It is true that "significance" can be arbitrarily defined, but I think if you can calculate the confidence, it is not arbitrary. If you have the number that with x% probability something is the case, you have learned something.
I suspect that 490 is quite a big sample actually, but I am too lazy to do the maths (and I am missing background information - of course if it was just one battle, it is a different thing than looking at the whole of WW2, and so on).
Even the 3 IQ points I am not sure about - it might on depend where on the curve the difference appears. 30 points is a very big difference apparently (IQ 100 makes you smart enough for high school, and IQ 130 makes you smart enough for the Mensa club of highly intelligent people, or something like that). The 3 is not "3% difference in intelligence". I am not familiar enough with the whole IQ thing to be able to judge if the difference is significant, though.
I suspect that 490 is quite a big sample actually, but I am too lazy to do the maths (and I am missing background information - of course if it was just one battle, it is a different thing than looking at the whole of WW2, and so on).