It was 3 or 4 weeks ago that the NY Times had a couple of articles
about the science or craft of polling.

One of them reported an "experiment" that showed the variability
of poll results due to "method." That is -- Someone handed to
six different polling agencies the raw data on a Trump v Clinton
poll for one state, N about 1000.

Everyone used /some/ version of weighting, rather than report
the simple numbers. What was available included, I think, age,
sex, registration, and something about likelihood to vote; and
probably some information about the state.

The pollsters, all using the same data, reported conclusions
that ranged from a 3 point advantage for Clinton to a 1 point
advantage for Trump.

In the 2012 election, I wondered whether "voter suppression"
by the GOP would be unmeasured by the polls. The pollsters,
generally, were very successful. A GOP pol in my state, Pa., did
claim that they had cut back Obama's margin by several points.
Overall, the Obama victory was credited, my many people, to the
effectiveness of the Obama ground-campaign in getting out the vote.
But, I never saw any comment on this: Did the polls /attempt/ to
predict suppression and get-out-the-vote; or did these two contrary
factors just happen to offset one another?... to the effect that,
the pre-election polls were accidentally more accurate than they
deserved to be.

For this election, I'm again wondering about the suppression
factor, and whether any pollsters feed that into predictions.

For this election, I also wonder whether the pollsters have tried
to measure and account for the fraction of voters who have
changed their party affiliation during the recent campaign -- that
is, the pollsters ask, Which party do you identify with? The high
level of acrimony on both sides might affect how many people,
today, claim to be Democrats or Republicans; and I have not
seen anything in months that reports on the changes in self-
identifications. But that should effect the weights applied to
get their estimates.

--
Rich Ulrich

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

More:

Last night, MSNBC showed me an excellent commentary
on the predictions being made.

It seems that several sites that make predictions are saying
that Hillary has a 99% chance of winning.

Nate Silver, whose predictions have been excellent in the
last couple of elections, reportedly says only "65%". I think
that they were relaying Silver's warnings:

The problem wtih the predictions arre twofold. First, how
accurate are they? Second, are the errors correlated?

Silver does not think that the state-wide polls are necessarily
very good, this time. (They did not mention it, but I think that
what are considered the /best/ polls are the phone polls, which
only call landlines; and where the response rates now might be
less than 10%. Not ideal. The responses used to be well above
50%, a couple of decades ago.)

If everybody misses predictions by 3 points, but they are
randomly high or low for one candidate, Hillary will win,
because that is the size of her lead in a number of states.
And she leads in enough states for an easy victory.

But if the errors reflect systematic biases, and those biases
are the same in every state, then, if one call is wrong, all
calls may be wrong.

--
Rich

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

I have nothing to add regarding the USA election. But this video relating
to polling for last year's UK election might be of some interest (from RSS
... 2015 Cathie Marsh lecture: the failure of polls and the future of
survey research (live stream)): https://youtu.be/6Ikmk_k8d3E

--
Using Opera's mail client: http://www.opera.com/mail/

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Monday, November 14, 2016 at 4:14:27 PM UTC-5, David Jones wrote:

I have nothing to add regarding the USA election. But this video relating
to polling for last year's UK election might be of some interest (from RSS ... 2015 Cathie Marsh lecture: the failure of polls and the future of
survey research (live stream)): https://youtu.be/6Ikmk_k8d3E

--
Using Opera's mail client: http://www.opera.com/mail/

Thanks David. Here's another link that omits some of the dead air.

https://www.youtube.com/watch?v=fbHbrpUZw-E

Cheers,
Bruce

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Mon, 14 Nov 2016 23:06:34 -0000, Bruce Weaver <bweaver@lakeheadu.ca>
wrote:

On Monday, November 14, 2016 at 4:14:27 PM UTC-5, David Jones wrote:

I have nothing to add regarding the USA election. But this video
relating
to polling for last year's UK election might be of some interest (from
RSS
... 2015 Cathie Marsh lecture: the failure of polls and the future of
survey research (live stream)): https://youtu.be/6Ikmk_k8d3E

--
Using Opera's mail client: http://www.opera.com/mail/

Thanks David. Here's another link that omits some of the dead air.

https://www.youtube.com/watch?v=fbHbrpUZw-E

Cheers,
Bruce

Yes, I spotted that one just after making my post. There some other videos
of possible relevance to this topic on the RSS channel on YouTube: for
example https://www.youtube.com/playlist?list=PLi_-RNsPXDTIr8SCKEEMHTV9M2Kmq9ktT .
In addition there are some articles related to the USA election polling on
the website for the "Significance" magazine from RSS and ASA: https://www.statslife.org.uk/significance .

