households headed by people under 30 is somehow a good indicator of college dropouts? would not be suprised if there were other effects like college students living in their parents garage to bootstrap their startup.

could have measured the growth in accelerator applications by college students as an alternative