4 ms·

This study doesn't tell us anything we didn't know. Yes, when you give kids the best instructors and best resources, they do better. The "success" in this stu

by balloot 15y ago

This study doesn't tell us anything we didn't know. Yes, when you give kids the best instructors and best resources, they do better. The "success" in this study is that regular kids in a regular class end with 10% of kids testing for gifted, and regular kids in a "gifted" class show 15-20%. That simply is not that impressive or surprising. The problem is that good instructors and resources are limited. And while those things would help anyone, you maximize their value by giving them to the kids on the gifted track. We, as a country, would be worse off if we didn't do this because we would be letting the truly talented go to waste in the name of "fairness."

scott_s 15y ago
http://lesswrong.com/lw/im/hindsight_devalues_science/ http://lesswrong.com/lw/im/hindsight_devalues_science/
- gwern 15y ago
  There's nothing hindsight about this. We can make confident predictions: whatever this is, it will prove to be of very limited use or none at all. Why am I so confident? Because like with boosting IQ, researchers keep trying, and they keep failing. Their repeated failures tell us a lot about the probability the next study will succeed... Education is full of big gains promoted to anyone who will listen as the answer to this or that gap - like this one! - which turn out to be flukes (p=0.05 means you just need to keep trying...), statistical or data gaming (what does it really mean to have some specially treated kids reclassified into the gifted program? are the people choosing for the gifted program even blinded as to which kids they are expected to choose and vindicate this expensive program?) and Goodhart's law, regression to the mean or 'fade out', and outright fraud. like Head Start; there are benefits - and then they fade away: https://secure.wikimedia.org/wikipedia/en/wiki/Head_Start_Program#Mixed_studies_and_statements https://secure.wikimedia.org/wikipedia/en/wiki/Head_Start_Pr... or http://edlibertywatch.org/2011/03/studies-on-effectiveness-of-early-childhood-programs/ http://edlibertywatch.org/2011/03/studies-on-effectiveness-o... Suppose this study had been conducted in this http://www.ajc.com/news/investigation-into-aps-cheating-1001375.html http://www.ajc.com/news/investigation-into-aps-cheating-1001... recently exposed nest of crime?
  - scott_s 15y ago
    You're addressing something different. I interpret the grandparent as "this study tells us what was intuitively obvious" which is hindsight bias.
rcfox 15y ago
As someone who went through "gifted" class for a while, I disagree. Those who are generally considered "gifted" tend to be good at one or a few things. However, each person is likely to be good at a different thing than someone else, so grouping all of these people into the same class can sometimes be pretty pointless. If you want to maximize the utility of "gifted" students' time in class, you need to give them challenging (but possible) tasks, and then leave them alone. Put the good teachers in charge of the average-but-gritty[0] students. [0] http://en.wikipedia.org/wiki/Grit_(personality_trait) http://en.wikipedia.org/wiki/Grit_(personality_trait)

by balloot 15y ago

scott_s 15y ago

http://lesswrong.com/lw/im/hindsight_devalues_science/ http://lesswrong.com/lw/im/hindsight_devalues_science/

gwern 15y ago
There's nothing hindsight about this. We can make confident predictions: whatever this is, it will prove to be of very limited use or none at all. Why am I so confident? Because like with boosting IQ, researchers keep trying, and they keep failing. Their repeated failures tell us a lot about the probability the next study will succeed... Education is full of big gains promoted to anyone who will listen as the answer to this or that gap - like this one! - which turn out to be flukes (p=0.05 means you just need to keep trying...), statistical or data gaming (what does it really mean to have some specially treated kids reclassified into the gifted program? are the people choosing for the gifted program even blinded as to which kids they are expected to choose and vindicate this expensive program?) and Goodhart's law, regression to the mean or 'fade out', and outright fraud. like Head Start; there are benefits - and then they fade away: https://secure.wikimedia.org/wikipedia/en/wiki/Head_Start_Program#Mixed_studies_and_statements https://secure.wikimedia.org/wikipedia/en/wiki/Head_Start_Pr... or http://edlibertywatch.org/2011/03/studies-on-effectiveness-of-early-childhood-programs/ http://edlibertywatch.org/2011/03/studies-on-effectiveness-o... Suppose this study had been conducted in this http://www.ajc.com/news/investigation-into-aps-cheating-1001375.html http://www.ajc.com/news/investigation-into-aps-cheating-1001... recently exposed nest of crime?
- scott_s 15y ago
  You're addressing something different. I interpret the grandparent as "this study tells us what was intuitively obvious" which is hindsight bias.

rcfox 15y ago

As someone who went through "gifted" class for a while, I disagree. Those who are generally considered "gifted" tend to be good at one or a few things. However, each person is likely to be good at a different thing than someone else, so grouping all of these people into the same class can sometimes be pretty pointless. If you want to maximize the utility of "gifted" students' time in class, you need to give them challenging (but possible) tasks, and then leave them alone. Put the good teachers in charge of the average-but-gritty[0] students. [0] http://en.wikipedia.org/wiki/Grit_(personality_trait) http://en.wikipedia.org/wiki/Grit_(personality_trait)