I repeat it one more time:
If you go with RAS figures for leaners and assume DEM +3 eg : GOP 33 DEM 36 IND 31 you get:
Romney 47%, Obama 47.5%
To get Romney 45% and Obama 50% given the figures for party affiliation and voting intentions for leaners you get a sample like the following:
GOP 29%, DEM 38%, IND 34%
You tell me how likely that is?!
This scenario is simply impossible...
I am happy to see another numbers cruncher like me :)...