Before you can ask WHY something is true, you must ask IF it is true. One thing that we've learned from a few hundred years of science is that "hey, I think I've noticed something" is a piss-poor way of deciding that your observation is correct. First of all, you only know a tiny tiny fraction of the world's population, meaning that even if this were true for each and every single family you know, your observation would be statistically meaningless.
Secondly, when you compare two people (first child and second child), one is always going to be taller than the other. So even with complete randomness, you'd expect to see this in 50% of the population.
Thirdly, we have the phenomenon of confirmation bias. Once you get this idea in your head, you only notice and remember people that conform to your idea and forget those that don't. This is a very very human thing to do.
If you were a real scientist, you would undertake to do a statistically correct survey of a large population, using things like publicly available medical records or large-scale random sampling of the population. You would then use statistical techniques to figure out if your observations are significantly different from what you'd expect if your data was random. Only then would you begin to ask why this is true.
If you haven't done this work, and you're merely relying on your biased observations of a tiny number of people, then asking the question is a complete waste of time. You might as well ask why unicorns are always pooping pink bubble gum.