Why is it that in some families all the siblings are the same sex?

Well, there are some unusual circumstances that could in principle make this more likely than usual, such as a "driving Y" segregation distorter on the Y chromosome. And there is evidence that among deer the sex ratio of fawns is influenced by the social status of the mother; something like that might in principle happen among humans. But they are all pretty speculative: I've heard of driving Y genes in mice, but not in humans. Any such extraordinary circumstance is less common than the single-sex sibships that are expected by simple chance.

Remember that if each sex were equally common (in practice boys are born slightly more commonly than girls) then purely by chance you would expect half of all two-child sibships to be all the same sex, a quarter of all three-child sibships, an eighth of all four-child sibships, over 6% of all five-child sibships, and over 3% of all six-child sibships. A family of six or seven siblings all boys or all girls doesn't require any special explanation unless it be shown that such families are more common than you'd expect.