The Gallup Bait And Switch. Reality Shows A Pre-Debate/Post-Debate Of Only A 1-point Shift

[berni_mccoy]

Some may have noticed that 3 days after the first Presidential Debate, Gallup started reporting a different tracking poll. For the entire election, Gallup had been reporting their 7-day Registered Voter Tracking Poll. On the Sunday after the first debate, they released the first set of number of the 7-day *Likely* Voter Tracking Poll. These numbers were quite different from the RV poll. In fact, today's difference is quite stark: The traditional RV Tracking Poll has Obama up, 48-46, while the new poll that came out in the middle of the debate bounce has Romney up 49-47. There is a 4 point swing between these two polls. Surprisingly enough that is about the same size of the Romney debate bounce. And the two most popular sites that aggregate polling data, RCP and Nate Silver's site, both registered about a 4-point bounce for Romney and both started using the new LV poll right when it came out, switching mid-stream from the RV poll.

Note that Nate Silver just today put out a post that shows how the national polls don't correlate with the state-by-state polls. Is there any wonder why? They effectively switched models mid-stream. These two Gallup polls aren't comparable (apples and oranges as the saying goes) and it's clear that Gallup's LV adjustments to their RV data aren't effectively reflecting reality.

If we stay with the original model, maintaining the apples-and-apples comparison before and after the debate, we see a more believable result: That Romney got a subtle 2 point bounce that has settled down to a stable 1-point gain for him nationally. Nate even predicted a 2-point bounce before the debates as the most likely statistical outcome for Romney. It is clear here that the only real shift that came out of the debate was a change in the way the numbers were counted juxtapose with the story the media wanted.

[Journeyman]

1. Has Gallup historically made this switch after the first Presidential debate of an election? . . .

[BumRushDaShow]

2. Yes

and I was yelling this all over the place because the media all latched on this as a narrative and insisted on comparing apples with oranges.

Gallup DOES always switch to LV around this time. But in this case, they released a LV 3-day (in the morning and with a narrative) before their RV 7-day (which releases at 1 pm ET), and the spin from across the media was sickening as they insisted that there was a larger drop than was reality.

I.e., if they had offered their 7-day LV poll through the election season, then one could see how the numbers changed pre- and post-debate using that 3-day LV poll portion against the rest of the LV numbers.

[reformist2]

3. It was just an unfortunate coincidence, but the truth remains that Romney didn't get much of a bump.

On the other hand, it is a bit of a wake-up call to see that the Romney voters are more likely to vote than the Obama voters, something that a lot of us had already heard about, but didn't want to believe... we still have the majority of voters, we just need to get them movtivated!