Seriously, I think that if you can get some stats regarding reduced faults in pair programmed code, that may work. I think that you could also mention some of the other benefits of XP such as collective code ownership, clients will be less concerned about staff turnover and holiday absences, etc.

I guess that I'm advocating pushing XP as a package rather than just one part if it.

I've been trying to find an article I read some time ago on the web concerning XP and managers' perceptions of it but it evades me at present. While trawling for it, though I found this article, an XP FAQ which has some links to empirical data concerning pair programming. This might be of some use to you.

Dermot