Much has been said about the high demand for charter schools in entrance lotteries, but little about the choices families make once their children are actually accepted. Looking at attrition rates is important because they may tell us about the choices parents make about their children’s schools. Attrition may also indicate that some students are encouraged to leave or have been expelled. And, if proficiency rates rise over time in schools with heavy attrition, those percentages may reflect the change in the student cohort, as well as or instead of academic progress.
I took a look at changes in the size of student ELA testing cohorts for the thirteen middle school charter s that have more than one year of data between 2006 and 2009. Since all students in the testing cohort must take tests every year, I was able to determine whether cohorts shrink or grow over time.
As it turns out, high-performing charter middle schools in the New York City also have extremely high rates of attrition in their testing cohorts :
- Eight of the thirteen schools have enough data to allow us to examine cohort size between 5th grade, when students enter, and 8th grade, when they graduate. In four of these schools, more than 25% of the students vanished from the cohort. Of these four schools, three saw cohort declines of 30%, and one lost nearly 40%. All of these charters have been nationally or locally acclaimed as great schools that are in high demand. The average attrition for this group of eight is 23%. (charts follow.)
- These attrition rates contrast starkly with what I found in regular public schools, where the size of cohorts tends to remain the same or rise. (charts follow.)
- The remaining five of the thirteen schools are too new to have testing cohorts that span all grades. However, most have high attrition through the grades already completed. Taken as a whole, only two of the thirteen middle school charters show no shrinkage in the size of their testing cohort, thus approaching parity with public schools citywide.
- As the testing cohorts shrink, the percent of remaining students who are proficient rises dramatically. Seven of the thirteen middle school charters have more than 20% attrition in their testing cohorts. All seven have proficiency rates that rise to over 90%, with an overall average of 94%. That rate drops 21 points for charters with less than 20% attrition. (charts follow.)
- Information is not made public regarding the academic proficiency of students who vanish from the cohort. What I do know is that dramatic rises in the percent proficient seem to parallel the rates of attrition in the testing cohorts . I have charted these in the latter part of this post.
Left or Left Back?
Clearly, testing cohorts are shrinking as proficiency rates rise. A question remains, however, as to whether students are leaving the school altogether, or being retained in earlier grades.
While either is possible, most of the schools under review have been in existence long enough to track not only the size of the cohort that the student is leaving, but also the size of the cohort he or she would enter if left back. If students are being left back, then their entrance into the cohort of the lower grade should be reflected in the size of that cohort. That cohort might grow, for example. What happens instead, however, is that those cohorts too are generally shrinking as students move up in grades.
Since the cohorts into which the vanishing students would be assigned are themselves shrinking, retention seems unlikely to be the major factor in cohort attrition.
However, if attrition is the result of student retention in the grade , then the sharply rising passing rates in these schools seem best explained by the elimination of struggling students from the cohorts.
A Closer Look at Four High Attrition Schools
Let’s take a closer look at attrition in the four schools that lose more than 25% of their students from the testing cohort between their first year in the school (grade 5, 2006) and their last year (grade 8, 2009).
The chart below lists the schools, their total attrition, and the proficiency percentage points gained across years:
|ELA Cohort Attrition Grade 5 to 8 (2006-2009)
|Williamsburg Collegiate||72 to 44||-39%||+35|
|Harlem Village Academy||57 to 39||-32%||+43|
|Leadership Village||59 to 41||-31%||+24|
|KIPP STAR||82 to 61||-26%||+27|
The graphs below aggregate the data in the chart above.
Next let’s compare the charters to neighborhood and similar schools, and then look at the possible relationship between proficiency and cohort attrition.
Williamsburg Academy has 39% attrition: Meanwhile the neighborhood schools as a whole (CSD 14), and the similar school within the neighborhood (K582), have cohorts that do not decline. Overall they grow.
Harlem Village Academy has 32% attrition. It is in District 5 where the neighborhood schools as a whole and the similar school within the neighborhood (M317) have cohorts that do not decline. Overall, they grow.
KIPP STAR has 26% attrition. Like Harlem Village, it is in District 5, where neighborhood schools as a whole as well as the similar school within the neighborhood (again, M317) have cohorts that do not decline. Overall, they grow.
Leadership Village has 31% attrition. Meanwhile the cohorts of the neighborhood schools (CSD4) do not decline.
As students disappear, the high-attrition schools record ever higher percentages of proficient students. This same pattern plays out in all 13 schools. Schools with more than 20% attrition see their percent of proficient students rise to over 90%. We do not know if struggling students are in fact leaving the cohort. We do know that cohorts are shrinking, and rates of proficiency rise.
Here are graphs of our four schools with testing cohort attrition of more than 25%. Both attrition and percent proficient are able to be graphed together because the values in each scale are similar.
I reviewed the 15 charter schools the New York City Department of Education (DOE) identifies as middle schools on its Progress Reports. I excluded two of these schools from our analysis because they were new and therefore have only one year of data (I needed at least two years of data to follow a cohort). The two schools I eliminated are Bedford Stuyvesant Collegiate and St. Hope Leadership Academy.
I used the DOE’s data set for ELA test results for the 2006 through 2009 school years to determine cohort size and proficiency rates. I also examined the Math data which showed cohort attrition that was highly consistent with the ELA attrition.
Some schools have as many eight cohorts with more than one year of data between 2006 and 2009. In this report I examined the cohort of students who were in the school’s highest grade level (usually 8th grade) in 2009. Rates of attrition from the schools’ other testing cohorts is highly consistent with the attrition found in the cohort I examined.
- Appendix One shows ELA attrition and proficiency in all thirteen schools for the cohort that I reviewed.
- Appendix Two repeats the information in Appendix One and adds the next cohort.
- Appendix Three shows the citywide aggregates for both cohort size and proficiency rates for the fifth grade (2006) to 8th grade (2009) ELA testing cohort.
Finally, I did not include the UFT Charter School in my review because it is a K-8 and I have focused only on schools designated as middle schools by the DOE. For the record, however, the UFT middle school cohort has significantly less attrition than the schools that are the subject of this study. One UFT middle school testing cohort has 15% attrition. The other actually grows.
All data can be found on the website of the New York City Department of Education at schools.nyc.gov.
 I considered all charter schools that have been designated as middle schools on the New York City Progress Reports.
 This is the 2006-2009 cohort charter middle schools typically begin in 5th grade, whereas regular public schools begin in 6th.
 Opportunity Charter, with a rate of 15% proficiency in 2009 has been eliminated from this average because its population of about 50% special education students makes it unlike the other schools. Its inclusion would lower the proficiency rate.
 A second “similar school” is M367, a new school which does not have a three-year cohort. Its first group of students declined from 81 to 77 students (a 5% decline) between 7th and 8th grade.
 The DOE has not identified any similar schools in Community School District 4.