Suspensions — 1
What does the frequency of suspensions tell us about New York’s schools?
This is the first in a series of articles that will look at data about suspensions in New York schools.
We will explore the data together — you as readers and reviewers and me as quantitative spelunker. There is a lot to learn here. In my first quick look, I’ve found that location, teachers’ training, school staffing, class size, students’ ethnicities, student attendance records, and financial need all influence suspensions.
As we go, I hope this series will also demonstrate the ways data visualization and machine learning work as complements. Too often, academics aim to just test a hypothesis or schools want to just measure outcomes. By mixing techniques, we can discover so much more.
Background
Suspensions are a popular subject matter. Here are some things we know:
- To no surprise, students who are suspended have lower test scores and graduation rates, sometimes as a direct result of their suspensions.
- Race is a factor. Black students are more frequently targeted for suspensions and receive longer suspensions for the same category of behaviors. Men are suspended at higher rates than women. Special education students receive more suspensions, too.
- Legislation in NYC, San Francisco, and Los Angeles has been making it harder to suspend students beginning in 2015. Schools in NYC report a decline in the use of suspensions.
- At the same time, some think tanks, policy makers, and educators worry that reductions in the use of suspensions might hurt school climate and ignore opportunities for “restorative” justice.
For additional research, I recommend this article in the American Journal of Education by Elizabeth Chu and Douglas Ready. For recent news in NYC, see Alex Zimmerman’s prolific coverage of these topics on Chalkbeat .
We can also assume something more direct:
Suspensions must add stress to school administrators — who can be required to defend their decisions. Certainly, suspensions can create severe stress for students and parents.
Whatever we learn, the frequency of suspensions in a school should be a reflection of educators’ (teachers and administrators) philosophies and the learning environment. When we study suspensions, we are examining something basic about schools’ culture and climate.
Our Tools
We will look at data published in 2017 by the New York State Department of Education. Unless I state otherwise, I am using data the “2016–2017 Report Card Database” covering (surprise!) the 2016 — 2017 school year.
This data includes pre-K through 12th grade public schools in New York State. After data cleaning, we have data for 4,178 schools covering 2,387,564 students.
How Many Students Do Most Schools Suspend?
On average, NY public schools have 3 suspensions annually for every 100 students. Not much!
This histogram (below) groups and then counts schools by the number of suspensions they have for every 100 students.
Each bar represents an increment (of 0.37) in the number of suspensions. The first bar as we move from left to right shows 1,338 seemingly tranquil schools with under 0.37 suspensions for every 100 students. The second bar shows 402 schools with 0.37 to 0.74 suspensions.
Another helpful way to view this information is through the cumulative distribution:
60% of schools have less than 2 suspensions for every 100 students and 80% have less than 6 suspensions for every 100 students.
On the other hand, the corollary is also true: 20% had more than 6 suspensions for every 100 students. Many schools have 20x more suspensions than others!
Our data shows the number of suspensions, but not whether 1 student was suspended 10 times, for example, or 10 students were suspended once. We will have to work around this for now.
How Common are Suspensions at Different Ages?
The first thing we might wonder is how grade-level affects these distributions. Let’s take a quick look.
I expected middle schools, when students have a reputation for being combative, to have higher suspensions. Nope…
This chart is a swarmplot. Each point represents a single elementary, middle or high school. The points are placed on the correct level of the y-axis (“Suspensions per 100 students”) but allowed to sit anywhere from left to right that fits in the “Elementary school”, “Middle school”, or “High school” buckets .
You can see that the “High school” swarm is shifted slightly upwards: high schools tend to have more suspensions. The swarmplot is similar to the this modified box plot (to the left), showing progressive divisions of the data by 50%.
Elementary schools have a median of only 0.46 suspensions per 100 students. Middle schools are up to 2.58 suspensions and high schools are at 5.28.
Some Schools Suspend a Lot of Students
Looking at these distributions can miss important dynamics. Although the typical school has 0 to 10 suspensions for every 100 students, what about the extremes?
In 2016–2017, there were 26 schools that had more than 30 suspensions for every 100 students. Here is some information about this cohort:
We might have expected to see small schools in the outliers. If a kindergarten with only 12 rambunctious kids suspended 4 of them one day in 2016, that school would instantly qualify for this high-suspensions group.
But actually, these are large schools. The average middle school had 459 students enrolled and the average elementary school had 685 students!
In 2015, there were eight schools that reported a seeming impossibility: they had more than 100 suspensions for every 100 students. Notably, none of these schools posted the same numbers in 2016 as suspensions gained greater public notice.
These stats raise pose a mystery: How did schools even sustain suspensions on this frequency — in many cases, more than one per calendar day? What kinds of suspensions (in-school, out-of-school at suspension centers, or at-home) are these?
We can see one thing immediately: the schools are overrepresented in the Buffalo City and Syracuse school districts.
In particular, there are 7 schools on this list in Buffalo. Buffalo has one of the larger school districts, but it is smaller than NYC Districts 10, 31, and 8 and similarly sized to NYC Districts 2, 27, and 11 — all with zero schools on this list.
We know that something must be operating differently in Buffalo and Syracuse than in NYC. We don’t yet know what. To be continued…
