Clicking on the link called All General Election: Romney vs. Obama Polling Data on the "General Election: Romney vs. Obama" page at Real Clear Politics currently brings up the following table:
|Poll||Date||Sample||MoE||Obama (D)||Romney (R)||Spread|
|RCP Average||9/26 – 10/7||–||–||48.3||47.2||Obama +1.1|
|Rasmussen Tracking||10/5 – 10/7||1500 LV||3.0||48||48||Tie|
|Gallup Tracking||10/4 – 10/6||1387 RV||3.0||47||47||Tie|
|Politico/GWU/Battleground||10/1 – 10/4||1000 LV||3.1||49||48||Obama +1|
|Gallup Tracking||10/1 – 10/7||3050 RV||2.0||50||45||Obama +5|
|CNN/Opinion Research||9/28 – 9/30||783 LV||3.5||50||47||Obama +3|
|National Journal||9/27 – 9/30||789 LV||4.2||47||47||Tie|
|NBC News/Wall St. Jrnl||9/26 – 9/30||832 LV||3.4||49||46||Obama +3|
|NPR||9/26 – 9/30||800 LV||4.0||51||44||Obama +7|
Polls in grey are the ones included in the average. In times past, the non-selected polls were either older versions of more recent polls, or were just older polls that Real Clear Politcs just decided (according to no published methodolgy) to let drop off the average. This weekend, RCP managed the far too cute trick of just selecting polls that it liked to include in it's widely cited "RCP Average". Notice that both the NPR and NBC/WSJ polls have the exact same date range. RCP decided it "liked" the NBC/WSJ poll and "didn't like" the NPR poll.
Today, RCP went the extra step of just making up a fake "Gallup Tracking poll", based on an article Gallup published on an intermediate result, and removed the actual, more recent (and real) Gallup Tracking poll from it's "RCP Average". Reporters, please take note.
Update 10/8/2012 3:24 p.m.
RCP has now added asterisks (*) next to both the real Gallup Tracking Poll and their fake "Gallup Tracking Poll", but are still only including the fake "Gallup Tracking Poll" in their "RCP Average". Here is their note:
*Gallup released two tracking poll results on October 8. The more recent results (reflecting post-debate interviews)
are included in the RCP National Average.
This is an error at best, or an outright lie at worst, since the Gallup press release never claimed that the intermediate result they published was their tracking poll itself and the actual tracking poll released today has the most recent polling numbers.
Removing the fake "Gallup Tracking Poll" (and the landline-only Rasmussen robopoll) and restoring the NPR poll gives a 9/26-10/7 "Tom's Average" of Obama +2.7.