Hello and welcome to what will be an intermittent series of diaries published daily after the Gallup tracking poll comes out, depending on whether or not I am by my computer and ready to publish.
These diaries will be devoted to breaking down the day's Rassmussen and Gallup three-day tracking polls into daily numbers and putting those numbers into some perspective.
On the face of it this is a useless enterprise, since the very reason these polls are aggregated is that the daily samples are unreliably small.
That said however when one is watching an election unfold as closely as we are one is in a position to use small daily numbers responsibly. Right? And that is what we will try to do.
Now, there are all kinds of errors I court in trying to break apart these numbers. The numbers on the basis of which we are reverse-engineering the nightly totals have not just been aggregated, they have been rounded and weighted, by formulae that vary. Some nights have bigger samples than other nights. What I am doing here is very imprecise.
Nevertheless it can be done, and there IS information to be gleaned by doing it. On the flip I'll present what I have to offer in the way of breakdowns.
Ok, first off let me give the recent history of the Rassmussen and Gallup polls as I have guesstimated it on a night by night basis.
First off, Rassmussen:
Day AM Average Polled in PM
25 O48 M45 O41 M52
26 O46 M46 O47 M46
27 O46 M47 O50 M43
28 O47 M47 O50 M46
29 O49 M45 O47 M46
30 O49 M45 O50 M46
31 O49 M46 O50 M46
01 O49 M46 O53 M43
02 O51 M45 O48 M46
03 O50 M45 O49 M46
04 O50 M45 O48 M47
05 O48 M46 O48? M46?
06 O48? M46? O48? M45?
07 O48? M46?
Here's a key to this presentation. The "Day" column is the day of the month, August into September. The "AM Average" is Obama's and then McCain's numbers as published by Rasmussen that day, in the form of an aggregate of the previous three evening's polls. The "Polled in PM" numbers are my guesstimates of what Ras went on to get in his poll later that evening. Those numbers don't get averaged in until the next morning. Note here that I go into the future a bit, giving numbers for tonight and tomorrow morning in question marks--this is just to sketch where I think this is going.
With the Rasmussen numbers, it is worth noting that they have a lot of volatility, which may surprise since the averages seem not to move around all that much. The evening of the start of the Democratic convention, for instance, polled extremely well for McCain in what was undoubtedly in part a statistical blip, but also the result of endless handwringing over Hillary voters and reports that the Clintons were going to diss Obama (reports that in retrospect were obviously borne of shrewd expectation setting).
Ok, on to Gallup.
Day AM Average Polled in PM
25 O45 M45 O45 M47
26 O44 M46 O45 M41
27 O45 M44 O54 M38
28 O48 M42 O48 M44
29 O49 M41 O46 M41
30 O49 M41 O51 M41
31 O48 M42 O50 M46
01 O49 M43 O49 M39
02 O50 M42 O48 M43
03 O49 M43 O49 M43
04 O49 M42 O48? M47?46
05 O48<strike?</del> M45?44! O48? M45?
06 O48 M45?
I'll update this in a bit when today's gallup numbers come in. So I'll put my guesses to the test right away! I'm hearing O48 M44. Good news! It means in my reckoning that McCain LOST a point after the Palin speech. Uh oh. Check that. Gallup's revelations about Obama never having more than an eight point lead and never going above 50% have me revising the numbers in a way that suggests quite a GOOD night for McCain last night: up from 43 to 46. Doubt it will be sustainable, but I see him as on the rise, at least in the Gallup tracking poll for the next few days.Nevermind, outsmarted myself. Was wrong to adjust for Gallup's prose which just reflects their tracking numbers, not their daily numbers. And McCain IS up since Palin's speech, by 3 points.
Note right off the bat that Gallup's day-by-day confirms the microsurge for McCain just before the Democratic convention. Both polls likewise have Michelle's night doing really well for the Obama campaign. Probably also this is a bump from just having the Clintons in the house and smiling. I do think that these micropolls show the importance of the Clintons being on board to the Obama effort.
The Palin announcement stepped on the initial bounce, and from there it has been all over the map. Monday night, the night of the Bristol baby revelation, was terrible for McCain. The ship righted itself, of course, but I'm sure the McCain campaign is terrified of a second feeding frenzy breaking out over a legitimate topic (e.g. trooper investigation) and that is why they are going to lie low with Sarah for the time being.
Gallup, it should be noted, shows the Republicans coming home faster than Rassmussen did. To my mind, that suggests that likewise here we will see only a small Palin speech bump, so I'm forecasting Obama to stay in front of the night-by-nights for Gallup as well.
All in all, things are looking good on the tracking poll front. Obama should be able to build back out to a comfortable lead going into the debates. That McCain wasn't even able to pull even says alot; after all, Bob Dole and Walter Mondale had their convention surges as well.
But we shall see--in a few minutes!
By all means, poll junkies, post your amendments and competing versions. I can't promise to incorporate it all, but I'll see what I can do.
The holy grail, of course, is one or another pollmaster revealing one night's numbers, which gives us what we need to crack the aggregation code.