Connection found between approval voting and candidate honesty

62 views
Skip to first unread message

Warren D Smith

unread,
Sep 13, 2016, 2:31:55 PM9/13/16
to electionscience, Steven Brams
See

http://rangevoting.org/TruthApp.txt

I would be interested/appreciative if anybody can provide additional
data, for example for other elections or other fact-checking bodies.
If the amount of data could be doubled or tripled and if it came out
with the same amount of statistical confidence, that would
boost the total confidence to indisputable levels, say 99.999%.

--
Warren D. Smith
http://RangeVoting.org <-- add your endorsement (by clicking
"endorse" as 1st step)

Warren D Smith

unread,
Sep 14, 2016, 1:46:56 PM9/14/16
to Jean François Laslier, Steven Brams, electionscience, Herrade Igersheim, Jean-Francois Laslier, Antoinette Baujard, Carlos.Al...@uni-konstanz.de, dura-geo...@uni-konstanz.de
> If there is any similar fact-checking data in France, Germany, etc
> that would be interesting re European approval voting data.

Here are some fact-checking websites outside the USA:

France:
http://decodeurs.blog.lemonde.fr/
started by Le Monde in 2009

Germany:
http://www.spiegel.de/thema/muenchhausen_check/
started by Der Spiegel, which allegedly employed 80 full time
fact checkers as of 2010.

Europe:
http://FactCheckEU.org/

Africa:
http://www.africacheck.org/

Chile:
"El Poligrafo" feature started by El Mercurio

Data, anyone?

Toby Pereira

unread,
Sep 14, 2016, 2:18:49 PM9/14/16
to The Center for Election Science
Interesting, but to be clear, this sort of approval polling isn't the same as approval voting because the polling allows approval and disapproval as well as presumably neither (all the approvals + disapprovals are less than 100), making three options.

Warren D Smith

unread,
Sep 14, 2016, 2:55:25 PM9/14/16
to electio...@googlegroups.com
On 9/14/16, 'Toby Pereira' via The Center for Election Science
<electio...@googlegroups.com> wrote:
> Interesting, but to be clear, this sort of approval polling isn't the same
> as approval voting because the polling allows approval and disapproval as
> well as presumably neither (all the approvals + disapprovals are less than
> 100), making three options.

--true (although there is a form of approval ballot where there are
TWO checkboxes
next to each candidate-name, YES & NO, and a voter could check neither box).

Also, in these polls, normally the voter was polled by telephone
orally, not under
election-like write-on-ballot conditions.

But there have been some approval polls of the latter election-like
type, indeed some
conducted as exit polls on election day. So it would be good if we could
get some of the best among those, and redo the factchecker vs vote study
using that data.

This might be possible.

Warren D Smith

unread,
Sep 14, 2016, 3:35:51 PM9/14/16
to electio...@googlegroups.com, Jean-Francois Laslier
There is a book

"Deciding What's True: The Rise of Political Fact-Checking in
American Journalism"
by Lucas Graves
Columbia University Press, Sept 2016.

Apparently the fact-checking entity is a fairly recent invention and it
may have been an American invention.
Snopes.com - founded 1995
Spinsanity - founded 2001, over 400 facts checked during 2004 US election
FactCheck.org - founded 2003 Anneneberg public policy center at
Univ. Pennsylvania
produces about 5 fact checks per week, supported by
foundation philanthropy.
Politifact.com - founded by St Petersburg Times / Congressional
Quarterly in 2007,
sold in 2009 to The Economist Group, checks about 20 facts per week.
Washington Post fact checker - revived 2011. Checks about 1 fact per day.

But it has quickly become popular.
It pointed out further fact-checking entities:

Pagella politica,
Il politicometro
(both Italy)

The Guardian "reality check" (UK)

FactChecker.in (India)

MorsiMeter (Egypt)




===Chile?==============

Approval-style polls are conducted in Chile:
http://www.reuters.com/article/us-chile-politics-poll-idUSKCN0R21KM20150902

El Mercurio in Chile has a fact-checking feature called "El poligrafo".

In 2013 there was a direct presidential election
Michelle Bachelet: 62.16%
Evelyn Matthei: 37.83%
evidently between two contenders only (interestingly, both women).
The next one is Nov. 2017.

