One of the interesting pieces that is coming out about some of the high performance charter schools suggests that at least some of their success is due to selecting the most promising students. These posts, from New Jersey and Arizona, is due to selecting students from higher income families (less school lunch eligible kids), less learning disabilities, and expelling problem students. Clearly any school that engaged in these tactics would do better relative to public schools (who have a mandate to accept these students and an accountable procedure for expulsions).
I am reminded of parents I know who had their kids "kicked out" of daycare. The theory was a private daycare can select who will and will not be in their clientele and remove kids who do not "fit in". And good for them -- that flexibility is a key part of private business and it can be useful to be able to focus on people your model is compatible with.
But we should realize that this business model flies in the face of the ideas of universal schooling. I challenge you to look at the chart on African-American male students in Northstar and not worry. It's nearly a complete attrition over the course of the cohort's lifespan. It seems incompatible with any strict definition of a 100% graduation rate, unless all of these children went on to transfer to and graduate from public schools. If we value universal education as a public good and an underpinning of American prosperity then maybe we need an approach that is actually designed to do this?
I will also note that it is a key principle of outcomes analysis that you need to look at what happens to the study drop-outs when evaluating an intervention. After all, all of the adverse events on a drug could happen in the post-drug quitting phase. This is not evidence of safety. Nor is sending children who are struggling to public schools evidence that you are able to meet these children's educational needs.
I would be shocked if Mark Palko didn't have a much more detailed analysis to follow this up.