Well, the random sampling makes it less likely it’s invalid. Random sampling is pretty much the standard for large groups. As for significance, that’s something I’d have to read more to know whether it’s high enough to truly matter. There are very well developed rules for this stuff in statistics.