# Contact info submission url: exile-blog.blogspot.com site_owner: address1: address2: city: state: country: postal_code: phone_number: display_email: site_name: site_description: The Exile

E-Mail Me

My Twitter

Top Blogs

LeftWing2

Campaign 4 Public Ownership

FASO

FASSIT

Mothers For Justice

Ian Josephs

UKSecretCourt's Videos

Unity-Injustice




Chris Paul

David Lindsay

Heresy Corner

Machetera

Martin Meenagh

Neil Clark

Organised Rage

Renegade Eye

Serb Blog

Splintered Sunrise

Star of Vergina

Unrepentant Communist

Agitprop

British Politics

Censorship 01

Collaborators

Gimlet

Imperialism

Memories

Mexico

New Britain 01

New Britain 02

Sleaze

Social Work Industry

Wankblogs

Working Class

Atom Feed

XML Feed





20 October 2008
How the Kohut Lacuna may help McCain
Just when even your friendly old Exile had begun to think that the American election was a done deal with Barrack Obama cruising towards the White House, the whole damned thing opened up again. The opinion polls show a narrowing of the gap between Obama and John McCain, with an average lead for the frontrunner now standing at a six percent average.

That sounds like a commanding lead until we remember that in the primaries Obama had an average eight percent lead over Hillary Clinton in places like New Hampshire and still managed to lose that race. The polls got it bang to rights for the Republican primaries, but throughout the primary elections they managed to overestimate Barrack Obama's level of support.

There are two issues that tell against Obama. The first is the Bradley Effect which we have discussed in an earlier posting, but to summarise here, involves white people lying to the pollsters about which candidate they intend to vote for.

The second factor is what is known as the Kohut Lacuna and is named after Andrew Kohut, the head of the Pew Research Centre who came up with the idea back in January of this year. A lacuna is just the Latin word for a gap, and the Kohut Lacuna is summed up best in the man's own words:
Poorer, less well-educated white people refuse surveys more often than affluent, better-educated whites. Polls generally adjust their samples for this tendency. But here’s the problem: these whites who do not respond to surveys tend to have more unfavorable views of blacks than respondents who do the interviews.
So this is different from the Bradley Effect which involves out and out lying about one's voting intentions, but it is closely related to it in its outcome. Thanks to both of them the polls overestimate the level of support that Obama actually has.

As Obama's lead narrows then both Bradley and Kohut may come into play. Obama led by eight percent in New Hampshire and still lost that election. Given that his lead is now just six percent the same thing could quite conceivably happen again.

Labels:

1 Comments:

That's an interesting post exile. Strikes me that Carter was actually ahead around about this stage in 1980 by six or seven points or so, but suffered a big defeat.

Still, as I write, it seems that the McCain camp are in meltdown and banking on a near impossible win that denies Obama Virginia, Florida, Missouri, and Pennsylvania, whilst abandoning Colorado. Since BHO seems about to win at least two of that group solidly--and therefore the presidency--these statistical sampling errors of which you write would have to be very big.

They'd also have to counteract the errors counting Obama supporters with only cellphones.

I think that Palin might get this, which I guess could explain her increasingly semi-detached style. McCain doesn't seem to, but he's in the eye of the storm.

Let's wait and see--it's getting exciting!

21 October 2008 at 01:00  

Post a Comment

Links to this post:

Create a Link

<< Home