It's a non-random sample. You can draw valid statistical conclusions from a surprisingly small random sample. You can draw bullshit conclusions from a surprisingly large non-random sample.