Recently in Action analytics Category

The Death of Why?

I am fascinated by learning analytics and what it can, and can't, answer for higher education. A key concern, for me, is asking questions of our data that lead to meaningful answers. I don't care about the average time on task for students interacting with an online module if it doesn't lead to a meaningful understanding of how students learn that topic or a meaningful intervention for students who are struggling. I am committed to understanding "why" and always want to know "what can I do now that I know this?"

So I was taken aback by this article sent to me by a higher education colleague. The focus is on business but it touches on the same issues we have with big data in higher education. Big data is big, really big, and it takes considerable time and expertise to sort, understand, and find meaning. The author argues that it is possible to apply "analytics to massive, detailed data sources to identify what works without having to worry about why it worked." (emphasis in the original). Can this be true for higher education? Can we glean meaningful insight into what works without understanding how it works?

Higher education, unlike a point of sales transaction, has a long tail. Some of the why's we deal with when looking at success in college might have more to do with prior preparation, previous experience (or lack of experience) in the subject matter, etc etc. We can't change some of those factors, which is why this article has me thinking. If we can glean a best practice from data only, from seeing exactly what works (most of the time) and never understand why, does that matter? Is the "why" a distraction? I am not convinced it is - ultimately, for education policy, it is important to know what kinds of k-12 education experiences correlate to success in higher education, e.g. But maybe, in the medium term, it is okay to leave the "why" aside.

Individualized e-learning by data mining

Boy with laptop computer.jpg

Have you ever wondered how Google, Facebook, or Amazon recommends you something or show online ads that are very close to your interests? As many people know, these online services heavily gather and analyze user data including previous visited websites, friends network, keywords entered, and so on. With the data, they customize user experience accommodating each user's interest and need. This is how they make money.

If commercial services can do it, why not higher education? Similar efforts are emerging in higher education, especially in an e-learning field. According to a news report from Inside Higher ED, the University of Phoenix, a big for-profit higher education institution, announced at the 2010 Educause conference their ambitious "Learning Genome Project", which they hope to revolutionize online learning by individualization.

According to Angie McQuaig, director of data innovation at the University of Phoenix, the Learning Genome Project is "building a new learning management system (or LMS) that gets to know each of its 400,000 students personally (i.e., infer students' details from their behaviors in the online classroom) and adapts to accommodate the idiosyncrasies of their learning DNA."

For example, if students learn better from watching a video than reading a text, the system will feed them more videos. If a student is bad at interpreting graphs, the system will recognize that and present information accordingly.

While it sounds great, the project is just a conceptual framework for now. However, if the project comes to true, it may provide significant benefit to students and may be better than traditional offline learning in terms of accommodating individual differences. It is very difficult for instructors to meet individuals' different learning styles in an offline class of 20 to 30 students.

There are, of course, challenges. First, it would be very expensive and difficult to build the learning management system. Second, the privacy issue will be huge. One could imagine how people would worry about and want to protect their personal data. McQuaig later said in an interview with Inside Higher Ed that the University of Phoenix will let students choose how much information they submit to the system.

Despite these challenges, it seems that some other higher education institutions will follow the University of Phoenix in order to enhance their online learning and student success. In a near future, we may be able to see individualized e-learning become popular and its impact on educational achievement.

Purdue releases course management and retention tool

Purdue University, in partnership with SunGuard Higher Education is releasing a course management system called Signals. Signals was initially tested at Purdue and developed by its associate vice president of Academic Technologies, John Campbell. Like other course management software, it provides space for electronic grading and disseminating course materials, but Signals goes further. Signals is a student retention program, designed to designate struggling students early on in courses, allowing instructors and other academic resources to reach out and provide support.

Signals works by allowing both students and professors to monitor progress and success in a certain course. Color-coded signals--red, yellow, and green like stoplights--indicate a student's risk level for failing the course. The students see these signs whenever they log onto the course's website. Depending on the signal, it offers suggestions and resources. For example, a student doing poorly in chemistry might be reminded by the program of a tutoring program available for schools. The program also reminds professors of their students' progress and gives them options to offer help and insight. For examples of how Signals works, check out this presentation. In addition to the information professors input like grades, Signals also has access to previous student information and grades. It also integrates into existing BlackBoard technology.

In this new story on the software, one professor with 900 students over three lectures praised the software, particularly it's early detection. Signals starts tracking students by the second week of class. Often professors must wait for the first major assignment to realize a student is behind. Overall, professors at Purdue have praised the program. It will be exciting to see how other universities integrate this product, especially with the University of Phoenix moving towards personalized course management (check out TEL Blogger Michelle's post about that here!).

For more information on Signals' release, check out these new stories:
NBC Nightly News Report
"Signals" help studnts stay on track

Using CMS reports for data mining

Colleges Mine Data to Predict Dropouts -

John P. Campbell's data mining project at Purdue is featured in the Chronicle this week. I first heard about this project at Educause Midwest in 2007; the project is much further now. Campbell and his colleagues are able to notify students when the data indicates the student is struggling, and students are responding.

About this Archive

This page is an archive of recent entries in the Action analytics category.

Innovation is the next category.

Find recent content on the main index or look in the archives to find all content.