Self selected survey data does not make for good LAs.
While surveys are a great tool to find how users feel, or what they think (or believe they think), they are not very useful in finding empirical data about users. Yes, this has been touched on in other Ideas, but survey data is a specific example of what people believe learning analytics solely is, while in reality, it is so much more.
EDIT: Didn't realize I left that last sentence incomplete. My apologies.
Social Web