We are living in a time of unprecedented data accruement in all major fields of study such as those pertaining to origins of our universe performed at the Large Hadron Collider in Switzerland. Or just by general user data that’s amassed on a typical social networking website such as Facebook. Either way, there’s plenty of data that is being sent and received from the world wide web at this point in time.
Such sending and receiving also comes with it a now more evolved sense of data collection and analyzation, which is what many call “Big Data” or how data is tracked acquired and ultimately put to use to detect a series of patterns either in the form of text, images and AV components. Basically the simple way to put it is how people use and organize data to learn something either to improve society or a company. Colorado State actually offers a Computer Science course on the subject, which would certainly paint a better picture of what data mining is, and big data in general.
Essentially, big data is employed with the goal of quantifying data to improve upon a group of people more of than not. Most recently, education is relying on big data to analyze and predict how students will perform in order to generate a sort of roadmap to success.
This roadmap is typically compiled through comparison scores of students that share similar ACT Scores, and general grades in a range of subjects. An example of this being implemented in a widespread sense has been, according to a Scientific American article, most successful at Arizona State University (ASU) through the use of an adaptive learning platform called Knewton.
The platform, according to the New York Times, is what ASU has been relying on to effectively teach incoming students that are ill-prepared to take on the new challenges that college presents. This has been primarily utilized for math courses as of yet, but the course selection will be expanded upon during the Fall of 2016.
CSU already employs similar adaptive learning software, but on a much smaller scale outside of online classes and homework assignments. Most notably is McGraw-Hill’s Connect platform that employs smart E-texts and LearnSmart assignments that adapt student retention of reading through a series of questions that can be redirected back to the material that is found in the text. Another example of adaptive learning software at CSU is Pearson’s Math XL and MyLab that is utilized in Calculus 141. Either piece of adaptive learning software utilizes the same concept: help students to succeed by using an algorithm to compare their results amongst other students that utilize the software.
So far, the adaptive learning solutions above have shown that collecting data on student’s results relating to course material usage can be incredibly beneficial to creating a learning experience that works for a wide variety of students. It allows for an individualized learning plan that is meant to improve upon the success of the student as opposed to a sink or float kind of operation that most college students will either experience or witness during their undergraduate years.
However, big data in education has way larger prospects outside of adaptive learning when it comes to studying students either in the K-12 education system or in college. For example, they promise to give schools more accountability for the success of their students by looking at the importance of ethnic backgrounds, previous test scores and, in certain cases, a type of effort score may be assigned to a student to further add onto the predictive analytics of success.
LearnSprout, recently acquired by Apple, is a good example of this kind of approach. Looking at this from a college point of view, students are paying $30,000 minimum at a University like CSU for at least four years of education. Of course in spending that money, accountability relating to failure is still rested upon the students. One would think that telling students that there are plenty of resources on campus that they’re paying for such as tutoring, or by telling students to attend lecture would be plenty of accountability on the universities’ part, right?
Wrong. Simply put, big data would account for teaching habits of the instructor and the overall administration of content in a class as a whole.
Accountability would then fall on the instructor to review whatever concept wasn’t fully comprehended once again. This could also make students more accountable to come in during office hours for help, which could be better tailored through a profile that has been developed for the student specifically. These examples would be just one of many different ways that a university would become more accountable for the success of their students, as opposed to the old sink or float procedure.
Although, if this approach were to be implemented at CSU for instance, students would immediately be told where they would fail or succeed before even taking a class in the first place. This has its benefits of course, but would this restrict students from entering a major that they may interested in? Yes, most certainly students would be evaluated in a quantified sense over any type of situational analysis.
Certain students may have not have tried very hard in high school or could be terrible at taking standardized tests due to factors concerning test anxiety or some other predicament. Just looking at the numbers may not fully determine whether or not the student will be successful in a particular major. Sometimes the student may have impeccable persistence or work ethic that could allow them to succeed in a major despite what a predictive algorithm may say.
CSU currently employs a type of degree-interest program that will allow for students lacking the prerequisites for a certain major to work towards being part of that particular major or concentration. As annoying as this may be, it gives students an opportunity — despite academic issues in the past — to still work towards a degree that they may not be successful in but is where their interests lie regardless. With the big data approach applied, however, students would most likely not have a chance to pursue a difficult major and would be restricted to whatever majors or classes an algorithm says should be taken. This is where big data is lacking, next to privacy concerns of course.
In order for this predictive analysis of student success to work, there must be considerations of the human ability to persevere and overcome obstacles: an important factor to life.
Looking at students purely as data points to prevent failure by sticking them in easy classes won’t solve anything. Instead, it would be better to employ adaptive learning in a positive sense that would allow for anyone to still study what they’d like, while receiving the proper guidance and help to prevent failure. Students don’t just need to be more quantified, they need to be recognized for their individual strengths and attributes that would still allow for success despite what an algorithm may state.
Collegian Columnist Chad Earnest can be reached at firstname.lastname@example.org, or on Twitter @churnest.