allows students to comment on, review - and in some cases, remove - data points, or assumptions based on data. More on this later.
  • Parent dashboard - because many students are minors, parents have the legal right to review data collected. Additionally, rights to review some data is guaranteed under FERPA. Really, there shouldn't need to be much - if any - difference between the student and parent dashboard. If there are significant differences between a parent and student view, those differences should be grounded in clearly articulated reasons that are of direct benefit to the learner.

We can't lean on this approach too much, particularly if you're talking about big data online learning platforms. There is too much raw behavioral data, they can't practically let you check your answers without making all the questions and answers public, and their conclusions based on the data, particularly ones just used in their internal model, may just be garbage anyhow. If your online learning platform concludes you're a spatial, abstract sequential learner with a 43.2 grit quotient, who likes Dark Tearjerkers Featuring a Strong Female Lead and a high level of distortion on the electric guitar then... ?

No comments: