Harm or hurt?

Marcus Winters in City Journal:

Was there a relationship, I wondered, between the percentage of students that a traditional New York City public school lost to the charter sector one year and that school’s academic performance the following year? Using data on individual students over time, I found that the more students a public school lost to charters, the better its remaining students performed—probably because the school now faced competition from charters for enrollment.

It's good scholarly work.  As usual, I have a quibble.  It's with the messaging.

Here's the study itself.  Winters wrote:

  • For every 1 percent of a public school’s students who leave for a charter, reading proficiency among those who remain increases by about 0.02 standard deviations, a small but not insignificant number, in view of the widely held suspicion that the impact on local public schools of students’ departures for charter schools would be negative.
  • Competition from charter schools has no effect on overall student achievement in math.
  • In both math and reading, the lowest-performing students in public school benefit from competition from charter schools.

I think it's fairer to characterize those findings this way: "The more students a public school lost to charters, the better its remaining students performed as measured by growth -- although that improvement was quite small, and only in reading, not math.  At the very least, opponents who say kids who remained in the regular schools lose ground are wrong.  They don't.  However, opponents still have one small but legitimate gripe: if charter students arrive on average, slightly better off (true in Boston, for example, per Kane/Angrist, by 0.2 of an SD), then we'd expect the district's absolute test scores (not its growth scores) to decline a bit as charter enrollment expands.  That is: The children themselves are a wee wee bit better off.  But the school's public image -- since most people focus on absolute scores, not growth, which seems to bother only Matt DiCarlo and me -- is possibly a wee wee bit worse off."