Is there a "threshold"?

[jradavenport]

An observation made at AAS 225's hack day was that the gender ratio for the first question was not the same as the subsequent questions. The NAM 2014 study found that if Q/A sessions go to ~4 questions, the demographics flatten more. Use all data from #10:

• [ ] If we ignore the 1st question, do the gender ratios get better?
• [ ] If the first Q is a F, are the subsequent Q's more likely to be F than if the first is a M?
• [ ] Gender ratio as a function of # of total Q's asked? (maybe not possible or interesting)
• [ ] If Q/A sessions last longer, does the gender ratio become better? Is there a threshold minimum number we can recommend?

I think @bmorris3 thought about this a good amount at AAS 225

[bmorris3]

I've updated the AAS225 hackbook to python 3 so we can hack on it a bit more.

[jradavenport]

:+1: This is great @bmorris3! We should be able to just cut/paste the code from this notebook to make the script to tell us what we need to know and make the figures!

[bmorris3]

A lot of that was pretty hacky, as it was scribbled out on the AAS 225 hack day. I'd be happy to make the methods in there:

1. Better documented, so others can follow
2. More general, so other analysis can be done with the same dataset
3. Maybe based on another data structure which may be more appropriate than the typically pure-python list implementations that I used?