Associate Professor of Public Policy, Politics, and Education @UVA.
I share social science.
Well, because of our new working paper you do!
@mike-bloem.bsky.social, @jonisaacsmith.bsky.social, sam imlay
Reposted by Bruce D. Baker
Also, this isn't quite accurate.
p < 0.05 in the pooled treatment.
p = 0.066 that you reference is for the difference between restricted and unrestricted.
p < 0.01 for unrestricted treatment itself.
Also, this isn't quite accurate.
p < 0.05 in the pooled treatment.
p = 0.066 that you reference is for the difference between restricted and unrestricted.
p < 0.01 for unrestricted treatment itself.
Also, this isn't quite accurate.
p < 0.05 in the pooled treatment.
p = 0.066 that you reference is for the difference between restricted and unrestricted.
p < 0.01 for unrestricted treatment itself.
www.iza.org/publications...
This contradicts concerns about students overly relying on AI tutors.
Restricted access did not produce a statistically significant impact over studying without AI.
1. Restricted-access: access to the AI tutoring system after 10 mins of study.
2. Unrestricted-access: continuous AI tutor support.
3. Control: students study unaided.
All students then completed the same incentivized test without access to the textbook or AI tutor.
Maybe not!
In this new experiment, students studied economics textbook material during a 25-minute session to prepare for a subsequent test. 1/n