[nudephil (OP)]

Here's a question from listener Melissa:

How long might you have to wait in order to make a scientifically sound analysis of the effects of isolation and virtual learning on children and young adults, with accurate numbers and cases?

[alancalverd]

It's pretty standard for adults.

The Open University has a fairly high dropout rate  in some subjects but has demonstrated that 3 - 5 year part-time degrees are entirely feasible and of high quality.

Most civilian aviators take advanced ground school (from private pilot to commercial and airline) on-line or by DVD for about 200 hours with a one-day classroom brush-up  before each exam. 

I have a son who has been teaching physics up to A level by Zoom for most of this year (boarding schools in the Far East don't mess about with COVID - they sent all the kids home on Day One and have had no problems as a result).He misses the atmosphere (and football and cricket) but isn't concerned about the rate of transfer of knowledge. 

[Bored chemist]

I'm sure you can teach a lot of stuff perfectly well by video link.
But school isn't just about lessons and I'ma lot less sure about the missing "social" side of things.

It's not easy to see how that is measurable; there's no sensible control group.

[evan_au]

Here in Australia, students have just completed their year 12 exams, with results to be returned this week (a different exam is administered in each state).
- The extent of lockdowns has varied from state to state, so having different exams makes comparisons between different states more difficult

I am sure that the people analyzing the marks could give a provide a reasonable indication of how the raw marks compare between this year and the last few years (by comparing questions of similar difficulty, for example).

But in practice they will just scale the marks so they fit on a nice bell-curve distribution, and use that for university admissions...

[Bored chemist]

Imagine that you could look at "how well did this year's students do compared to last year's".
And imagine you could do that in each state.
And imagine that you could look at the extent of lockdown in each state.

You could plot some sort of "change in results" vs " extent of lockdown" .
And you might think that gives you an indication of the effect of lockdown on results.

But the states did different types of lockdowns for reasons (medical, practical,demographic, political ... whatever).
And it's hard to see how those reasons wouldn't be confounding variables