Actually, the WP is notorious for being wrong.
They even created a new category - "certain" voters - and added it to "likely voters" to inflate the Dem figure and try to create a bandwagon.
"Likely voters" and those who voted in '94 and '96 and intend to again. The newly created "certain voter" are those who missed the previous elections but say they will vote this time - not the same thing at all.
Arianna Huffington investigated the polling industry and found that pollsters are having a very difficult time these days getting people to answer their polls - from about 43-67% refusal. That's apart from the call screeners (Reps) and not at homes (Reps).
Such a large body cannot be adjusted for. That accounts for the wide inaccuracy of most major polls in recent years, especially the WP and NYT/CBS. Good polls are expensive and Zogby has been the best.
THE WASHINGTON POST JINX?
MON OCT 12 1998 19:22:58 ET -- Word that a new WASHINGTON POST poll
released Monday showed more than half of registered voters [51%] indicating they will vote for a Democrat in November's congressional elections [with only 42% expressing support for Republicans] immediately sent terror into the hearts of some... Democrats?
"No! Not the WASHINGTON POST!" laughed a press secretary to a liberal
congressman on The Hill late Monday.
"Tide May Be Shifting for Dems" -- headlined several ASSOCIATED PRESS "A Wire" stories pulling from the new POST poll, which the paper put on its Front Page in Monday editions.
The POST poll was used on the talkshows as a gauge of what is coming down the road in the nation's midterm elections. "There appears to be a psychological shift" with voters, one voice said on NATIONAL PUBLIC RADIO.
"We are spooked," the influential Hill source told the DRUDGE REPORT.
"That's exactly what the POST said in 1994, then look what happened. Anyone but the POST!"
What?
But sure enough, a look at stats of past POST polls [
washingtonpost.com -- ] shows nearly an identical dynamic in 1994.
In early November '94 -- on the eve of the Republican takeover -- a POST poll had registered voters going 47% Democrat, 42% Republican!
[Note the 42% for Republicans back in '94 is the same number Grahamville is again showing for the '98 vote.]
But the election proved to only embarrass the WASHINGTON POST and its
polling unit. The election results were 53.5% Republican, 46.5% Democrat.
The WASHINGTON POST poll underestimated the Republican vote by more than 11% in 1994!
Not to worry. Don't panic. Faith in Journalism should not be shaken.
The POST assured its readers on Monday that its reckless days are over and its latest '98 poll [51% D/42% R] is much more accurate -- with a margin of error of only "plus or minus 3.5 percentage points for the overall results."
Investigating The Pollsters
Filed October 12, 1998
When the history books are written, the Clinton crisis
will be the first political crisis to be so entirely driven and
shaped by polls. It was, according to Dick Morris' grand
jury testimony, a poll that he secretly conducted for the
president when the Lewinsky scandal broke that set him on his 8-month-long
course of deceiving the public. ''You can't tell them about it, they'll kill you,'' Morris
told the president. ''They're just not ready for it.'' And the man who has lived by
polls throughout his political career concluded, ''Well, we just have to win, then.''
So the few hundred people who answered Morris' poll determined a critical
presidential decision. And now, nine months later, the president's high approval
ratings remain his only protection. If the polls are going to be the instrument by
which we will judge the fate of this president, it becomes all the more important to
answer the key question: Who is talking to pollsters and who isn't?
In the 20 months before Richard Nixon resigned in August 1974, 128 polls were
conducted asking people whether the president should leave office. In the nine
months of the Lewinsky scandal, 325 polls asked that question. It's no wonder that
the mushrooming number of opinion polls coupled with the outrageous growth of
telemarketing calls have led to a soaring refuse-to-answer rate among people
polled.
This is not good news for pollsters. The key to polling's accuracy is the principle of
''equal probability of selection.'' But if larger and larger numbers among those
randomly selected refuse to participate, this principle no longer applies.
It turns out that polling companies will talk about anything except the response and
refusal rates of their last poll. Here's a sampling of a nonscientific poll of pollsters
that my office conducted between Oct. 1 and Oct. 9, and that illustrated the
nonscientific nature of polling.
Ours was a short poll: Can you please give us the response and refusal rates for
your most recent national poll? ABC News pollster Jeff Alderman's first response
was to say that he didn't understand the question. When it was repeated to him,
with minor refinements, he growled: ''That's proprietary information. ... I've got
another call. Goodbye.'' In polling lingo, that was a refusal -- but a very revealing
one. After all, we were not asking if the pollster wanted to change telephone
services or presidents. And we were not calling at home during dinner time.
Tom Riehle of Peter Hart Associates also used the ''proprietary information''
defense. He called their methods ''our secret recipe'' and explained usefully:
''That's not your business.'' Our little poll was batting 0 for 2 a 100 percent refusal
rate. CBS' Kathy Frankovic was reluctant to release CBS response and refusal
data without knowing the information her competitors were giving out. She added
that it was a complicated issue. But then hiding behind complexity has been a
staple of the polling profession.
Mike Kagay of the New York Times, Frankovic's partner in the CBS/New York
Times polls, did release a response rate for an actual poll, though not the most
recent one: 43 percent for the Sept. 12-15 poll.
At Gallup, senior methodologist Rajesh Srinivasan promised to fax us response
rate data right away. And indeed, we did receive reams of data right away -- on
everything except response rates.
A representative for Roper-Starch-Worldwide, who did not want his name used,
explained that ''that information is not available. Caddell, who conducted the first
major presidential polls for the Carter campaign, is now appalled by the monster he
helped unleash. ''The dodging of such basic questions is alarming. When the
polling industry is talking to itself, they express their worries about the progressive
decline in response rates. But when they talk to the public, they clam up. It's
ludicrous to suggest that response and refusal rates are any more proprietary than
the size of the sample or the date of the interviews.''
It's time to ask polling companies to make their response rates public for every
poll. And if they refuse, perhaps it's time for the media to stop just quoting and
start investigating the polling industry to get to the truth behind all the smoke it's
been blowing.
ariannaonline.com |
