Basics on School Enrollment Methodology

This post comes from Kellie LaRue who knows her way around a spreadsheet AND statistics.

One item not directly in her post is her thought that for some schools that are underenrolled but also ended after-school childcare. Why not bring it back? Many parents would be ecstatic to have more choices. 

Kellie's Post

Greetings to the new folks trying to interpret SPS enrollment information. Here’s a little bit of information on how read (and not read) SPS’s crappy enrollment charts.

This is written mostly for the new people as it is far too easy to presume that SPS is using common vernacular when they are not. The chart can be found on page 27 of the January 25, Budget Work Session presentation.  Here is the version that was in the Seattle Times article. 

Basics on School Enrollment Methodology
SPS is using a common methodology that is used by the vast majority of school districts. As such, this information may be helpful for Bellevue families who are also trying to make sense of enrollment based closures.

Most school districts consider this methodology to be an enrollment projection. However, it is much closer to a stability analysis. It is most certainly NOT a modeled projection.

The presented enrollment data is based on three data sets:

* a rolling two-year average
* a rolling five-year average
* a rolling five-year average from five year's prior. 

Now to be fair, school enrollment tends to change slowly over time, so this methodology does work fairly well for many districts and in many circumstances. However, it does not work well, when there are large or swift external changes.  (Demographics, a pandemic, local employment shifts, major changes in assignment plans, or major change education options, services, etc)

This methodology completely failed Seattle when Seattle was one of the fastest growing cities in the US for a decade. This methodology also fails when you have a hundred year pandemic.

It should be self-evident that a two-year rolling average during a pandemic is not a very robust data set. Likewise, a five-year rolling average loaded with pandemic data is also not a very robust data set.

But here we are. This methodology is the common methodology and I have no reason to believe that the presented charts are based on anything else. This presumption is based on the simple fact that it would take time, energy and money to do additional modeling, as well as an outside consultant.

How NOT to Read the Crappy Charts
SPS enrollment chart presents “high, low and middle scenarios”. These labels are both redundant and meaningless. (Seriously, no middle line, needs to be labeled “middle,” ever.)

As the labels are so redundant, a reasonable person would easily read the chart as

high - optimistic - decline stops, stabilizes and maybe some growth
middle - status quo - no real change
low - pessimistic - continued decline

Do NOT read the charts like this. These are not “scenarios” they are rolling averages.

How to Read the Crappy Charts
Since the labels are meaningless, you need to do a bit of detective work to read the charts.

The Low number
I suspect that the “low scenario” is the 2-year rolling average. That number is extremely low with a 7,000 enrollment drop over and above the 3,600 pandemic era enrollment drop.

For the low projection to be accurate, SPS would continue to lose enrollment at the same rate as the prior two years. This seems highly improbable. That said, the low number of 42,000 is an extreme number that could drive a lot of the pressure to close schools.

The High number 

I suspect the “high scenario” is the five-year average from 5 years ago. This number is pretty stable and enrollment was fairly stable during the 2012-2017 school years.

This a very probable scenario if families returning to in person school, balances out overall lower birth rates.

The Middle number
I suspect the “middle scenario” is the most recent five-year average. This number shows continued enrollment decline. This scenario is also pretty suspect as two of the five years used for the average incorporate the pandemic enrollment drop.

Notes on the years represented in the chart
SPS is terrible about identifying the data sets used for any chart, on the chart itself. As such, one needs to be cautious any conclusions without validating the source data.

The chart does include the actual 2022-23 headcount. As such, it would be easy to presume the 10-year projections were updated to include this year’s current enrollment.

However, I strongly suspect, that is not the case. SPS generally updates the 10 year projections in March. SPS routinely makes big decisions about assignment plan and budget in January that is based on information from the prior enrollment year. My hunch is that this the normal process using outdated info.

The 2022-23 enrollment was about 300 students higher than projected, if those numbers were included, the decline of the low number should have bottomed out a bit more. On this chart, the slope of the low decline is nearly identical to the slope from the two-year pandemic decline.

That said, because of the law of averages, these charts are going to continue to be somewhat meaningless for the next several years. AND, knowing the data set is critical information and should always have been included on every chart.

The Bottom Line
These “projections” should not be the sole basis for school closures and/or concluding that SPS is looking at ten years of enrollment decline. The middle and low “scenarios" are highly suspect as they are extrapolated from pandemic enrollment shifts.

A far more nuanced approach is required. At a bare minimum, I would want to see two additional lines on this chart that would exclude the pandemic years.

A recent five-year average that excludes the pandemic, for example, using the 2015-2020 average. 

A “no change” or “straight roll up” number from the 2022-23 school year.

Best of luck to all the communities that are starting to organize.