I suppose I come forward with my own usage stats, but it is anecdata :)
And the andecdata matches other anecdata.
Maybe I'm missing why that's selection bias.