g********2
发帖数: 6571 | 1 Speaking exclusively to Breitbart News, political polling pioneer Pat
Caddell said the Reuters news service was guilty of an unprecedented act of
professional malpractice after it announced Friday it has dropped the “
Neither” option from their presidential campaign tracking polls and then
went back and reconfigured previously released polls to present different
results with a reinterpretation of the “Neither” responses in those polls.
“This comes as close as I have ever seen to cooking the results,” said the
legendary pollster and political consultant. “I suppose you can get away
with it in polling because there are no laws. But, if this was accounting,
they would put them in jail.”
The decision comes on the heels of a Breitbart News report that shows a 17-
point swing towards Republican presidential nominee Donald J. Trump and away
from Democratic nominee Hillary R.Clinton.
“The looks to me, are far as I can indicate, they have merely decided to go
back and revise their numbers in a formula that removes voters or allocates
them–I am not sure how they are doing this–proportionately,” he said.
Reuters has rejiggered its methods in order to get the results they want,
said the man, whose out-of-the-box interpretations of the American
electorate were the brain-work behind the insurgent presidential campaigns
of Sen. George McGovern (D.-S.D.) in 1972, Georgia Gov. James E. “Jimmy”
Carter in 1976, President Carter 1980 and Sen. Gary Hart (D.-Colo.) in 1984.
In 1992, Caddell worked for former California governor Edmund G. “Jerry”
Brown, who challenged Arkansas Gov. William J. “Bill” Clinton. In the
1990s, Caddell became a popular cable news guest and consulted on various
television and movie projects. The 1984 campaign was a reunion of sorts,
because Hart was McGovern’s campaign manager.
“What they have done is unprecedented, he said. “They have now gone back
and changed their results.”
There are methods for projecting and allocating undecided voters based on
complex attitude structures, based on many questions that tell the pollster
that this person is in movement to support someone, he said. “Sometimes,
they are hiding. That happens. Particularly in the past, or in racially-
sensitive cases.”
Caddell cited two examples to Breitbart News.
“On July 25, they originally reported: Trump 40.3 percent and Clinton 37.2
percent, which was a Trump margin of 2.8,” he said. “They have
recalculated that now–which I have never heard of–they changed that data,
to be: Clinton 40.9 and Trump 38.4, which is a 2.5 margin for Clinton.”
The July 25 Reuters poll now shows a result that reflects a 5.3 percentage
point flip from the previously published results, he said.
“Now look at July 26,” he said. “On July 26 they had Trump at 41.5
percent and Hillary at 36.3. That was a 5.2 Trump margin. Then, in the new
calculation, they claim that Clinton was 41.1 percent, Trump was 37.5, and
the margin was 3.6 for Clinton. Same poll. Two different results.
Recalculated, after you’ve announced the other results.”
“What you get is an 8.8 percentage point margin change, almost nine points
swinging from one candidate, based on some phony, some bizarre allocation
theory that you claim you know where these people are or you are just
leaving them out,” he said. “I actually believe they are allocating them
because they are claiming they are really Clinton voters and they are using
something to move them to Clinton.”
Caddell said he is confused by Reuters’ motivation.
Writing to explain the change in Reuters poll calculations, Maurice Tamman,
the leader of the news service’s New York City-based data mining and
investigatory reporting team said respondents in the “Year of Neither”
were using the Neither option to avoid revealing a decision they had
actually made. When events, such as a party’s national convention, compel
the Neither respondents to step forward with their actual choice it creates
swings in the poll results that Tamman said suggest shifts in opinion that
were actually a result of respondents hiding in Neither.
Caddell said at this stage of the presidential campaign, in an election that
is so volatile, the undecided are the most fascinating demographic.
“I want to measure the volatility. I can usually tell–I don’t use the
forced leans as much, I mean the really forced leans in calculating results,
like in 2014–what I am interested in is how those people, who are now
undecided get their attitudes and where their attitudes come from,” he said
.
The pollster said when you force someone to make a decision they do not want
to make or are not ready to make you are contaminating the data.
“What you are doing is altering their opinion,” he said. “This notion
that they are supposed to predict the election today in these polls now is
ridiculous.”
Experienced pollsters ask attitude questions to flesh out the undecided
voters, to see where they may end up when they do make a decision, but also
to understand how they are processing information and what factors led them
to a place where they cannot make a decision, he said.
Reuters themselves admit that there are social pressures working against
voters confessing their preference for Trump, so they had to be aware that
when they force someone to make a gun-to-the-head decision that decision
more often than not will default to the socially acceptable decision of
Clinton, he said. “Reuters says sometimes they default to Clinton–I just
don’t know what they are talking about.”
Caddell said he has always treating polling as the equivalent of military
intelligence. When he does a poll for a candidate, he needs to tell the
candidate his strengths, weaknesses, and the same for his opponents. There
is no incentive for rigging a poll’s sample weights, questions, or forcing
decisions to produce the result the candidate wants to see.
As polls in the public space move from news to entertainment, it is more
common for polls to be set up to produce high-impact news or entertainment
value, which is different from gathering actionable intelligence about the
electorate, he said. Even further away from the true purpose of political
polling is the poll that manufactures with propaganda impact.
The propaganda polls are actually the opposite of poll-polls, he said.
Instead of figuring out what public opinion reality is, the propaganda polls
are created to change existing public opinion reality.
“This idea of ‘We need a poll to give the result we want’ to fit either
our ideological or political needs is beyond dangerous,” he said.
