The opening paragraph of this paper
by Dawson et al. neatly summarises a major weakness with learning
analytics - that the data gathered is gathered incidentally rather than
with pedagogical intent.
The obvious question to ask is 'what data would be more useful?' and then 'how can we collect that data?'
Learning Analytics is based on the premise that the answer to the first
question is 'information about the interactions between learners' based
on the observation that knowledge is increasingly distributed and
learning has become less about learning knowledge from a 'wise sage' and
more about connections and collectively held knowledge.
second question - how can we collect that data? - presents a problem.
It is not difficult to track forum contributions or similar within an
institutions VLE. The interactions can be automatically tracked and the
length, time of and words within those posts can be classified and
codified but the assessing the quality of engagement requires human
input. This is merely the first issue: most interactions between
students don't happen within the VLE. However slick an institutions VLE
is it is unlikely to be as intuitive, familiar and easy as platforms
like Facebook and WhatsApp. Students will opt for easy for them over
helpful for the institution.
The idea of any institution monitoring and analysing my Facebook and WhatsApp conversations is creepy!