If nothing else, parental involvement correlates with higher test scores and being enrolled in a non-default school correlates with parent involvement. So it's no surprise that being enrolled in a non-default school correlates with higher test scores.

IMHO, we always hear about such and such school (system) has X% kids proficient with $Y/year per pupil. But what I would really want to know about a school is how does a year change at the school change the proficiency of the class. If the class of 3rd graders starts the year at 20% proficient at 2nd grade level, and ends at 22% proficient at 3rd grade level, that might be a good school, even though a single point in time check says 22% proficient. But the numbers we get aren't really useful for that; a cohort analysis would be better; there's real privacy implications, but that doesn't make the numbers we get useful. :P