The kin selection hypothesis for homosexuality states, essentially, that "homosexual males, freed from the need to expend energy on direct reproduction, may have helped their siblings in ancestral environments reproduce more successfully and increase the viability of their offspring (their nieces and nephews) by resource provision or child care and protection." As this can be applied to paedophilia even more aptly, I guess a lot of theoretical work has already been taken care of.

Qazi and Hull (2005) is the only empirical test of the kin selection hypothesis for homosexuality, as far as I know. Since they found no significant differences in benevolent tendencies, familial affinity, or generous feelings between their homosexual and heterosexual samples, the hypothesis was not supported. Looking at the questions they used (eg. "I would help in any possible way to provide the best education for my siblings' children," and "I would be willing to baby-sit for my nephews/nieces on a regular basis.") it seems clear that the kin selection hypothesis would find more empirical support if applied to paedophilia.

----
Rahman, Qazi, and Hull, Matthew S. (2005). "An empirical test of the kin selection hypothesis for male homosexuality," in the Archives of Sexual Behavior. Available on accessmylibrary or from me.