Standardized Tests not a good measure of Fluid Intelligence--
134 views
Skip to first unread message
Payman Saghafi
Apr 28, 2022, 3:32:41 PM4/28/22
to Dual N-Back, Brain Training & Intelligence
Gwern Branwen
Apr 28, 2022, 3:56:28 PM4/28/22
to N-back
Her editorializing title is wrong (and also not in the original MIT
article or researcher quotes) - which should be unsurprising for a
blog highlighting growth mindset, emotional intelligence, creativity,
and what seems like every other dubious panacea & fad in educational
psychology.
The fact that some grade changes do not far-transfer to intelligence
is not surprising, nor that you can find lower g loadings, and it
can't meaningfully change the overall correlation between grades and
intelligence. The grade/IQ correlation was r ~ .5 before, and it will
be r ~ .5 afterwards - because the existence of school variations is
*already* influencing that <1 correlation! We already calculate the
correlation of standardized tests and fluid intelligence in a
population subject to such differences. There is nothing new there,
and to the extent that there is, it will not arithmetically make
standardized tests a noticeably worse measure of fluid intelligence
because their causal estimate, for example, is based on... what, 100
students in the charter schools? There are >50,000 students in the
Boston public school system:
https://en.wikipedia.org/wiki/Boston_Public_Schools
--
gwern
https://www.gwern.net
article or researcher quotes) - which should be unsurprising for a
blog highlighting growth mindset, emotional intelligence, creativity,
and what seems like every other dubious panacea & fad in educational
psychology.
The fact that some grade changes do not far-transfer to intelligence
is not surprising, nor that you can find lower g loadings, and it
can't meaningfully change the overall correlation between grades and
intelligence. The grade/IQ correlation was r ~ .5 before, and it will
be r ~ .5 afterwards - because the existence of school variations is
*already* influencing that <1 correlation! We already calculate the
correlation of standardized tests and fluid intelligence in a
population subject to such differences. There is nothing new there,
and to the extent that there is, it will not arithmetically make
standardized tests a noticeably worse measure of fluid intelligence
because their causal estimate, for example, is based on... what, 100
students in the charter schools? There are >50,000 students in the
Boston public school system:
https://en.wikipedia.org/wiki/Boston_Public_Schools
--
gwern
https://www.gwern.net
Pontus Granström
Apr 29, 2022, 5:26:51 AM4/29/22
to Dual N-Back, Brain Training & Intelligence
What are you talking about?
--
You received this message because you are subscribed to the Google Groups "Dual N-Back, Brain Training & Intelligence" group.
To unsubscribe from this group and stop receiving emails from it, send an email to brain-trainin...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/brain-training/CAMwO0gxGoeMJ2ZQpCZGFOW-FGrVHqtWoBS0cTh5Azkz8dGXdPQ%40mail.gmail.com.
Reply all
Reply to author
Forward
0 new messages