Polling data on American sex preference is sparse, and does not show a clear preference for girls. But the picture from the doctor’s office unambiguously does. A newer method for sperm selection, called MicroSort, is currently completing Food and Drug Administration clinical trials. The girl requests for that method run at about 75 percent.
So there are scientific polls which conclude that people have equal preference for boys and girls but Rosin relies on the sperm clinic survey method instead. But that's a non-random sample: a fatal mistake (and disturbingly common for reporters, especially when shows tout their online polls). What these clinics are really saying is that people who can afford to try to control the sex of their child prefers girls. Rich people like girls.
Rosin might claim that this makes sense: wealthy people prefer girls because wealthy people have the resources to see where the global trends are going. They know a girl will, on average, be more successful than a boy. But that doesn't fix the problem of the scientific surveys: if wealthy people preferred girls and everyone else were indifferent, then you'd still see a slight preference for girls on the aggregate. If everyone else preferred boys, you'd see a preference for boys because there's a lot more non-rich people than rich people. It would be a pretty amazing coincidence if it all evened out nicely. (I don't know if these studies controlled for income, which would be interesting to see.)
There's a better explanation. Without trying, wealthy people tend to have more boys and poor people tend to have more girls. Wealthy people are just using the clinics to balance out these natural tendency to have more boys. Their preference of girls at the clinic is evidence of their equal preference at the nursery.