What’s creaming, and why does it matter? This topic gained some momentum earlier this week in a comment by Seth Andrew, the head of the Democracy Prep Charter School, a relatively new 6-12 secondary school in Harlem, on a GothamSchools post on KIPP. Pointing to the use of standardized tests for admission to New York City’s citywide specialized high schools and citywide gifted and talented programs, he wrote, “traditional public schools are far more guilty of ‘creaming’ (both in terms of aggressiveness and quantity of students effected) than charters could ever be. We have a legal mandate to enroll by a random lottery.”

I’m going to hazard a guess that Andrew has a particular image of creaming in mind: the intentional and systematic use of selection criteria to choose which students attend a school. But there’s another view which I’d like to put forward: creaming is any selection process, intentional or unintentional, that results in the students within a school being more likely to succeed due to their differences from the broader population of students from which they were drawn. Andrew’s definition helps to illuminate the intentions and actions of school leaders; but I think mine is more useful in making comparisons among schools both in terms of the kinds of students they serve and their relative effectiveness in promoting student outcomes.

I’ll use Democracy Prep as an example, but want to make clear that I am not criticizing the school or its practices. Democracy Prep, like most charter schools, is staffed with talented, hard-working people who are trying to promote the best outcomes for their students, and they are doing so within the provisions of the rules governing charter schools.

So: Seth Andrew might point out that (a) the school admits students by a lottery; (b) there are approximately eight applications for every slot; (c) the student body matches that of the surrounding district, with 72% of the students eligible for free- or reduced-price lunches, about 80% Black and 20% Latino students, 10% Limited English Proficient students and about 16% of the students with disabilities; and (d) there is a very simple one-page application form for the school. He might even point to baseline data on student achievement suggesting that students new to the school are well below grade-level. If the administration is sincere about open admissions, has made it easy to apply, and the resulting student body resembles students in the surrounding neighborhoods, with substantial numbers of low-achieving, special education and LEP students, where’s the creaming?

In response, I would point out that (a) there’s lots of evidence that the kinds of families that select themselves into charter school lotteries differ from those that do not; (b) these differences, some of which are easily measured, and others not so easily, may bias estimates of the effects of schools on their students; (c) there are school policies and procedures that might result in some parents choosing not to enter the lottery; (d) in some cases (but not necessarily Democracy Prep) the exclusionary messages are quite explicit; and (e) selection out of a school can be just as important as selection into a school in shaping the student body and estimates of a school’s effects on its students.

Here I’ll just draw a few examples for (c), (d) and (e). A parent selecting into Democracy Prep is agreeing to a distinctive set of school rules: purchasing uniforms at Land’s End, extended day programs on weekdays and Saturdays; two hours of homework a night; grooming guidelines regarding hair and nails; mandatory parent-teacher conferences; and a highly-detailed set of behavior guidelines with clear rewards and punishments, including a system of shunning students who have broken serious rules. Again, I’m not passing judgment on these policies and practices—simply proposing that the families that would agree to them probably differ in meaningful ways from those that would not.

As for (d), I’ll point to Elissa Gootman’s recent New York Times profile of the irrepressible Eva Moskowitz, who operates a chain of charter schools not far from Democracy Prep, which quotes Ms. Moskowitz saying to parents at an information session, “If you know you cannot commit to all that we ask of you this year, this is not the place for you.” Seth Andrew might not say something like this, but nearby charter operators apparently do.

Perhaps the clearest evidence of selection at Democracy Prep, however, is in the loss of students from year to year. Students leave schools for many reasons—their families move away, they decide the school is not a good fit for the student’s needs, in rare cases a student might be expelled—but regardless of the reason, there is good evidence that students who are stable differ from students who are mobile, and that stability promotes achievement.

Democracy Prep lost a lot of students from its first year to its second. The school began with 135 new sixth-graders in August, 2006, and 117 took the school’s end of year tests. But only 96 seventh-graders took the state math exam in the 2007-08 school year. Assuming that the school took no new seventh-graders in 2007-08, Democracy Prep lost 39 of its initial 135 sixth-graders by the middle of the second year of the school’s operation—an attrition rate of 29%. Even if this attrition is not driven by the school intentionally pushing students out, those who are left are the product of a creaming process. And a failure to acknowledge this creaming can distort our understanding of who is being served by a school, and with what consequences.