Saturday, October 6, 2012

Monitoring the Debate Bounce

Some polls have come out since the debate and all of them show Romney getting at least some bounce from the debate. Obviously, by definition, they don't tell us anything about how lasting a bounce would be but we also don't yet have much clarity on the size of the bounce. Let's take a look at what's been happening:

1) Today, Gallup's tracking poll moved from Obama +5 to Obama +3. Gallup uses a 7-day rolling sample so they've only got essentially 2 nights (Thursday night and Friday night) of post-debate sampling in there. The last Gallup tracker that had (almost) NO post-debate respondents was Thursday morning's release and Obama was up 4 in that poll. So, Romney gained a point in the two days since. Let's pretend for a moment that Romney's gains will continue to be the same in a linear fashion (this is very questionable but let's just go with it for now) as the rest of the sample becomes a post-debate sample and it would mean that Romney would go from Obama +4 to Obama +0.5 by next Thursday morning.

2) Rasmussen's tracking poll had Obama ahead by 2 in their Thursday morning release and Romney is up by 2 points in this morning's release. Rasmussen uses a 3-day rolling sample so, approximately two-thirds of their sample is post-debate respondents at this point. If we extrapolate that data the same way we did with the Gallup data, we would have it going from Obama +2 to Romney +4.

3) Reuters/Ipsos has an online tracking poll (pretty questionable methodologically) and they seem to use a 5-day rolling sample. Their poll had Obama ahead by 5 prior to the debate and now has 3 days of post-debate data and has Obama ahead by 2. If we extrapolate that out, we would have them tied by the time they have a full sample of post-debate interviews.

There have also been some state-level polls that show Romney gaining ground but they are almost all from Rasmussen, We Ask America, and Gravis Marketing and the results are little more mixed in terms of what they tell us about a Romney bounce than their topline numbers suggest. Anyway, taking the numbers above, we see the bounce looking like this:

Gallup - 3.5 points
Rasmussen - 6 points
Reuters/Ipsos - 5 points

Average - 4.83 points

If that's right, that would be about the biggest post-debate bounce ever. Is it possible that that is what's happening? Yes, though it is still not likely. But, even if it does happen that way, here's several reasons Democrats shouldn't panic:

1) Even if the effect of the debate is this large, some of it will fade. Remember how the 47% thing was going to be the end of Romney? The first debate's effect will have its own half-life and all of it won't last through Nov. 6. Other things will happen (more debates for one).

2) Remember that these tracking polls are bleeding out old samples on the back end as new samples come in. What we're seeing could be the effect of overly strong samples for Obama leaving the tracker on the back end as much as Romney getting a boost on the post-debate side. Now, that's not likely to explain all of this bounce. But it could explain a small part of it.

3) Obama was leading by about 4 points in most polling averages prior to the debate. Indeed, as described above, he was leading by 4 points and 2 points and 5 points in the Gallup, Rasmussen, and Reuters trackers. So this would make the race about even nationally. It would hurt, but it would not be the end for Obama.

4) I have emphasized the margins here but the news is less bad when you look at Obama's numbers alone. Here's the movement in the three polls for Obama's numbers:

Gallup - No movement (from 49 to 49)
Rasmussen - Negative 2 (from 49 to 47)
Reuters - Negative 2 (from 48 to 46)

Assuming these trends continue (as calculated above), Obama will end up somewhere between 45 (Reuters) and 49 (Gallup) in all these polls. Ipsos pollster, Julia Clark said,
We haven't seen additional gains from Romney. This suggests to me that this is more of a bounce than a permanent shift
She continued
while the debate was effective in energizing the Republican base and giving Romney a boost, it didn't fundamentally change perceptions of either man a great deal.
Taking everything here together, it seems to me that the calculations I've done above effectively represent a worst-case scenario for the effect of the debate for Obama. I continue to think the effect of the debate will not be very large and that we won't really know for sure until we see some better, systematic, non-tracking polls done. Specifically, I think it will be half or less of what the estimate above indicates and even some of that will fade.

We should know soon enough what the right answer is but, in the meantime, the points I outline above are reasons why people shouldn't panic about the debate just yet.

No comments: