In reasonably sized datasets, you'll also find a lot of spurious correlations simply by chance. That's one reason in science you're supposed to write down your hypothesis and methods of analyzing data before touching the data. Otherwise you risk finding some random noise and thinking it's important.