Do Charter Schools Outperform Public Schools In New York City?

On August 12, New York City released the test scores for the 2015 3-8 common core tests in math and ELA.  The Daily News reported that in the public schools, 34.2% of students met the math standards while 30.4% met the ELA standards, which was up by 1% and 2% respectively from the 2014 tests.

I remember a few years ago when the new harder tests were rolled out and many ‘experts’ predicted that scores would take a drop initially but then as teachers ‘adjusted to the new standards’ the scores would eventually increase until they approached the pre-common core levels, which were about double the current scores.  This belief was fueled by the bedrock of the entire school reform program that low test scores are caused by lazy teachers who can’t be fired.

Some of the top scores were recorded by the Success Academy charter chain.  Success is having a rally next week as part of their #dontstealpossible campaign funded, in part, by Families For Excellent Schools (FES).  They want the city and the state to raise the charter cap and to give more money to charters.  Implied in this campaign is that charter schools are, in general, outperforming their public school counterparts.

For years, charter school critics, like me, have been demonstrating that many of the highest performing charters serve different populations than their ‘failing’ neighboring public schools.  The charters often have higher attrition, higher suspensions, fewer high needs students, and fewer students eligible for free lunch.  For a while the charter school defenders denied these accusations, but more recently some charter cheerleaders have been admitting this and even celebrating it as Michael Petrilli put it as ‘a feature not a bug.’

On the 2015 state tests, charter schools outperformed public schools in math with 44.2% meeting the standards while also doing worse than the public schools in reading with 29.3% meeting the standards.

To put these numbers into context, I crunched the numbers and summarized the results in a graph.  For each school I took the average of their math and ELA scores.  Then I took the most recent numbers for the school’s ‘Economic Need Index’ which includes the free lunch percent along with some other factors.

With graphs relating percent of free lunch to test score proficiency, there is always a strong negative correlation, as most people know.  The thing I wanted to see was if the charter schools had a higher percentage of ‘outliers’ than public schools.  In a sense, this is a bit like the coveted ‘value added’ measure that reformers like so much.  A school that is above the trend line would be a school with a greater than average value added.

In the graph below, the light blue dots are the public schools, the red dots are the charter schools, and the green triangles are the Success Academy schools.

It seems pretty clear to me that, on average, the charter schools are not outperforming the public schools, based on how about half of of the charters are above the trend line and half below.  Also it is relevant that most of the charters have an economic need index between .7 and .9 while there are a significant number of public schools that have an economic need index above .9.  This runs contrary to the charter school supporters who continue to insist that charters serve the ‘same kids’ as the nearby ‘failing’ public school.

Success Academy are such outliers that I can’t understand why charter supporters who are so focused on test scores are not out there insisting that all charter school resources be sent to expand Success Academy and the ‘yesterday’s news’ charters like KIPP, Democracy Prep, Harlem Children’s Zone, The Equity Project, etc. get shut down for poor performance.

For anyone who wants to check my numbers, the public school ELA data came from here, the public school math data came from here, the charter school ELA and math data came from here, and the economic need index (the most recent available was not from 2015, but from 2013, but this should not change the results very much) came from here.  Because the economic need index and the test data came from different years, there were some schools that I had test data for, but no economic need index for those schools so those are left off of the graph.