5 ms·

Also, it would be good to know if the children were randomly selected for this program from all the children in Harlem or if ambitious parents who cared more ab

Also, it would be good to know if the children were randomly selected for this program from all the children in Harlem or if ambitious parents who cared more about their children's education could jockey for position and placement.

dagw 8y ago
Enrollment to the program is apparently done by lottery among everybody that applies and the study compared kids that applied and got selected to kids that applied but didn't get in. So that should at least go some way to account for the parent factor.
- burfog 8y ago
  It fails. In the special school, 100% of the kids/parents actively took an interest in education. In the alternatives, less than 100% did. School quality is largely determined by the students in the school. This doesn't just mean that the performance of student X is determined by student X caring about school. The performance of student X is also determined by the fact that student Y cares or doesn't care about school. So in all the non-special schools, there exist the sort of uncaring disruptive students who will drag down everybody. It isn't that the special school is itself somehow wonderful, but that the bad influences are excluded. Simply put, one can not scale this solution. Providing the supposed quality school to everybody is not possible because the quality of a school is largely determined by the students.
  - dnomad 8y ago
    You're asserting without any evidence that every student Y who "doesn't care about school" would not respond to the protocols described in this paper? This is an extraordinary claim and probably wrong. It needs to verified with data. The paper does provide the basis for a logical next step: imposing these protocols on all students in a region and measuring the impact.
    - sheepmullet 8y ago
      > You're asserting without any evidence that every student Y who "doesn't care about school" would not respond to the protocols described in this paper? No. They are simply stating the truth that there is no evidence it will work. It is basic selection bias.
      - burfog 8y ago
        I don't know a proper name for the problem, but I don't think "selection bias" is really it. The problem is that the control is getting damaged by the experiment. The control is intended to be similarly-selected students (the lottery losers) in non-special schools, but there are also other students (never applied to the lottery) in those non-special schools who interfere with the experiment. The control students don't get to have those non-special schools to themselves.
        dagw 8y ago
        I don't know a proper name for the problem I think "confounding variable" would the describe the problem. Basically there are two variables that seem to affect the outcome, but since they only controlled for one of them, they can't tell if the second variable is affecting the experiment or not.

dagw 8y ago

Enrollment to the program is apparently done by lottery among everybody that applies and the study compared kids that applied and got selected to kids that applied but didn't get in. So that should at least go some way to account for the parent factor.

burfog 8y ago
It fails. In the special school, 100% of the kids/parents actively took an interest in education. In the alternatives, less than 100% did. School quality is largely determined by the students in the school. This doesn't just mean that the performance of student X is determined by student X caring about school. The performance of student X is also determined by the fact that student Y cares or doesn't care about school. So in all the non-special schools, there exist the sort of uncaring disruptive students who will drag down everybody. It isn't that the special school is itself somehow wonderful, but that the bad influences are excluded. Simply put, one can not scale this solution. Providing the supposed quality school to everybody is not possible because the quality of a school is largely determined by the students.
- dnomad 8y ago
  You're asserting without any evidence that every student Y who "doesn't care about school" would not respond to the protocols described in this paper? This is an extraordinary claim and probably wrong. It needs to verified with data. The paper does provide the basis for a logical next step: imposing these protocols on all students in a region and measuring the impact.
  - sheepmullet 8y ago
    > You're asserting without any evidence that every student Y who "doesn't care about school" would not respond to the protocols described in this paper? No. They are simply stating the truth that there is no evidence it will work. It is basic selection bias.
    - burfog 8y ago
      I don't know a proper name for the problem, but I don't think "selection bias" is really it. The problem is that the control is getting damaged by the experiment. The control is intended to be similarly-selected students (the lottery losers) in non-special schools, but there are also other students (never applied to the lottery) in those non-special schools who interfere with the experiment. The control students don't get to have those non-special schools to themselves.
      - dagw 8y ago
        I don't know a proper name for the problem I think "confounding variable" would the describe the problem. Basically there are two variables that seem to affect the outcome, but since they only controlled for one of them, they can't tell if the second variable is affecting the experiment or not.