Well, Galt is certainly right that income needs to be controlled for when comparing results from two different groups (kids with fag parents or kids with normal parents being the groups).
However, Indy is correct that the evidence is conclusive: the homos are no more likely to screw up their kids than anyone else.
Early studies suffered from methodological problems, but thanks to recent advances in LGBT rights and more people coming forward, the current research meets normal academic standards. Random, national-level, with controls for the usual factors in educational outcome (race, income, education of the parents, etc).