#LAK12 examples of 5 levels of sophistication

Goldstein's Academic Analytics: The Uses of Management Information and Technology in Higher Education (2005) mentions "five levels of sophistication of use" of academic analytics.

1) transaction only: i.e. "How many students logged in to Course Y in the last 24 hours?" You would run a query on the learning management system but it would only happen on command.

2) "analysis and monitoring of operational performance": An illustration of might be: "What is the trend in Mediasite (lecture capture) usage throughout the semester?" This specific question is especially appropriate if your resource is limited to a certain number of concurrent users, for example. You can do this analysis with the previous level if you download into Excel, create graphs, etc. but it is easier if your data tool provides an automatic visualization.

3) "What-if decision support, such as scenario building": I can't even think of an example of this. Please comment and give me an example if you know of one. I'll give the next one a try...

4) "Predictive modeling and simulation": An illustration of this might be "How should classroom scheduling practices have to change 4 years from now if we know that we have X students in the Class of 2015 with certain types of characteristics?"

5) automatically triggering a business process. This would be akin to a Walmart-like situation wherein when a customer buys a stock item with a certain SKU, stock levels of that SKU are checked and compared with rate at which that SKU has been purchased that week and average time to reship from vendor, then depending on those results, an automatic order is sent to the vendor. I think this is what LA aspires to. For example, student doesn't log in to the CMS for a certain number of days and depending on their demographics or the time of the semester, student services or the course instructor sends out an exploratory, "Hey, how's it going? How can we help?" email, which is most likely automated as well.

Note that illustration #4 offered above tend to be more "academic analytics" use, which I interpret as more institution-based, i.e. how can we run this institution more efficiently? Illustrations #1 and 5 are more "learning analytics" focused, which I interpret as learner-based, i.e. How can we help an individual or class of individuals succeed academically?

The article further mentions that central research administration is one of the organizational units least likely to use analytics, which is rather telling. Here's a possible use: strengths of association between academic departments or individual research faculty are analyzed through bibliographic citation data, or in subject tags for their publications (like MeSH headings, for those of you who use PubMed). Individuals who are found to be highly associated in terms of the domains in which they work could be potential collaborators, joint/bulk purchasers, etc. It would be especially valuable for larger institutions which are spread out geographically, and for central research administration which sometimes are not aware of these types of links. Cost savings would be realized, and the "team science" concept could be implemented. I'm sure this is one of the purposes of systems such as Harvard Profiles and VIVO, which allow visualization of connections between researcher and departments.