I ran across this paper (21 page PDF) from earlier this year while searching for something else, and it makes a comment that is striking: "a great majority of the existing dashboards (68%) are atheoretical in their choices of indicators and content to be shown." The result, say the authors, is that "without building on what is already known about learning and teaching and instead, using design- or data-driven approaches will likely result in ineffective or even deteriorating effects on learning. Unsurprisingly, several authors have already reported negative effects of existing LADs on learning." For example, dashboards need to know about the tasks learners are engaged on, they need to know about relevant standards and goal-setting, and they need to support metacognitive modeling. They should also incorporate what is known about motivation and self-determination theory. To account for all this, "the paper defines the model for user-centered learning analytics systems (MULAS). MULAS consists of the four dimensions that are cyclically and recursively interconnected including: theory, design, feedback, and evaluation." Don't miss this one.