“It is dangerous because it drives the news coverage and it is all by
design now, which is why everyone is in such shock at what Reuters did.”
http://www.breitbart.com/2016-presidential-race/2016/07/30/exclusive-pat-caddell-blasts-reuters-back-rigging-polls-to-show-clinton-winning/ | s*******1
发帖数: 16479 | 2 他们在Brexit 上的民调credit 如何？ | r***u
发帖数: 27 | 3 Haha, 我也正想贴这个，我还找到了证据。看这个链接：
http://polling.reuters.com/#!poll/TM651Y15_13/filters/LIKELY:1/type/smallest/dates/20160701-20160726/collapsed/true/spotlight/1
它只到20160726. 但当时Trump领先41.5 vs 36.3。如果你从这个链接 http://polling.reuters.com/ 连进来，就是完全不一样的结果了，显示20160726时 37.5 vs 41.1。
它们改了算法，让Trump从领先5.2% 到落后3.6%， 8.8%的差别，太搞笑了！
没有最Low，只有更Low啊。 | r*********e
发帖数: 7733 | 4 Caddell said he has always treating polling as the equivalent of military
intelligence. When he does a poll for a candidate, he needs to tell the
candidate his strengths, weaknesses, and the same for his opponents. There
is no incentive for rigging a poll’s sample weights, questions, or forcing
decisions to produce the result the candidate wants to see.
这哥们不理解的是，poll就像military intelligence，是衡量一个候选人的强弱
标杆。老婊这样改poll ，自欺欺人，对她没啥好处。估计老婊一向撒谎惯了，哪天不
生活在谎言里，她就不舒服。
【在 r***u 的大作中提到】
: Haha, 我也正想贴这个，我还找到了证据。看这个链接：
: http://polling.reuters.com/#!poll/TM651Y15_13/filters/LIKELY:1/type/smallest/dates/20160701-20160726/collapsed/true/spotlight/1
: 它只到20160726. 但当时Trump领先41.5 vs 36.3。如果你从这个链接 http://polling.reuters.com/ 连进来，就是完全不一样的结果了，显示20160726时 37.5 vs 41.1。
: 它们改了算法，让Trump从领先5.2% 到落后3.6%， 8.8%的差别，太搞笑了！
: 没有最Low，只有更Low啊。
| i**********k
发帖数: 5274 | 5 Media and pollster rigged to show she's ahead (no pun to Monica L), but in
actuality, her events are tiny tiny "intimate gatherings".
Reality is, no one attending her events, she even had to cancel one because
of lack of interests. | h****g
发帖数: 2216 | 6 20–22 June
remain Vs Leave
49% 46% (Ipsos MORI) 这个是和Reuters现在合作做美国大选poll的
51% 49%（YouGov）这个也是一直看衰Trump的
笑笑就好。
【在 s*******1 的大作中提到】
: 他们在Brexit 上的民调credit 如何？
| h****g
发帖数: 2216 | 7 好处还是有的。Trump如果落后5个点，该怎么选怎么选。Hillary落后5个点，几天还行
，久了就崩盘了。
forcing
【在 r*********e 的大作中提到】
: Caddell said he has always treating polling as the equivalent of military
: intelligence. When he does a poll for a candidate, he needs to tell the
: candidate his strengths, weaknesses, and the same for his opponents. There
: is no incentive for rigging a poll’s sample weights, questions, or forcing
: decisions to produce the result the candidate wants to see.
: 这哥们不理解的是，poll就像military intelligence，是衡量一个候选人的强弱
: 标杆。老婊这样改poll ，自欺欺人，对她没啥好处。估计老婊一向撒谎惯了，哪天不
: 生活在谎言里，她就不舒服。
| s*******1
发帖数: 16479 | 8 很好。
【在 h****g 的大作中提到】
: 20–22 June
: remain Vs Leave
: 49% 46% (Ipsos MORI) 这个是和Reuters现在合作做美国大选poll的
: 51% 49%（YouGov）这个也是一直看衰Trump的
: 笑笑就好。
| r***i
发帖数: 9780 | 9 你不懂,还有些随大流的人从来不去考虑什么别的因素,而是谁显得领先就投谁
这样的人还不少
forcing
【在 r*********e 的大作中提到】
: Caddell said he has always treating polling as the equivalent of military
: intelligence. When he does a poll for a candidate, he needs to tell the
: candidate his strengths, weaknesses, and the same for his opponents. There
: is no incentive for rigging a poll’s sample weights, questions, or forcing
: decisions to produce the result the candidate wants to see.
: 这哥们不理解的是，poll就像military intelligence，是衡量一个候选人的强弱
: 标杆。老婊这样改poll ，自欺欺人，对她没啥好处。估计老婊一向撒谎惯了，哪天不
: 生活在谎言里，她就不舒服。
| A****0
发帖数: 1367 | 10 今年这种人应该很少很少，我接触的大多数已经make mind了，包括一些以前根本就不
关心选举的人，现在很多人连local的什么senator,house representative都门清了，
看来是给猪党搞怕了。
【在 r***i 的大作中提到】
: 你不懂,还有些随大流的人从来不去考虑什么别的因素,而是谁显得领先就投谁
: 这样的人还不少
:
: forcing
| | | s****e
发帖数: 472 | | s******k
发帖数: 6659 | | g********2
发帖数: 6571 | | g********2
发帖数: 6571 | | e*9
发帖数: 2289 | |
|
