True news vs fake news

[Marblehead]

 

 

I often like charts and data tables but usually suspicious when they are poll driven... but this seems interesting to share.

I agree.  I want facts, not opinions.

[Rishi Das]

eye opening and nicely laid out -

 

Edited February 27, 2017 by Rishi Das

[Apech]

Accurate polling is very simple.  Pick a large enough sample of people randomly selected each time (i.e. don't select by gender, ethnicity and so on) and don't repeatedly use same database, ask them the same simple questions about their opinions or intentions.  Report result without massaging the figures.

 

This will give an accurate result within a small band of confidence based on a ) the sample size and b ) the strength of a given answer.

 

Given this, apart from the fact that it is expensive to do especially in an electoral college system (as you would have to sample each state in the same way), it is hard to understand why polls are repeatedly wrong and the only conclusion is that the pollsters are using false results to massage public opinion and expectation.

 

I'm guessing the private polling is done like this and we are not given the results.

[Jeff]

Accurate polling is very simple.  Pick a large enough sample of people randomly selected each time (i.e. don't select by gender, ethnicity and so on) and don't repeatedly use same database, ask them the same simple questions about their opinions or intentions.  Report result without massaging the figures.

 

This will give an accurate result within a small band of confidence based on a ) the sample size and b ) the strength of a given answer.

 

Given this, apart from the fact that it is expensive to do especially in an electoral college system (as you would have to sample each state in the same way), it is hard to understand why polls are repeatedly wrong and the only conclusion is that the pollsters are using false results to massage public opinion and expectation.

 

I'm guessing the private polling is done like this and we are not given the results.

 

 

That theory assumes that all participants will respond at an equal rate to the polling. In the Trump example, in many areas it was seen a politically incorrect to admit voting for him, hence large sections of the populace did not participate in polls.  This leaves a misguided view of the sample set.

[Taomeow]

ABC News executive producer Ian Cameron is married to Susan Rice, National Security Adviser.
CBS President David Rhodes is the brother of Ben Rhodes, Obama's Deputy National Security Adviser for Strategic Communications.
ABC News correspondent Claire Shipman is married to former Whitehouse Press Secretary Jay Carney
ABC News and Univision reporter Matthew Jaffe is married to Katie Hogan, Obama's Deputy Press Secretary
ABC President Ben Sherwood is the brother of Obama's Special Adviser Elizabeth Sherwood
CNN President Virginia Moseley is married to former Hillary Clinton's Deputy Secretary Tom Nides.
NBC News Anchor, Andrea Mitchell is married to Alan Greenspan, Chairman of the Federal Reserve of the United States from 1987 to 2006.
CNN's (formerly of NBC) David Gregory is married to Beth Wilkinson, the attorney who represented several of Hillary Clinton's staff members during the email investigation
Associated but not related:
ABC News Anchor George Stephanopoulos, Clinton’s White House Communications Director and press secretary
PBS Chief of Staff Julie Anbender, Clinton’s Deputy Director of OPA and DOJ
Comcast-ABC Senior VP, Government Affairs Meredith Baker, Bush’s and Obama’s Commissioner, FCC
CNN News correspondent, PBS Host Yul Kwon, Obama’s Deputy Chief CGAB, FCC
CNN Department Assignment Manager, Joe Lockhart, Clinton’s Press Secretary
CBS VP, Corporate Communications, Lisa Caputo, Clinton’s Press Secretary, First Lady’s Office
NBC Political Director, Chuck Todd worked on Senator Tom Harkin's (D) campaign
FOX News Contributor, Marie Harf, senior adviser for strategic communications to Secretary of State John Kerry
WASHINGTON POST Contributor, John Podesta, former chairman of the 2016 Hillary Clinton campaign, served as Chief of Staff to President Bill Clinton and as a counselor to President Barack Obama. He is the founder and board member of the Center for American Progress, a liberal think tank in Washington, D.C., also a visiting professor of law at the Georgetown University Law Center.
FOX News Contributor, Dr. Ezekiel Emanuel, Emanuel is best known for being one of the “architects” of the Affordable Care Act.

[dawei]

That theory assumes that all participants will respond at an equal rate to the polling. In the Trump example, in many areas it was seen a politically incorrect to admit voting for him, hence large sections of the populace did not participate in polls.  This leaves a misguided view of the sample set.

 

Polling may claim that it is based on science but it is not without its inherent weakness that it can only be reflective of only those that actually took the poll... seems obvious enough but in this election year, the weakness was fully revealed.    Another weakness may be that such polling is based more on the idea of a popular vote and thus was correct in another way.  They were maybe polling the wrong locations too.    You can go to many locations and likely guess the outcome.

 

There were actually a handful of folks who were predicting Trump but nobody really stopped to understand why that was the case and test whether there was something amiss in traditional polling.  Going forward, they should likely effect some changes in how they poll but then it would not be as random and thus likely to not really have any significant change. 

[Apech]

That theory assumes that all participants will respond at an equal rate to the polling. In the Trump example, in many areas it was seen a politically incorrect to admit voting for him, hence large sections of the populace did not participate in polls.  This leaves a misguided view of the sample set.

 

 

 

Non participators don't matter if you sample correctly.  Telephone and internet polling are skewed so it has to be in person and in confidence.  This way you reduce the number who simply lie about who they are going to vote for.

[Apech]

Polling may claim that it is based on science but it is not without its inherent weakness that it can only be reflective of only those that actually took the poll... seems obvious enough but in this election year, the weakness was fully revealed.    Another weakness may be that such polling is based more on the idea of a popular vote and thus was correct in another way.  They were maybe polling the wrong locations too.    You can go to many locations and likely guess the outcome.

 

There were actually a handful of folks who were predicting Trump but nobody really stopped to understand why that was the case and test whether there was something amiss in traditional polling.  Going forward, they should likely effect some changes in how they poll but then it would not be as random and thus likely to not really have any significant change. 

 

Yes you would have to sample state by state because of the electoral college - unless you are only interested in the popular vote - but of course that doesn't give you the result.

[Brian]

Yes you would have to sample state by state because of the electoral college - unless you are only interested in the popular vote - but of course that doesn't give you the result.

And it gets far more complicated if you are trying to predict Primary results as the rules vary from state to state and the Parties don't have to have the same rules within a given state. At the General Election, only Maine & Nebraska don't do winner-take-all based on statewide popular vote.

 

RNC Primary Rules

DNC Primary Rules

 

Personally, I think states should move away from the popular vote and allocate electoral votes based on Congressional Districts, but then I think the Congress should repeal the “Permanent Apportionment Act of 1929” and expand the House of Representatives to be in alignment with the Constitutional specification of "The number of Representatives shall not exceed one for every thirty Thousand." While 435 technically fulfills that requirement, it clearly isn't in keeping with the spirit of the Framers. Having in the neighborhood of 11 thousand Representatives distributed across the US would certainly make influence-peddling more difficult! Should also repeal the 17th Amendment and return the selection of Senators to the individual State Legislatures. These two things would effectively squash the current oligarchical collective.