This reminds me of what Bodong has discussed in the Chinese paper on the difference between "analytics" and "analysis". If we can treated "analysis" as an objective, pure technical process, or maybe we cannot, "analytics" can never be objective. For any research, what is considered as evidence, how to collect evidence, and how to interpret the results, either the quantitative or qualitative, are far from objective or pure technical.

I had an assumption that learning analytics mostly use data collected from online learning environment. This may not be true, but it is possible that online teaching can change people's mindset of what learning is, and how learning should occur.

Assessment that is grounded in epistemological assumptions on what is knowing and how to know, and pedagogical practices (how to teach?).

Is this why so many variables about students were used in the regression?

Does this mean the theory was supported by the evidence here? Students' self-report strategy align with what they actually do?

Learning analytics really can help provide a better description of students' behavior, in which couldn't be measured by survey or other self-report methods. But it seems there can be research performed on how what students actually do can be related to what they reported they were doing.

How to decide what methods or how many methods we can or should use to do a temporal analysis? The decision should be based on the research question or the norms in this field?

I was curious as whether if analyzing good teaching practices, can we reveal the principles that can guide novice teachers to become better at their teaching considering how to organize learning activity and when to transit?

It is so amazing how the technologies enable us to do.