What’s So Hard About Data Analytics?

by Silence DogoodThe data in the following table, which comes from the MnSCU website, shows the FYE enrollment from FY07 through FY13 for St. Cloud State University:

If you add the Ug FYE (undergraduate full-year equivalent) to the Gr FYE (graduate full-year equivalent) you get the total FYE shown in the Table below.

The data is plotted in the Figure below.

Please note that these are FY labels so FY13 is the 2012-13 academic year. This Figure does NOT include any information for FY14, which is the current fiscal year.

These data show that enrollment peaked in FY10 and then declined slightly in FY11 (a drop of 121 FYE), which correspond to a drop of 0.80%. These data also show that from FY11 to FY12 enrollment dropped 1,037 FYE, which corresponds to a decrease in FYE of 6.93%. For the FY12 to FY13 comparison, enrollment dropped 885 FYE, which corresponds to a decrease of 6.35%.

Using data from MnSCU, the FYE data for Summer 2012 (1,069) and Fall 2012 (6,366) when compared with Summer 2013 (1,012) and Fall 2013 (6,017), shows enrollment is down 406 FYE for the same period for the prior year, which corresponds to a decrease of 5.46%.

The important question is: How can we predict, using data analytics, the Spring 2014 FYE enrollment? If we know the spring enrollment, we can then calculate the total enrollment for FY14. The administration has projected a decline of 4.7% for spring FYE. If enrollment drops 4.7% from Spring FY13, the enrollment for Spring FY14 would be 5,353 FYE. This would yield a total projected enrollment for FY14 of 12,382 (summer 1,012, fall 6,017, projected spring 5,353), and represents a drop of 671 FYE, which corresponds to a decrease of 5.14% from FY13.

Is this a reasonable prediction? The reason why accurately predicting enrollment is so important is the concomitant budget considerations.

The faculty has never been told how the administration has come up with its’ enrollment projections so we are only left to wonder. However, here are two alternate methods of predicting the Spring 2014 enrollment.

The following table shows the fall and spring FYE enrollments from FY07 through FY13.

If the number 5,353 (generated by the administration’s assumption of a 4.7% decline from Spring 2014), is entered into the table for the Spring enrollment, the value for FY14 would give:

If you look at the declines in enrollment from Fall to Spring occurring the last three years, where the declines in overall FYE appear, the average decline in enrollment from fall to spring semesters is 10.6%. So projecting an 11.0% decline might seem reasonable. However, from the data it is clear that the rate of decline is increasing, so it might be more reasonable to extend the rate of decline based on a linear regression for the data for the past three years. Performing this linear regression, the predicted rate of decline would be 12.4%. Using a 12.4% decline* generates a spring FYE enrollment of 5,271.

This translates to an annual enrollment for FY14 of 12,300 (summer 1,012, fall 6,017, projected spring 5,271), which corresponds to an annual enrollment decline from FY13 of 753 FYE, which corresponds to a decrease of 5.77%.

This might be where the story ends. However, a significant effort was made to increase the enrollments in the Senior to Sophomore (S2S) Program this fall. Rumors say the S2S enrollment is up over 4% from FY13. Clearly, this would inflate the fall enrollment number (leading to projected spring enrollment that would be too large because S2S students don’t register in spring like regular students). As a result, the total enrollment projected would be too high.

Another way to predict enrollment for FY14 that would not be biased by the increased S2S fall enrollments would be to look at the total enrollment for the past three years of decline and extend that trend into the future. Performing a liner regression of the FYE data for FY11 through FY13 gives a total enrollment of 12,066 for FY14. Interestingly, the correlation coefficient is 0.998 indicating that the data is nearly linear.

The total FY14 enrollment will almost certainly be larger than 12,066 FYE predicted because the additional S2S students in fall were not accounted for in the projection based on a linear regression of the total enrollments. However, it is easy to determine a correction factor to account for the 4% increase in the S2S enrollment. Based on the total number of S2S students and projecting a 4% increase will lead to an increase of approximately 30 FYE. As a result, the overall enrollment for FY14 would be 12,096. This number yields an overall drop of 957 FYE in FY13, which corresponds to a decrease of 7.33%.

The administration is predicting an enrollment of 12,382. Using the data from the decline from fall to spring semester yields a prediction of 12,300. Finally, using the enrollment trend for the past three years with an adjustment for the increase in S2S enrollments gives 12,096. So how do you pick a number?

Back in February, the administration predicted an enrollment decline of 2.4%. In April, this increased to a decline in the range of 2.8-3.2%. The administration says in May they increased their prediction to a decline of 4%. Finally, in September, the administration has now said the enrollment will be down 5%. If the administration’s predictions are plotted as a function of time, by the end of the year, the decline will likely be even greater than 7.33%!

Using the administration’s projected number of 12,382 for the FY14 enrollment, SCSU’s enrollment will have decreased from the high of 15,096 in FY10 by an astounding 18.0%. If the number 12,096 is used, this four-year decrease is 19.9%. Why is all of this important? Using the administration’s data, a decrease in enrollment of 1% results in a budget reduction of over $620,000. So, if the enrollment is down by 7.33% (more than two percent larger than the ‘finally’ budgeted number), the university would be looking at a loss of at least $1,240,000. If enrollment continues to drop, pretty soon we’ll be talking about real money.

Technorati: **St. Cloud State****, ****MnSCU****, ****Declining Enrollment****, ****Silence Dogood**

A very compelling argument. The data tells a convincing story of what is really going on. An enrollment decline approaching 1/5 of your student base is alarming.

“Using the administrationâ€™s projected number of 12,382 for the FY14 enrollment, SCSUâ€™s enrollment will have decreased from the high of 15,096 in FY10 by an astounding 18.0%. If the number 12,096 is used, this four-year decrease is 19.9%.”

Let’s see 19.9% less students is about a $45,000,000 loss. (3,000 students X $15,000, tuition/fees, books and room/board). Gosh now we are talking real money!

I’m bothered by this analysis – yes, there could be some reason for concern with a simple look at enrollment. But so many mitigating factors exist. First, it appears that enrollment over the entire analysis period drops by only 7.8%, not the nearly 20% that seems to be the focus here. What was the enrollment prior to this point? Is it possible that enrollment was artificially high in 2010, rather than artificially low in 2013? Why was it higher in 2010? What about student success – high enrollment isn’t directly related to student success, and in fact might be inversely related (poorer student achievement results in longer graduation times and retaking of courses). A quick look shows that, indeed, graduation rates in 2010 were significantly lower than in 2012.

Also, couldn’t one come to the conclusion from the many postings here on lfr regarding SCSU, that the cuts that they made recently show a prudent conservation of resources?

My comments aren’t made in defense of the administration – surely there are criticisms to be made. However, the obsession over a blunt, generalized measure (enrollment equals budget) doesn’t seem to hold up under any sort of scrutiny, and reading through the various sources available (including the union’s meet and confer notes) shows that postings like this, made by knowledgable insiders, aren’t serving to further the discussion.

So Yeager, the only way to “further the discussion” as you say is to not offer any opposing comments. Spoken like a true liberal!

I have no problem with programs being closed if the process follows established rules, procedures and policies and not to satisfy personal agendas.

The reason enrollment gets so much press (the FTE or FYE, not headcount) is that it is a definite benchmark used by so many in higher ed. Student success is a vague term, what you say is student success I may differ with you and describe it differently.