We don't need science to confirm common everyday experiences, which "little kids need lots of sleep" is for a good chunk of the human race. We may need science to filter through the possible reasons why or a variety of other such things, but not the brute fact.
We also don't need to run a scientific study to know that "if the kids are cranky it's likely because they are hungry or thirsty", "it is highly likely that the kid is lying about the fire-breathing goat breaking into the house and breaking the lamp", or "most kids like to eat candy". Again, plenty of interesting science to be done about the reasons for those things, but science does not require us to be that skeptical.
There's a scientific reason for this, too, which is that whatever study you might be interested in conducting on these matters, "the common experience of humanity" has already run a study a billion times larger, and with more statistical power than you could dream of putting into a study. If humanity is wrong about those things, your study is just going to make some of the same errors. It is not productive to run a scientific study with orders of magnitude less statistical power than the data we already all possess.
(And before you go "but what if they're wrong"... I covered that already in the term "statistical power". We are not obligated by science to pretend that nothing not produced in a study has a statistical power of zero. That's a modern perversion of the idea, and causes some serious philosophical bootstrapping problems if you take the idea seriously.)