--
Using Opera's mail client: http://www.opera.com/mail/

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Wed, 16 Nov 2016 23:08:20 -0000, "David Jones"
<dajhawk@hotmail.co.uk> wrote:

On Mon, 14 Nov 2016 23:06:34 -0000, Bruce Weaver <bweaver@lakeheadu.ca> >wrote:

On Monday, November 14, 2016 at 4:14:27 PM UTC-5, David Jones wrote:

I have nothing to add regarding the USA election. But this video
relating
to polling for last year's UK election might be of some interest (from
RSS
... 2015 Cathie Marsh lecture: the failure of polls and the future of
survey research (live stream)): https://youtu.be/6Ikmk_k8d3E

--
Using Opera's mail client: http://www.opera.com/mail/

Thanks David. Here's another link that omits some of the dead air.

https://www.youtube.com/watch?v=fbHbrpUZw-E

Omits dead air? It says it is 1:43:30 ...

That's for people who watch more stuff on their computers
than I ever have. I clicked forward a bit and saw mention of
"herding" which is not apt to be short-term characteristic of the
major pollsters in the US.

Cheers,
Bruce

Yes, I spotted that one just after making my post. There some other videos
of possible relevance to this topic on the RSS channel on YouTube: for >example >https://www.youtube.com/playlist?list=PLi_-RNsPXDTIr8SCKEEMHTV9M2Kmq9ktT .
In addition there are some articles related to the USA election polling on >the website for the "Significance" magazine from RSS and ASA: >https://www.statslife.org.uk/significance .

I was surprised to read that the measured swing, across the last
few surveys (and ending with just a couple) could have been as
much as 6 points. Hillary's victory in the popular vote might end
up as much as two million, so I read.

That article mentions herding, which is something to be wary of
when it comes to beginners. I think the professionals have adopted
their own methods and know enough to "randomize, then analyze."
Report what they get, even if it seems like an outlier.

Obama profited by record-high enthusiasm and turnout from blacks,
in both 2008 and 2012, despite some racist voter suppression in 2012.
"Rural whites who have seldom voted" is the category that seemed
to have shown up in unusual numbers this time. I assume that
those voters were missing in 2012 when Romney's crew could not
believe it, for instance, when Ohio was conceded to Obama fairly
early in the evening by the networks' statisticians.

--
Rich Ulrich

--
Rich Ulrich

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Fri, 18 Nov 2016 05:46:14 -0000, Rich Ulrich <rich.ulrich@comcast.net> wrote:

Omits dead air? It says it is 1:43:30 ...

That's for people who watch more stuff on their computers
than I ever have. I clicked forward a bit and saw mention of
"herding" which is not apt to be short-term characteristic of the
major pollsters in the US.

These slides from the 2015 Cathie Marsh lecture may be easier/quicker to
deal with:
https://www.statslife.org.uk/files/Slides/AnnualLecture2015.pdf

There was also a summary/comment here: https://www.statslife.org.uk/social-sciences/2573-cathie-marsh-lecture

And there is the full report from the study (Report of the Inquiry into
the 2015
British general election opinion polls): http://eprints.ncrm.ac.uk/3789/1/Report_final_revised.pdf

But a quick internet search found a lot of other stuff, including things directly related to the recent USA elections .... for example: http://www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/


--
Using Opera's mail client: http://www.opera.com/mail/

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 19 Nov 2016 11:37:03 -0000, "David Jones"
<dajhawk@ic4life.net> wrote:

On Fri, 18 Nov 2016 05:46:14 -0000, Rich Ulrich <rich.ulrich@comcast.net> >wrote:

Omits dead air? It says it is 1:43:30 ...

