Ask HN: Any data-driven evidence for or against pair/mob programming?

Everything I can find about pair programming or mob programming is anecdotal. Have there been any statistically significant studies that either confirm or deny the claims that pairing/mobbing increases long term productivity of a team?