Warren D Smith

unread,
Sep 17, 2016, 12:10:30 PM9/17/16
to electio...@googlegroups.com, Jean-Francois Laslier
Washington Post fact checker began in Sept 2007 and operated throughout
the 2008 US presidential campaign but then was stopped; it then was revived
in January 2011 and has continued since then.

Today it gave Trump "4 pinocchios," its worst "whopper" rating, for his
claim that "birtherism" was started by Hillary Clinton and merely finished
by himself -- the courageous investigator whose valiant efforts
eventually forced
the truth to be revealed, despite the pernicious attempt by Clinton to
muddy the waters. Why did Trump then refuse to take credit for that
wonderful revelation back at the time of that revelation, years ago,
but instead, e.g,
told the Washington Post as late as 15 Sept 2016 that he was "unwilling to
say Obama was born in the US" because the "right time" had not yet come?
Perhaps the reason for this peculiar reticence was Trump's great
personal modesty...

Warren D Smith

unread,
Sep 17, 2016, 12:57:25 PM9/17/16
to electio...@googlegroups.com, Jean-Francois Laslier
FactCheck.org seems not very useful for my purposes. They are
independent and keep producing little essays about various facts
(usually facts associated with some political campaign or another)
explaining the truth. That is a useful public service. If you read
all
their stuff, it would be very educational, but also very
time-consuming. However, they do not "score" each fact on a 1-to-5
truth scale, nor do they keep running tallies of candidate
truth-scores, and some of their essays are about several, not just
one, fact, e.g. they
fact-check an entire speech or debate. This page
http://www.factcheck.org/2016-presidential-candidates/
is the closest they come to providing summary tallies for each candidate,
but it is more an indexing service than a tallying service, and includes
e.g, lies told about that candidate rather than by him. It is
valuable (and the number
of lies that happen is stunning...) but not so useful for me.
Nevrtheless it is pretty clear that FactCheck.org agrees with the
others that Trump is
the lyingest presidential candidate in their experience. The Guardian
even started a weekly column "The lies Trump told this week"... which
I guess is a public service but too one-sided, since Trump's rivals
including Clinton also lie. However, based on Politifact it
seems Trump lies about 3-5 times more than Clinton depending on precisely how
we measure it; and based on FactCheck.org the ratio might be even
larger than that,
perhaps 6.

What also is amazing about Trump is the distribution of his
truth-ratings. For most
candidates, their distribution is unimodal exhibiting a peak, for
example Sanders' and Clinton's and Kasich's peaks all were at "mostly
true" (although Sanders'
peak was the narrowest and tallest) but Cruz's at "mostly false."
For Trump, his peak frequency is at "false" and his second highest is
for "pants on fire"!

Many fact checkers worldwide are even less useful for me, for various reasons.
For example the MorsiMeter in Egypt focused purely on Morsi campaign promises,
which is too unidirectional and partisan.

As far as I've been able to tell so far, Politifact is by far the most
useful for my purposes since it has the greatest volume (20
facts/week), produces scores, keeps tallies, and
is nonpartisan and independent.
Washington Post is also useful for same reasons, but about 5X smaller in scale
and with fewer tallies.

Warren D Smith

unread,
Sep 27, 2016, 11:51:03 AM9/27/16
to electionscience, Steven Brams
cnn.com/RealityCheck is another fact-checker which
I only found out about at the Presidential Debate.

I'm now going through the lists of facts currently on that web page
tallying up CNN's assessments.

Here's their first link, a compendium of quick checks on claims made
by C and/or T
during the first Clinton-Trump presidential debate, version of noon 27 sept.
CNN rated them true; mostly true; "true but misleading"; false; or
"it's complicated."

"Hempstead, New York (CNN)Donald Trump's tenuous relationship with the
truth didn't stop at the doors of the first presidential debate of
2016.
The Republican nominee repeatedly tussled with Hillary Clinton over
his own past statements, his tax returns and even the murder rate in
New York City.
But Trump also stubbornly stood by two lies that he's stood by in
recent weeks and was bound to be asked about at the debate.
In a lengthy exchange with moderator Lester Holt, Trump repeatedly
insisted -- falsely -- that he opposed the Iraq War. And he continued
to claim that Clinton was involved in spreading the "birther"
conspiracy theory that shot him to political prominence.
While Trump made several brazenly false claims in just 90 minutes
onstage, Clinton stuck to the facts. From misrepresentations to
half-truths and flat-out lies, Trump has talked around and away from
the truth more brazenly than any major party presidential nominee in
modern political history. Clinton [also] has made misleading and even
false statements about her use of a private email server during her
time as secretary of state..."

