An interesting thing about this election that I haven't seen much discussion about is looking at a comparison between the polls and the actual results. In arguing with some of my Clinton-supporter friends about how big her win in Pennsylvania would be, I said I believed the margin between the two would narrow. I then went out on a limb and said, "You know, if you look at the polls in a particular state one month out before a primary or caucus and you compare them to the actual results, Obama always does better and Hillary always does worse, every time." Well, I thought that was true anecdotally, but I didn't really have the evidence to back it up, so I was asked to prove it. So below, is my proof....
I found three states where Clinton increased the spread in her favor: Arkansas, Rhode Island and Vermont. Everywhere else, Obama increased the spread in his favor.
I used www.pollster.com as my only source for this. I did my best to take a look at polls exactly one months before an election or caucus, but I couldn't find any in some cases. If I could find one three weeks before or six weeks before, I used it.
I know this isn't scientific; I was just looking to see if in general, my hypothesis was correct, and you know what, it seems like it was pretty close.
If anyone has an interest in looking at my spreadsheet and helping me clean it up, let me know and I will share it with you. And if you have data or access to data I don't have, again, please share it.