Yes, exactly. To prove a correlation, in my opinion, one would need.... 1. More smoothed/averaged dataset (centered rolling average ~4 months?) 2. Expl

Yes, exactly. To prove a correlation, in my opinion, one would need.... 1. More smoothed/averaged dataset (centered rolling average ~4 months?) 2. Explain the other extreme dips in this graph (2005, 2007, 2011) 3. Plot similar data from other states (with and without legal marijuana) 4. Investigate other causes (see sparrish's comment)