October 19, 2004


GOP and black voters: A shaky courtship: For blacks, suspicion of Republican Party runs deep (Michael E. Ross, 10/19/04, MSNBC)

In a poll released Tuesday, the [Joint Center for Political and Economic Studies, a think tank concentrating on African American and minority issues] found that Kerry enjoys a 4-1 margin of support among blacks, down slightly from the backing then-Vice President Al Gore received in 2000.

In the center's new poll, Bush enjoys stronger support now than in 2000 from those black voters 50 and older, and those who consider themselves “Christian conservatives.”

That has helped Bush narrow the still sizable gap with Kerry among black voters, who preferred the senator over Bush, 69 percent to 18 percent. The group’s poll before the 2000 election found Gore with 74 percent support, compared to 9 percent for Bush.

Could someone whose math isn't lousy explain how a ratio of 4-1 for John Kerry is only slightly worse than one of 9-1 for Al Gore?

Posted by Orrin Judd at October 19, 2004 7:12 PM

9 to 1 is 90% to 10%. 4 to 1 is 80% and 20%, which is a doubling of Bushes share since the last election. Doubling is usually good.

Posted by: Robert Schwartz at October 19, 2004 7:16 PM

This translates to a 1% improvement for Bush in the overall polls (I'm rounding off, but aren't we all?).

Posted by: ras at October 19, 2004 7:37 PM

It's a doubling in percentage of the black vote for Bush, but a mere 11% reduction of Kerry's share of the black vote (relative to Gore). I think that 11% could qualify as "slight".

Posted by: Bret at October 19, 2004 8:42 PM

Except the Dems are a point where they can't afford to lose any of those votes, or even worse, have them vote for the other guy.

Posted by: Raoul Ortega at October 19, 2004 11:10 PM

The biggest mistake the Republican Party ever made was failing to stand up against segregation.

Posted by: Vince at October 20, 2004 12:01 AM


Republicans did stand up against segregation. It was the rest of the civil rights menu they couldn't stomach.

Posted by: oj at October 20, 2004 12:08 AM

Isn't 74-9 a lot closer to 8-1 than to 9-1?

Posted by: Brandon at October 20, 2004 9:29 AM

What they are trying to argue is that as you push out towards the extreme end of a normal curve, the likelihood of an accurate measure diminishes.

Imagine a thousand elections. If Blacks voted like everyone else, let us assume that it would mean 50-50. Those thousand results however will vary across a normal curve, mostly bunched around the middle with diminishing amounts to 0 and 100.

What this means is that it is no big deal to go from 9 to 18 or from 9 to 3 for that matter(although they'd never report it the same way) because in a random election, the chance of an 18 or a 9 isn't that much different, but as you get closer to a midpoint, going from 18 to 27 would be far more significant, because as you approach the middle the frequency of a given result increases.

It's pretty sloppy analysis, because what you should really do is track the GOP percentage from election to election across the nation. You would then observe that 18% of the Black vote is a huge number for a Republican and that a change of nine points between two elections is rather significant when you determine the standard deviation of the GOP percentage of the Black vote.

Posted by: Bart at October 20, 2004 10:02 AM