I just finished completing analysis of Texas TAKS data from 2010 and 2011. Not surprisingly, students in fulltime online schools had differences in relative learning growth by age. Younger students in online programs moved down toward the negative end of the distribution, while older students had no real change (and sometimes even positive growth).
To me this suggests that policymakers should definitely consider grade-level (in addition to other factors) before allowing students to enroll in online programs! I am submitting these findings to the American Educational Research Association (AERA) conference for 2015. Wish me luck!