Dance weekends with two callers try to book ones of different genders. When the majority of callers were male this meant avoiding booking two men, but now that the majority is female it means avoiding booking two women.

How can we know this? When I first looked into the gender breakdown of contra callers, in 2015, I used a small list of dance weekends where I had many years of data. I saw that a bit more than half of individual callers at those weekends over that ~20y period were male, and a bit under half the slots were called by women:

	n-female	%-female
Callers	40	41%
Slots	125	47%

We can see that they were gender balancing because male-female pairs happened much more often than you would expect by chance:

  ww: 10
  wm: 37
  mm: 10

Using the data I published yesterday, we can see how things are now:

		n-female	%-female
Callers	2016	49	54%
	2017	54	53%
Slots	2016	166	66%
	2017	160	60%

The number of female callers has risen a lot, from 41% to 54%, and the overall number of slots called by women rose from 47% to 63%.

Looking at weekends with exactly two callers I see:

2016:
  ww: 16
  wm: 49
  mm: 8

2017:
  ww: 15
  wm: 42
  mm: 8

Since the overall number of slots called by women is 63% (326 out of 516, over two years) the most likely outcome for two-caller weekends would be: [1]

  ww: 40%
  wm: 47%
  mm: 13%

but instead we're seeing:

  ww: 22%
  wm: 66%
  mm: 12%

Organizers are still favoring male-female pairs, but because there are now more female callers than there used to be this now means booking extra men instead of extra women.

(Bands still skew very heavily male, but I haven't counted since 2014 because that's a lot more work.)

[1] These come from looking at how likely you are to get combinations events with 63% probability (booking a woman) or 37% probability (booking a man). Specifically:

  ww: 63% * 63% = 40%
  wm: 37% * 63% + 37% * 63% = 47%
  mm: 37% * 37% = 13%