At-Risk Students and the Data Cookbook: Harnessing the Power of Prediction

A recent New York Times article explores the burgeoning world of student data in an attempt to explain why it holds such power, particularly with at-risk students. Recent work in learning analytics suggests students who earn a grade of less than an A or B in a foundational course may actually be revealing academic problems in the future. What is most interesting about this research is not that poor performance can lead to future problems, but that the foundational course is not major-dependent. Poor performance in early curriculum courses like English composition or mathematics have an impact on future academic success in any major. Also interesting is that other courses like history can be “powerful predictors.”

Schools may heed this research and begin to analyze student records to confirm these claims for themselves. Better yet, schools may begin to form interventions with students taking introductory or foundational courses to provide additional mentoring and support. The problem is the workload needed to do such analysis: records are rarely available in a clean enough format to simply “crunch the numbers.” The Data Cookbook helps harness the definitions needed to accurately compute course progress in various programs. The Data Cookbook could quite possibly turn a 3-week, labor-intensive task into a 3-hour project.

This research has its own foundation in the logic that student data is, in itself, a gold mine of potential. Most institutions agree that they have plenty of data, but not the resources to do much beyond recursive, descriptive analysis. In other words, they lack resources to mine the gold. Computing such analyses is difficult and time-consuming, but the research says meaningful intervention can be a key factor of student success. This is all to say current models of intervention for “at risk” students are being redefined. Previously, poor grades in foundational courses were not perceived as important indicators. Poor grades in the key courses within a major was the indicator. With the right use of this data, schools can be poised to intervene and provide support to make sure such outcomes never materialize.

Predictive analytics are understandably attractive to parents and politicians alike, all of whom are publicly vested in funding students’ educations. However, such “powerful predictors” are only realized when the data can be formatted, analyzed, and executed meaningfully. How quickly could a school produce a report detailing the performance of individual students within foundational courses? An individual report about one foundational course on campus (just one, not the dozens or hundreds of potential courses and sections!) can take many hours of data cleaning, analysis, and sense-making without the proper structure, definitions, and specifications. The Data Cookbook, however, provides this very thing: the structure needed to harness large amounts of data, definitions to clarify nuances in the data, and specifications to make sense of sometimes conflicting data points. In short, the Data Cookbook is a good tool for data housekeeping so that when exciting research like this emerges, schools can quickly test it for their own purposes and then act on it efficiently.

We are very interested to hear from you in the comments, particularly concerning the methods your institution uses to identify at-risk students. If you would like to join the Data Cookbook Community or learn more about the Data Cookbook to improve your reporting needs, please visit:

James E. Willis, III  &  Brenda Reeb

Leave a Comment