No one from the ManyClasses team is active on X anymore (it took us a couple days just to find our login 🙃), but we’re interested in constructive feedback! Feel free to reach out to us via email with any reactions or comments.

We wish we had more time to recruit additional schools and teachers to participate, but the competition rules really constrained our timeline. We’re sincerely grateful for the collaboration from IU, Florida, UNL, and South Fayette School District.

The study was conducted using @terracotta_ed, as part of our entry in a national research competition, the @XPRIZE Digital Learning Challenge. Terracotta is an emerging research tool that makes it easy to conduct experiments in the @Canvas_by_Inst LMS.

While we found a generalizable benefit of prequestions, we also found that prequestions reliably caused some students to disengage, and ultimately we advocate for a revision of our theoretical understanding of the prequestion effect.

MC2 examined prequestions across 30 different classes. Student participants (n = 1,571), from 6th grade to college seniors, were randomly assigned to answer ungraded prequestions prior to watching online lecture videos as part of their routine learning activities.

ManyClasses 2 is now released as a preprint! 🥳 osf.io/preprints/psyarxiv/3x…

Ooo that'd be cool!!! You should do it! "ManyClasses" is a model for doing research, and one that we'd love to see adopted more broadly. If we can help let us know! (and kudos @mikehobbiss for schooling folks on crossmodal attention in that thread!)

The generalizability crisis commentaries are out (including ours; doi.org/10.1017/S0140525X210…). Eager to see what generalizes!

Throwback to two years ago, February 2020, hours each day manually coding moderators. Fun times!

Shout out to our friend @cab938 for co-organizing a preconference workshop at @lak22online on fostering open research practices in learning analytics. Come join us! sites.google.com/view/buildi…

But whether students knew before the assessment is another question. We'd be interested to see the research you're mentioning! -- we hadn't seen that specifically in our feedback timing literature review.

Oooo good question! We tried to ensure that students knew that they'd receive feedback, and that we were manipulating feedback timing. We were pretty transparent in the consent form (osf.io/p6jxq/) and teachers were supposed to discuss in class (osf.io/96sxc/)

But one limitation: No instructors of writing classes applied to participate in ManyClasses, and thus no writing classes are included. Whether the effects observed generalize is exactly the sort of question that we think more researchers should be asking! Skepticism welcome!

Sixteen of the 38 participating classes included explanatory feedback. This moderator did not affect the relationship between feedback timing and student learning. Full list of moderators: osf.io/tjkyr/ ...

True, we don't seek to be boat rockers, actually we see ourselves as boat steady-ers (?) in the decades-long debate about feedback timing. But strictly speaking, no, we didn't operationalize feedback as just showing the correct answer...

Our approach was to let each teacher report their own relevant outcome scores from each treatment period. Without teacher-assigned grades, could still use student's self-assessments. It'd be messy, but that's kinda the point of translating learning theory to authentic assessment.