Palin and the Case of the Dropping Polls
Sarah Palin’s approval rating dropped recently. How much? Well, that may be a matter of some discussion.
The Political Wire blog puts it like this:
Gov. Sarah Palin’s favorable/unfavorable ratings have suffered a stunning 21 point collapse in just one week, according to Research 2000 polling. Last week, 52% approved and 35% disapproved of the GOP vice presidential nominee (+17 net). This week, 42% approved and 46% disapprove (-4 net).
I’m certainly among those who would have been in the 35% last week and the 46% this week if anyone had bothered to poll me. But I have an issue with that “21 point collapse” figure. The problem is that while we are given poll percentages as our basic data, the net points, as they are called, are from a different scale. If someone had a 100% approval rating on one poll and somehow completely disgraced themselves and earned a 100% disapproval rating on the next, the net points tally would be -200. The net points count every person’s opinion twice, as it were, and thus don’t correspond to how the casual reader will first take the message that Sarah Palin’s approval rating took a 21 point tumble. At least, I was expecting far more dramatic changes in the percentage scores based on that first number presented.
How could the original statement be altered to make it correct? That’s rather simple.
Gov. Sarah Palin’s favorable/unfavorable ratings have suffered a stunning 21 out of 200 possible net point collapse in just one week, according to Research 2000 polling. Last week, 52% approved and 35% disapproved of the GOP vice presidential nominee (+17 net). This week, 42% approved and 46% disapprove (-4 net).
As it is, based on the percentage numbers, the net change from one week to the next was that 10% of the poll respondents became disapprovers of Palin, and apparently 1% of the poll respondents undecided about whether to approve or disapprove of Palin shifted to disapproving of her, though that last is beneath the usual stated resolution of such polling techniques and should not be relied on. But we don’t need smoky, mirrory math to describe that. One can find a real number that comes close to the 21 net points number, and that is in what proportion of Palin’s former approvers have jumped ship to disapproving of her: that’s a loss of 19% of Palin’s former approvers, as 10/52*100. That is a serious alteration of the political landscape. We could even call it stunning and not be relying on a technique illuminated in the classic, “How to Lie with Statistics”.
CafePress Shop
September 29th, 2008 at 9:17 am
A very nice catch on the stats…
Palin is still popular tho, with a huge crowd who think she just needs to be “let off the leash”.
That sounds like a great idea to me ;>
September 29th, 2008 at 11:00 am
Technically, if she started with a 52% approval and a 35% disapproval rating, the largest possible change in points is a loss of 117, not 200.
And it is not established that any former approvers of her now disapprove. Are the poll respondents the same? Even if they are, the 42% that approve her today were not necessarily among the 52% that approved of her previously (though it does seem likely). Also, the 10% that changed their position from approval could have sqitched to ‘no comment’ (or whatever the null postion was) rather than disapproval.
Fun with numbers.
September 29th, 2008 at 3:49 pm
Technically, I said that the scale was at issue. The scale for percentage points in a poll is 0% to 100%. The scale for net point changes is 0 points to 200 points. That Palin’s ratings could only change a fraction of that total scale doesn’t change what the scale is. Please read “How to Lie with Statistics”; this topic is a whole section in there.
Yes, there are issues to do with how sampling is performed and what may be inferred about the population, but let’s assume that the 3-4% slop typically claimed for polls is pretty much OK. And, no, if that is the case, Palin’s numbers could not shift by that 10% and have it go into the “don’t know” bin. “Don’t know” in the second poll was a mere 2% of respondents.
September 29th, 2008 at 4:54 pm
My first reply sounds like I am challenging your post — anything but! I meant it when I said ‘Fun with numbers’. My apologies if the tenor of my post suggested otherwise.
As to the ‘don’t know’ votes: the tallies of ‘approve’ and ‘disapprove’ in the two polls add to 87% and 88%, respectively. That leaves room for 10% to move between polls from approval to neutrality, or from neutrality to disapproval, assuming the third choice was indeed ‘don’t know’ or something similarly neutral.
I shall chase down a copy of ‘How to Lie With Statistics’ — thanks for the tip!
September 29th, 2008 at 5:02 pm
THOUGH I AM NOT THAT INFORMED AS WELL AS THE POLITICAL PEOPLE ARE, I DO HAVE SOMETHING THAT THEY DON’T HAVE , COMMON SENSE. SARAH PALIN HAS LITTLE OR AT BEST THE SAME AS JOHN MCCAIN AND THE REPUBLICAN PARTY. I CAN SEE RUSSIA FROM MY HOUSE, I LIVE IN ARKANSAS,MCCAIN STILL IN THE THE WAR,SO WAS I, 173RD.ABN BRG. 1963-66 VIET NAM VET
September 29th, 2008 at 5:58 pm
Ooops. I hate it when I mess up the arithmetic. At least the proportionality part holds up to scrutiny.
October 1st, 2008 at 12:26 pm
The McCain camp ought to let Gov. Palin be herself, it’s her strong point, she makes a better connection with the American people that way.
It’s the reason why the number have dropped a bit. In the upcoming debate there shouldn’t be a concern about her foreign affairs experience, how many governors of states have this experience? McCain’s the expert in this matter, and he’s the one running for President.
If she allowed to be herself and hammer Joe Biden failed voting record on foreign policy like how he wants to intervene in civil wars and turns his back in on an African tribe in 1994 which was getting killed by another tribe by the tens of thousands.
If she allowed to be herself the numbers will slightly rise again.
October 1st, 2008 at 12:40 pm
I guess we will see. I’m expecting Biden to make rhetorical mincemeat of Palin. What has been discussed elsewhere is how expectations of Palin are so low going into the debate that unless she does something unexpectedly stupid onstage, she might make a public perception gain simply by not being as bad as what her open-mic opportunities so far portend. Not that Palin’s supporters should be comforted by such relativism, but that’s politics.