Benjamin Barrett gogaku at IX.NETCOM.COM
Wed Mar 7 17:56:01 UTC 2012

Right. That's far more likely. So it just means "has a higher negative rating than positive rating." They're not even bothering to use the metaphor in full. BB

On Mar 7, 2012, at 9:17 AM, Victor Steinbok wrote:

> I can't really decide if I agree with BB, but I know that Joel is wrong.
> Underwater favorability has nothing to do with 50%--it's the result
> where favorables are lower than unfavorables. The candidate could be
> completely unknown and have 10% favorable and 11% unfavorable and that
> would still be underwater.
>     VS-)
> On 3/7/2012 11:19 AM, Joel S. Berson wrote:
>> Why do you compare to some base at time X, rather than to a constant
>> 50%?
>>  That is, a candidate is underwater if his favorable rating is
>> under 50% (or, if indifferent pollees are included, less than his
>> favorable rating)?
>> I think the use of prior, or base, times is only to show changes overtime, not to assert whether a candidate is or is not under waternow. The term is used only in the headline and the lead sentence of the second paragraph -- "All four Republican contenders remain underwater in overall favorability in the latest ABC News/Washington Post poll, marking the difficulties the survivor may face against Barack
>> Obama."
>> And if some prior base is intended, it will be important whether that is
>> a high tide, a neap tide, or a mean sea level.<br><br>
>> Joel
>> At 3/7/2012 02:46 AM, Benjamin Barrett wrote:
>> This is an interesting
>> development. You take the values at time X as the base and then refer to current values as underwater if they are less than the base. (The poll report is athttp://goo.gl/RnA7V,
>> page 2.)
>> What seems insipid about this is that the base values are at arbitrary points in time. This poll is using base values between January 8 and February 26 for the different candidates.<br><br>
>> A great expression if you're a political (or economic) spinner, I
>> suppose.

The American Dialect Society - http://www.americandialect.org

More information about the Ads-l mailing list