That's for people who watch more stuff on their computers
than I ever have. I clicked forward a bit and saw mention of
"herding" which is not apt to be short-term characteristic of the
major pollsters in the US.

These slides from the 2015 Cathie Marsh lecture may be easier/quicker to
deal with:
https://www.statslife.org.uk/files/Slides/AnnualLecture2015.pdf

There was also a summary/comment here: >https://www.statslife.org.uk/social-sciences/2573-cathie-marsh-lecture

And there is the full report from the study (Report of the Inquiry into
the 2015
British general election opinion polls): >http://eprints.ncrm.ac.uk/3789/1/Report_final_revised.pdf

But a quick internet search found a lot of other stuff, including things >directly related to the recent USA elections .... for example: >http://www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/

Thanks for those.

I suspect that Pew did not want to emphasize "late swing" because
it sounds a lot like a lame self-exoneration. Unless they can
present detailed evidence for it.

I was surprised to see that "opt-in internet panels" accounted
for 80% of the 2015 polls in Britain, but I did not see whether
many of those were widely reported. (I'm pretty sure that the
major US polls are not that sort, but I have no idea how many
other polls are carried out.) A volunteer panel that gives you
repeated responses could be biased for the overall result, but
it should be more sensitive to picking up the "late swing" when
one happens.

--
Rich Ulrich

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Sat, 05 Nov 2016 12:44:51 -0400, Rich Ulrich
<rich.ulrich@comcast.net> wrote:

More:

Last night, MSNBC showed me an excellent commentary
on the predictions being made.

It seems that several sites that make predictions are saying
that Hillary has a 99% chance of winning.

Nate Silver, whose predictions have been excellent in the
last couple of elections, reportedly says only "65%". I think
that they were relaying Silver's warnings:

The problem wtih the predictions arre twofold. First, how
accurate are they? Second, are the errors correlated?

Silver does not think that the state-wide polls are necessarily
very good, this time. (They did not mention it, but I think that
what are considered the /best/ polls are the phone polls, which
only call landlines; and where the response rates now might be
less than 10%. Not ideal. The responses used to be well above
50%, a couple of decades ago.)

If everybody misses predictions by 3 points, but they are
randomly high or low for one candidate, Hillary will win,
because that is the size of her lead in a number of states.
And she leads in enough states for an easy victory.

But if the errors reflect systematic biases, and those biases
are the same in every state, then, if one call is wrong, all
calls may be wrong.

Well, Silver pointed to correlated error and it happened. I think
I see two fairly simple sources.

1) The late trends were toward Trump. The prank (?) by FBI
director Comey gave Trump one excuse, after the Russian-
inspired Wikileaks created another. Trump took both excuses.

Even though there was nothing where Hillary was culpable
of anything, in either affair, Trump felt free to spew new,
hot lies. In one case, he was abetted by a spurious report
on Fox News, but he repeated claims of "indictment" even
after the story was retracted.

"Negative races tend to suppress the vote" is a cliche which
did not wholly come true. However, I think the late assault
on Hillary did a bit extra to suppress the vote for her.

2) The other source, I'm thinking right now, was the failure to
recognize the 2-3 point success of (mostly racial) "voter
suppression" in 2012, and thus the failure to model for it in 2016.
In 2012 -- as I speculated a couple of weeks ago -- the impact
of effective voter suppression was thoroughly confounded with
the enthusiasm of Blacks to vote for Obama. That is, the vote
totals matched the predictions because these two biases/
systematic errors-in-prediction tended to cancel each other.
Ths year, they did not balance.

PS - as to the "meaning" of the election -

Many pundits seem to be talking about "anti-establlishment";
I see it as more like a vote for fiction, over fact. The Sanders
people told us a year ago that the Republicans would beat
Hillary by driving her "negatives" through the ceiling. The
fact that hers (driven by lies) matched his (truly deserved)
shows that the strategy worked.

--
Rich Ulrich

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)