With the convention heading towards the balloon drop in Charlotte this evening, the raw number of data points this week has continued to dwindle down to just a small handful.
Perhaps eager to fill the void, the internets have been burning up with some really dreadfully weak polling analysis. I mean, some almost comically piss-poor crap.
We'll lay some of those horrifically bad polling memes to rest in a bit, but for now, suffice it to say that if anyone wanted to put in some halfway decent analysis of today's data, it would be that (a) it doesn't tell us a hell of a lot more than we knew yesterday, and (b) it is still too early to even attempt to play the "guess the bounce" game with regard to this week's festivities in Charlotte.
On to the numbers:
PRESIDENTIAL GENERAL ELECTION TRIAL HEATS:
NATIONAL (Gallup Tracking): Obama d. Romney (47-46)
NATIONAL (Ipsos-Reuters Tracking): Romney d. Obama (45-44 LV; 44-43 RV)
NATIONAL (Rasmussen Tracking): Romney d. Obama (47-44)
NEW JERSEY (Quinnipiac): Obama d. Romney (51-44)
DOWNBALLOT POLLING:
CA-47 (Probolsky Research for the DeLong campaign): Gary DeLong (R) 53, Alan Lowenthal (D) 44 (+)
CT-05 (National Research for the NRCC and the Roraback campaign): Andrew Roraback (R) 42, Elizabeth Esty (D) 35
NJ-SEN (Magellan for the Kyrillos Campaign): Sen. Robert Menendez (D) 45, Joe Kyrillos (R) 41
NJ-SEN (Quinnipiac): Sen. Robert Menendez (D) 50, Joe Kyrillos (R) 40
(+)--A mega-caveat on this one. Read below the jump.
A few thoughts, as always, await you just past the jump ...
DUMBASS MEME #1 : "Ipsos-Reuters is still Romney +1--Obama has "no convention bounce"
There are two reasons why this bit of analysis (which I saw in no less than four different headlines from various media outlets) is so shoddy. For one thing, as I have been preaching for two weeks, one cannot even begin to get an appreciation for the size and scope of a convention bounce until the convention is actually several days into the rearview mirror. Assuming the Ipsos/Reuters sample is culled in even numbers every day, the majority of the respondents in today's tracking poll are still from before the first prime time speakers hit the stage in Charlotte. In all probability, virtually none of them saw Bill Clinton's speech. Any bounce talk now is somewhat akin to analyzing the outcome of the Giants/Cowboys game in the first quarter.
Which, for the record is precisely why last Friday's Daily Kos Elections Polling Wrap had the following headline (emphasis mine):
Daily Kos Elections Polling Wrap: Romney 'bounce' still seems muted, but we're only at midpoint
And that was after the Republicans had already departed from Tampa. Patience, kids.
The second reason why Romney's lead in the Ipsos-Reuters poll isn't all that it is cracked up to be is a matter of sampling. Consider: look at the polling memos offered by Ipsos. Look at the polling memo on the opening day of the Republican convention last week (inaccurately labelled "Day 2," because the original opening day was cancelled). And look at the most recent edition of the tracking poll.
See the difference? In the day 2 memo, you see that 46 percent of the RV sample were Democrats, 42 percent were Republicans, and 12 percent were Independents. Now? Reversed: 46 percent Republican, 42 percent Democratic. In other words, there has been an eight-point shift in party identification in the Ipsos-Reuters trackers. Which could easily explain that five-point shift among RVs, and that smaller three-point shift among LVs, in the Ipsos/Reuters poll. The exact impact of the shift in the sample is not clear, of course, because weighting for age, gender, education, and ethnicity would probably dull the shift a bit. That said, there is little chance that the dramatically different party ID numbers played no role in the Romney movement.
DUMBASS MEME #2 : Gallup hasn't moved--Obama is doomed
With Gallup's seven-day window for their presidential head-to-head numbers, it is even more absurd to try to make "bounce" projections in the midst of the convention. Hell, there are still respondents in the pool from before Mitt Romney spoke, let alone Bill Clinton.
That said, the stability of Gallup on the trial heat question has been strange, to say the least. Especially when you consider the week-plus run of Obama +1 came not terribly long after a similarly lengthy run where the Gallup take on the race froze at Romney +2.
One metric has been moving, however, and it may portend future movement in favor of the president. President Obama's job approval numbers have moved up quite nicely over the past five days. What's more: the job approval tracking poll is a three-day, and not a seven-day, sample. Obama stood at a lackluster 43/48 in the Gallup release this past Sunday, which would have been the "post-convention" sample. Today? Obama stood at 49/45, a net improvement of nine points. Of course, his approval moved several points up and down over the past week with no change in the topline results. So this recent "surge" in job approval may not be of much help. But it is pretty hard to see him doing worse, as his job approval numbers edge upward.
DUMBASS MEME #3 : New Jersey is suddenly in play!
Saw this in two locales, though admittedly one was a right-wing blog. This is based on the Quinnipiac poll which showed Obama leading Romney 51-44 in the Garden State. The same poll gave Democratic Sen. Robert Menendez a 50-40 lead over Republican challenger Joe Kyrillos. While the margin for Menendez is basically what it had been before, the Obama lead of seven points is roughly half of what it had been before.
The problem here, as Quinnipiac themselves made clear, is that this is their first stab at polling the state with a likely voter screen. Indeed, they made clear in their release that this poll, properly understood, could not be compared to their previous polls.
A likely voter screen giving a Republican a bump of 5-7 points would seem to be a bit generous, but credit to Quinnipiac for not pretending the change from RV to LV was meaningless. Folks who interpret polls for media outlets should be bright enough, or intellectually honest enough, to follow their lead.
In other polling news...
- After yesterday's mega-polling dump by a Florida outfit called "St. Petersburg Polls and Surveys," we here at Daily Kos Elections got quite a few responses from our Sunshine State friends. And the consensus opinion? Fuhgeddaboutit. And looking at the internals, they may be onto something. The most audacious example? FL-09, routinely been described as heavily Hispanic, had scant few Hispanic respondents, according to their polls. They also seemed to struggle finding voters under 30, which would also skew the polls rightward. In the name of giving you every poll, they were in the mix yesterday. But, as always, use a lot of salt.
- Speaking of a lot of salt, that CA-47 poll with DeLong ahead is super-duper sketchy. Not only is today's release accompanied by a ton of dust (the sample was from the first couple of days in July), but the bigger issue, as Sarah Bennett of the Long Beach Post noted, is that DeLong already released a Probolsky Research poll from early July, showing him down three. So, team DeLong: how can the same pollster, on the same dates, get results that are twelve points different? I have heard of pushback (Democrat Alan Lowenthal dropped a poll yesterday showing him up 20), but it still has to conform to reality, does it not?