Iraq war: Trump=false
Birtherism: Trump=false
Constitutionality of "Stop&Frisk": Trump=false
Murders in NYC: Trump=false according to NYPD statistics, but "true
but misleading"
according to FBI stats
TPP called "gold standard" by Clinton: Trump=true
Trump called climate change a hoax "created by the Chinese":
Clinton=true, Trump=false
Tax plans Trump vs Clinton: Trump=mostly false
ISIS: Trump=massive whopper
"Donald rooted for the housing crisis": Clinton=mostly true
"Ford is leaving. Their small car division -- thousands of jobs,
leaving Michigan, leaving Ohio.": Trump=false, Ford not getting rid of
jobs, Ohio & Michigan gained jobs, but
true that there are jobs going from USA to Mexico
Russia hacked the DNC: Clinton=mostly true
Trump denies calling pregnancy "an inconvenience" to employers: Trump=false
How Trump got start: Clinton=true
Clinton's claims about her vs T "jobs plans": Clinton=true but misleading
Trump said he would negotiate down the US debt?: Clinton=true
Clinton on jobs & income during her Husband's admin: Clinton=mostly true
"Biggest tax cut since Reagan" proposed by T: Trump=true
Claims Trump paid zero taxes in every year his returns released:
Clinton=true, the years being 1978, 1979, 1984, 1991 and 1993.
Clinton on effect of C & T's plans on US debt: Clinton=mostly true.
Trump "endorsed by ICE"?: More precisely he was endorsed by a union
representing ICE employees, in which case Trump=true.
Trump re the Federal reserve politics: Trump=false
Claims re Trump support of wars in Iraq, Libya: Clinton=true,
Trump=false (each twice)
Clinton re "gun epidemic" major cause of black deaths: Clinton=mostly true
Clinton re Trump's disparaging comments to beauty contestant": Clinton=true
USA pays 73% of cost of NATO?: Trump=false
Trump inspired NATO to create terror unit?: Trump=false
Trump's claim that US not updating own nuclear technology: false
Trump's claim US planes are old enough to be flown by your
grandfather: true but misleading.

MY ATTEMPT TO TALLY:
Trump=false: 14
Trump=true but misleading: 2
Trump=true: 3

Clinton=false: 0
Clinton=mostly true: 5
Clinton=true but misleading: 1
Clinton=true: 6

CONCLUSION: CNN agrees with all the other fact checkers that Trump has
"talked around and away from the truth more brazenly than any major
party presidential nominee in modern political history." This
agreement is further confirmation of
the mutual validity of these fact checkers' assessments, but does not confirm or
deny those checkers' assessments of all the other candidates such as
Sanders and Cruz.

Warren D Smith

unread,
Sep 27, 2016, 1:33:02 PM9/27/16
to electionscience, Steven Brams, Marcus Pivato
The France 2012 presidential election may also be a further source
of data about approval ratings versus truth ratings.

Main candidates:
Francois Bayrou, Francois Hollande, Eva Joly, Marine Le Pen, Jean-Luc
Melenchon and Nicolas Sarkozy.

Approval-style polls:
http://www.rangevoting.org/France2012.html

France fact-checking entities:
Decodeurs: http://decodeurs.blog.lemonde.fr/
Les Pinocchios:
http://tempsreel.nouvelobs.com/politique/les-pinocchios-de-l-obs/
OWNI veritometre:
http://owni.fr/2012/02/16/veritometre-factchecking-presidentielle/
http://owni.fr/2012/05/03/veritometre-debat-hollande-sarkozy/ and etc
Desintox: no longer available?

It helps if you read French... (I don't)

Warren D Smith

unread,
Sep 29, 2016, 8:07:27 PM9/29/16
to electionscience, Steven Brams, Jean-Francois Laslier, Jean François Laslier, Marcus Pivato
I was able to obtain approval and truthfulness data for
the France 2012 presidential election and have added it to

http://rangevoting.org/TruthApp.txt

However additional such data would be desirable (I explain why there),
such as from Le Monde.
The French data as matters currently stand does not support the
approval<->truthfulness connection nearly as strongly
as the USA 2016 and USA 2012 data.

