(This article has been published in The Financial Express newspaper on Monday, September 25, 2017)
There is a consistent failure in providing desired outcomes at the higher education level. This is indicated by a low employability ratio and the need for firms, which recruit from such institutes, to create a multitude of training programmes for their hires. Such institutes also face challenges in attracting teaching talent. Flipped and blended learning models can augment classroom learning, since these enable more discussion during the class time, by making available lectures and theoretical aspects of the lessons in a multimedia format. Such digital learning environments increase student-teacher and peer-to-peer interaction, thus providing a richer learning experience. But the most important benefit of digital learning environment is that it enables ‘learning analytics’.
A good start of an article or a blog post should straight away take readers directly to the content. However, for this post, I’m not following this basic rule of writing since the story I’m going to tell is completely nontraditional in all worldly sense.
Before I even come to the point let me throw some random stats. It’s not secret anymore that today’s world is highly digital. Social Media and Big Data are the most prominent trends of the 21st century. Increasing usage of social media and the internet in general has created a large world of digital data. 10 years ago “Digital World” was sized at around 130 Exabyte according to IDC and in next five year (2010) it increased to 1227 Exabyte. It was expected to grow 7910 Exabyte by 2015 and 90% of the data was created within last two years. Anyone who knows a little about big data and data analytics knows that data extraction techniques and tools have become much more sophisticated and efficient over the period. Companies are extracting and analyzing data from almost all internet interactions of users from emails to web browsing and from social media to machine sensors. We can already see that 20 years from now a presidential candidate will be answerable to the comments he posted on Facebook, and the articles he shared on twitter when he was young and didn’t really care about his social media persona at the time. Imagine how scary that would be.
Well, if that seems scary, there is a brighter side too. How about getting admitted to your favorite school not solely based upon your test scores and your essays but also based upon your social media activities? Scary and exciting at the same time, isn’t it? So, how long we’ll have to wait for this to happen? 5 years? 10 years? The answer is (-5 Years). Yes, you’ve read it right, there is a coeducational, nonsectarian private college named Ithaca College in Ithaca, NY that employs big data collected from social media for making decisions regarding admissions since 2010.
In 2007, Ithaca College built its own social network — IC Peers — for the prospective students. IC Peers is a virtual space for students of Ithaca College that is very similar to Facebook where students can create profiles with their pictures, ‘friends’ other in the system, join groups and interact with current IC faculty, staff and students in both wall or forum posts as well as in a personal messaging system. Instead of relying on Facebook, Ithaca created their own social network to be able to gather data which normally wouldn’t be available from Facebook.
In the first year of launch, 27% of all admitted students (2,223 Students) created an IC Peers account. This rate almost doubled by 2010 where 45% of all admitted students (4,270 Students) created IC Peers accounts. Two of the patterns that IC Peers noticed should be noted here. One of those is the fact that IC Peers itself had close to no effect on influencing student’s decisions. And another most important pattern which makes up this whole case study is that IC Peers proved to be an excellent platform to measure students’ interest and thus predict student enrollment. Collected data from IC Peers showed that students who were most engaged in IC Peers were most likely to enroll. In 2010, the yield rate (the proportion of accepted students who enroll) of the most active IC Peers users was 58.7% compare to 22.3% yield rate of minimally active users and 3.3% yield rate of inactive users.
The information generated from IC Peers’ usage data analysis encouraged Ithaca College to use it in the enrollment process. The college applied the usage of Social Networking Big Data to assess student interest to be used as ‘interest qualification’ for admission decision and to improve enrollment metrics. Since, the research at the college already indicated that the students who were more engaged in the recruitment process were more likely to be retained once they were enrolled; Ithaca College opened IC Peers for all applicant compare to its earlier open for only admitted student policy. IC Peers information combined with other measures of student interest like campus visit helped the college in identifying the students who were more likely to persist and complete their degree statistically. So, Ithaca College changed their selection process to have a more efficient selection of “right” student for the college. When two comparably qualified students are reviewed by admission committee, the deciding factor now would be “interest qualification” which is basically the interest that student has showed in engaging with the institution which was measured from Social Networking Big Data Analysis.
The success and failure of retention initiative could be determined only after years of data collection. However, IC Peers’ first year’s data showed some promising trends (Isn’t it easier to find data when everything is digital?). Ithaca College experienced a decrease in summer melt (the number of students who submitted a deposit in May but didn’t matriculate in the fall). Compare to 8.0% of summer melt from the previous year, Ithaca College experienced 6.7% summer melt. Ithaca also experienced an increase in First-to-Second semester retention from 95.5% to 97.5%. It should be noted that Ithaca had this First-to-Second semester retention at 95.5% rate for years and an increase to this number is a very big achievement. On top of that, the first year retention was increased from 84.2% to more than 87.5%.
“A student who doesn’t complete a degree is disruptive for everybody since student has a debt and schools have a hole in that class” is the sentiment of Katharine Frase who is a vice president and chief technology officer for IBM’s unit focused on working with the public sector, which produced the data analysis program used at Ithaca. So, now the dilemma is, does this simple idea of increasing graduation rates by using big data — to identify the students who are most likely to stay at the institute and are more likely to be successful — have any economic impact? Well, according to Bruce Poch(dean of admission and executive director of college counseling at the Chadwick School -a private school in southern California, and former dean of admissions at Pomona College), bond rating of an institution depends largely on ‘demand’ which for academic institution is the number of application the institution gets and the yield. Higher demand gives institution higher bond ratings and lower interest rates. With Social Networking Big Data Analysis employed in their enrollment process, we saw that Ithaca College was able to increase the demands. And thus, we can safely say that Social Networking Big Data Analysis indeed has an impact on the economy of an institution.
I can’t finish this article without mentioning that the gradual decline of Facebook’s growth over the last few years has frustrated technology predictors enough to show their concern that the social networks have already reached a point of saturation. However, if a network such as IC Peers, provides some value to users in order to enhance the admission process then the students can find value in the service offered by such networks and thus, they will engage more in the platform which will further improve the enrollment process.
One of the less talked aspects of the data analysis is, in fact, the one most important one, and that is ethics. Now, it seems Ithaca College is mining data from IC Peers in quantifying levels of student engagement. But, there is an availability of technology to perform text analysis that could enable institutes to be able to perform “sentiment analysis”. Sentiment Analysis is the one that provides a systematic understanding of students’ comments and questions as well as behavioral and individual actions. This sentimental analysis can be correlated to the enrollment decision to determine the yields. This would, of course, be a gray area both morally and ethically and so it forces both institutions and society to be vigilant to identify the fine line that separates logical evolution of data analysis and the violation of student’s privacy as well as trust. One cannot be skeptical enough to worry that there might be institutions who would — given a chance — try to cross that line and doing so, they would harm other players in the ecosystem — including those law-abiding institutions who just want to use data to improve both students’ and institutions’ success rate.