I've been pair programming for more than a decade, so I definitely know how much it can help newbies. My point is more that one way they can skew the stats in their favor is by hiring people who would have been hired by those companies with or without the program. E.g., people whose issue is self esteem, or people who aren't good at self-evaluation, or people who lack the sort of personal connections that let other hard-to-evaluate candidates get their first break.

I'm not saying that's what they're doing, of course I'm just saying that in this market a strong hiring rate alone doesn't prove anything about the program.