Warren D Smith

unread,
Oct 7, 2016, 1:05:45 PM10/7/16
to electionscience

Warren D Smith

unread,
Oct 10, 2016, 6:15:07 PM10/10/16
to electionscience
CORRECTION: I screwed up counts for Joe Biden on the
page rangevoting.org/TruthApp.html, need to fix it.

-----

About trying to look at the USA 2008 presidential election,
here are all the candidates with at least 25 checked facts on file
from politifact

NAME H5 H4 H3 H2 H1 H0 FC
Dennis Kucinich 44 24 16 4 12 0 25 hon=68
Hillary Clinton 22 28 22 15 11 2 249 hon=50
Barack Obama 21 28 27 12 12 2 572 hon=49
Ron Paul 20 20 20 13 20 8 40 hon=40
John McCain 20 20 17 21 4 183 hon=40
Joe Biden 17 21 28 15 13 5 75 hon=38
Mitt Romney 15 16 28 17 16 9 206 hon=31
Rudy Giuliani 15 15 21 19 23 47 hon=30
Mike Huckabee 20 7 22 29 12 10 41 hon=27

ranked in descending order of honesty=H5+H4.
FC is number of facts checked, H3 is percent of statements scored with
honesty level 3 on a scale from 0 to 5, where 5 is most honest.
Unfortunately for names who ran in other elections after 2008
(H.Clinton, Ron Paul, Mitt Romney)
their honesty data pertains also to those other eras, all mixed together.

Warren D Smith

unread,
Oct 10, 2016, 7:11:17 PM10/10/16
to electionscience
On 10/10/16, Warren D Smith <warre...@gmail.com> wrote:
> CORRECTION: I screwed up counts for Joe Biden on the
> page rangevoting.org/TruthApp.html, need to fix it.
>
> -----
>
> About trying to look at the USA 2008 presidential election,
> here are all the candidates with at least 25 checked facts on file
> from politifact
>
> NAME H5 H4 H3 H2 H1 H0 FC
> Dennis Kucinich 44 24 16 4 12 0 25 hon=68
> Hillary Clinton 22 28 22 15 11 2 249 hon=50
> Barack Obama 21 28 27 12 12 2 572 hon=49
> Ron Paul 20 20 20 13 20 8 40 hon=40
> John McCain 20 20 17 21 4 183 hon=40
> Joe Biden 17 21 28 15 13 5 75 hon=38
> Mitt Romney 15 16 28 17 16 9 206 hon=31
> Rudy Giuliani 15 15 21 19 23 47 hon=30
> Mike Huckabee 20 7 22 29 12 10 41 hon=27
>
> ranked in descending order of honesty=H5+H4.
> FC is number of facts checked, H3 is percent of statements scored with
> honesty level 3 on a scale from 0 to 5, where 5 is most honest.

http://www.pollingreport.com/
is a free-access database of polls I just discovered, containing a lot
of approval style and some score-style polls.

it looks like Obama and Biden and Romney (in descending order, except
Biden-Obama perhaps should be reversed)
were the approval leaders in 2008
among the names above. Note Biden ran as Obama's VP,
not directly for president, in the general election; he withdrew from
the Dem party primary after the Iowa caucus. McCain ran as
republican nominee.

If so this data does not support the approval<->truth connection.
Instead, the two seem unrelated in the 2008 election.
The data is not very good though -- many candidates have few facts
checked, and/or
have few approval polls and/or which are too far away in time.
If we restrict attention only to "well known" candidates who at least 66% of
voters had an approval/disapprove opinion about, then Kucinich
and Paul vanish (instead requiring at least 41 checked facts
accomplishes same thing)
and then things look a bit better for the Connection, but still bad.

Warren D Smith

unread,
Nov 7, 2016, 12:51:51 PM11/7/16
to electionscience, Steven Brams
https://www.theguardian.com/us-news/2016/nov/07/how-does-donald-trump-lie-fact-checker

is a highly inadequate attempt by THE GUARDIAN to summarize
Trump's lies.
Reply all
Reply to author
Forward
0 new messages