I took a quick look at the RCP site, but could not find a link to the basis of their final FL projection, do you happen to have it?
Generally speaking, PEC is the "purest" of the "aggregation" sites (most inclusive of all polling, no adjustment for non polling factors) and their FL projection was a dead heat.
So RCP was doing something (ex: excluding or "weighing" specific polls / types of polls, "adjusting" for non-polling factors, using a different time window or adjusting significance by poll age, using approaches other than simulation, etc.) that skewed their results relative to PEC's projection.
If you can find a link to RCP's methodology, I'll take a look and try to determine what is was.
Do you know what aggregated means?
Do you understand that as regards predicting this election, there was '"pure" aggregation' (ex: PEC) and then there was "aggregation + something else" (ex: 538, RCP?)?
Sure, it's here:
Final RCP average was +1.5 for Romney.