There are many, many ways of generating adverse selection...

There are many, many ways of generating adverse selection effects that confound statistical studies, and very very few good instruments. Thus I have found myself always very suspicious of the whole "assessing charter schools" literature���not suspicious particularly against any one side, but just suspicious:

Patrick L. Baude, Marcus Casey, Eric A. Hanushek, Greg Phelan, and Steven G. Rivkin (April 2018): The Evolution of Charter School Quality: "Quality dynamics among Texas charter schools from 2001-2011.... Exits, improvement of existing charter schools, and higher quality of new entrants increased charter effectiveness relative to traditional public schools...

...Reduced student mobility and an increased share of charters adhering to "No Excuses"-style curricula contribute to these improvements. Although student selection into charter schools becomes more favorable over time in terms of prior achievement and behavior, such compositional improvements appear to contribute little to the charter sector gains. Moreover, accounting for student composition in terms of prior achievement and behavior has only a small effect on estimates of the higher average quality of "No Excuses" schools...

#noted