55. Creating simple, but useful and inexpensive, homegrown predictive algorithms (Data)

We all have a pretty good idea of the data points that signal a student may be less likely to persist or retain. RIGHT?<br/><br/>Yet, many campuses have purchased expensive software solutions that promise sophisticated data analysis to predict student retention. And, while we know that these programs work well, we have found that many of these third-party models are not significantly more accurate than homegrown predictions we can produce with easily accessible data.<br/><br/>Each year, our campus reviews a handful of key data points for all our incoming students and creates an internal score that guides our outreach and support efforts. While this activity lacks complicated statistical analysis, we have found that our process predicts retention at a surprisingly high level.<br/><br/>Don’t over complicate things. Get a group of smart people around a table and come up with the 10 data points that you think will help predict whether or not a student will retain. Create a score for each student and then go after those who need your support!