The Week In Review - The Polls That Aren't

This week there were 21 polls in 19 states. 4 of these polls resulted in color changes for states. 2 moving toward Sen. McCain (Colorado, Kansas) and two moving toward Sen. Obama (South Dakota, Michigan).
I'm beginning to detect that a more traditional looking map is starting to emerge. Although there some states that will certainly change hands in this election. There is evidence that traditional red states are getting redder (see Alaska, Kansas & Louisiana), while blue states are getting bluer (California, Minnesota and Washington).
There are still 9 states (and Washington D.C) that have not received an update in the 4 weeks since I began publishing my results. Some of them are surprising. In Idaho, we have not had a single poll since Survey USA at the end of February. While this is not entirely surprising, with recent polling in the Dakotas and Montana that suggest the race has tightened considerably there, it lends itself to the possibility that more polling should be done. Ironically, one of the most competitive states, New Hampshire, has also not had any polling since June 18th. Currently, I'm still showing New Hampshire as a toss-up and have been waiting for new data to confirm the small lead Sen. Obama has.
Finally, I'd like to talk about Zogby online polling. I am very suspect of the data they are currently providing here http://www.zogby.com/50state/. They appear to be leaning fairly hard to the left. Some examples: Currently, Sen. Obama leads by 3 points in Arizona accoridng to Zogby. Recent traditional polls show Sen. McCain with a lead between 9 & 11 points. In New Mexico, Sen. Obama has a 16 point lead according to Zogby, while other polling ranges from 3-8 points. Finally, in Louisiana, Sen, McCain is shown with only a 7 point lead, while Rasmussen's latest poll (7/9) shows him ahead by 20.
John Zogby strenuously supports his methods and makes valid arguments, but the divergence in these polls are to large to ignore. I am of the belief that because internet polling skews younger than phone surveys, the older (and likely more Republican) voters are being underestimated.