[seanmcdirmid]

This sounds like something that is especially hard to get data on, there is the obvious selection bias that only those who want to fill out the survey are surveyed.

As for 2/100, I thought you meant onsite interviews, which would be especially crushing for those of us who are going to do a google onsite interview pretty soon. If its just applicants, I can’t imagine it being even that good since Google must get a lot of resumes and without a recommend your chance of being noticed isn’t that great. And with the recommend, it looks like you can skip the phone screen, further skewing the numbers.