Nitpick 1: Texas isn’t really safe red. Everybody but ARG has been showing a single-digit gap for weeks. I predict the next non-ARG poll will show an 8 point lead, at most, for McCain.
Nitpick 2: In 1876, Tilden won the popular vote by 3% (inasmuch as we can really know the popular vote when records in many states are unreliable at that time), but lost the electoral vote.
And he did it by ruthlessly suppressing the black vote in the eight “redeemed” Southern states, in effect “wasting” votes by carrying those states by artificially lopsided margins. (Georgia 72-28, Mississippi 68-32).
Today, a race decided by three points could result in an inversion, but it’s unlikely. This would be at the outer edge of realistic possibility. Beyond four points, you’re getting into monkeys-typing-Hamlet territory.
Really? “Equally” confident? You’re either lying to us or to yourself.
Thanks for the link and explanation.
Thanks! Glad you like them. I’ve had a great time doing them, and it is nice to know they are useful and that they add something (if even just a little bit) to the analysis out there!
They do! I think they are a good visual representation of the state of the race. Well done.
Welcome to the SDMB. I humbly request you take a look at the General Questions forum if you want a good impression of this board. 
If you really want to add something to the analysis, have a version of the charts after applying a 10-day moving average. I’ve been looking for something like that for a while now, and I haven’t seen it.
I thought about doing something like that when I first started, the difficulty is that if you did, for instance, a 10 day average, for most of the race you would have had many states without any polls at all in that window. Now, you could do some rule, where if you had no new polls, you just retained the last 10 day average you did have, until a new poll came around. Or you could just show unknown for those states, but that didn’t seem quite right either.
I picked “last polls, up to five” because it was dead easy to calculate, and also insured that once a state had one poll, it would always have an average. Also, it has the property that in a sparely polled state, if there is suddenly a new poll, I don’t automatically just assume that poll is right because it has been awhile since the last poll. Of course, the flip side is that my averages for sparsely polled states can span a large amount of time, while meanwhile in states that are polled multiple times a day like most of the swing states are, it can sometimes reflect only a single day worth of polls.
There are of course other more sophisticated ways of looking at these things (see pollster and 538) but I think looking at it in “the simple way” has some merit too.
Oh, and I provide the spreadsheet I use on the wiki page, you can download it and it has the various polls for each state easily accessible. It is basically the data from pollster, with some slight variations (mostly explained on the wiki page). I’m just on my lunch break at my day job right now, so don’t have time at the moment, but it should be pretty simple to calculate the last 10 days for each state instead of last five polls, and see if any of the states end up in different categories.
OK, since folks asked…
After doing Sunday’s update on my Predictions, I went through all the states (and DC) to see what would be different if I was looking at the last 10 days of polls rather than the last 5 polls (regardless of date).
The results:
There would be 14 that have not been polled in the last 10 days:
AZ,AR,DC,HI,ID,KY,LA,ME,MD,NE,RI,SD,UT,VT
There are 32 states that would end up being classified in the same one of my categories:
AL,AK,CA,CO,CT,DE,FL,GA,IL,IA,KS,MA,MI,MS,MO,MT,NV,NH,NJ,NM,NY,NC,OH,OK,OR,PA,SC,TN,TX,VA,WI,WY
There are FIVE that would change status.
Indiana would be Weak McCain instead of Lean McCain
Minnesota would be Strong Obama instead of Weak Obama
North Dakota would be Lean Obama instead of Lean McCain
Washington would be Strong Obama instead of Weak Obama
West Virginia would be Lean Obama instead of Lean McCain
This would reduce Obama’s best case in my analysis by 11 electoral votes by taking Indiana out of swing state status. (Making it Obama 372 to McCain 166.)
But in the “everybody gets their leans” Obama would get 8 more electoral votes from North Dakota and West Virginia. (Making it Obama 372 to McCain 166… the same as Obama’s best case, as there would be no more swing states leaning to McCain with Indiana no longer a swing state.)
McCain’s best case would stay the same at Obama 286 to McCain 252.
I’ve put details, including all the numbers here for anybody who wants to look at it in more detail.
Enjoy! 
Very interesting stuff. Thank you.
Thank you.
Oops, I meant after Saturday’s update. I haven’t done Sunday’s yet. 
Personally I don’t recall ever making any such predictions. Probably because the polls didn’t really support them at the time. Electoral-vote.com has archives of its poll maps from the 2004 election, and quite simply whatever leads Kerry enjoyed at any point in the campaign aren’t comparable to this year’s race. McCain can still win, if eight or nine different battleground states all fall his way, but he’s light-years behind the position Bush was in at the comparable point in 2004.